• Calling on LLMs: New NVIDIA AI Blueprint Helps Automate Telco Network Configuration

    Telecom companies last year spent nearly billion in capital expenditures and over trillion in operating expenditures.
    These large expenses are due in part to laborious manual processes that telcos face when operating networks that require continuous optimizations.
    For example, telcos must constantly tune network parameters for tasks — such as transferring calls from one network to another or distributing network traffic across multiple servers — based on the time of day, user behavior, mobility and traffic type.
    These factors directly affect network performance, user experience and energy consumption.
    To automate these optimization processes and save costs for telcos across the globe, NVIDIA today unveiled at GTC Paris its first AI Blueprint for telco network configuration.
    At the blueprint’s core are customized large language models trained specifically on telco network data — as well as the full technical and operational architecture for turning the LLMs into an autonomous, goal-driven AI agent for telcos.
    Automate Network Configuration With the AI Blueprint
    NVIDIA AI Blueprints — available on build.nvidia.com — are customizable AI workflow examples. They include reference code, documentation and deployment tools that show enterprise developers how to deliver business value with NVIDIA NIM microservices.
    The AI Blueprint for telco network configuration — built with BubbleRAN 5G solutions and datasets — enables developers, network engineers and telecom providers to automatically optimize the configuration of network parameters using agentic AI.
    This can streamline operations, reduce costs and significantly improve service quality by embedding continuous learning and adaptability directly into network infrastructures.
    Traditionally, network configurations required manual intervention or followed rigid rules to adapt to dynamic network conditions. These approaches limited adaptability and increased operational complexities, costs and inefficiencies.
    The new blueprint helps shift telco operations from relying on static, rules-based systems to operations based on dynamic, AI-driven automation. It enables developers to build advanced, telco-specific AI agents that make real-time, intelligent decisions and autonomously balance trade-offs — such as network speed versus interference, or energy savings versus utilization — without human input.
    Powered and Deployed by Industry Leaders
    Trained on 5G data generated by BubbleRAN, and deployed on the BubbleRAN 5G O-RAN platform, the blueprint provides telcos with insight on how to set various parameters to reach performance goals, like achieving a certain bitrate while choosing an acceptable signal-to-noise ratio — a measure that impacts voice quality and thus user experience.
    With the new AI Blueprint, network engineers can confidently set initial parameter values and update them as demanded by continuous network changes.
    Norway-based Telenor Group, which serves over 200 million customers globally, is the first telco to integrate the AI Blueprint for telco network configuration as part of its initiative to deploy intelligent, autonomous networks that meet the performance and agility demands of 5G and beyond.
    “The blueprint is helping us address configuration challenges and enhance quality of service during network installation,” said Knut Fjellheim, chief technology innovation officer at Telenor Maritime. “Implementing it is part of our push toward network automation and follows the successful deployment of agentic AI for real-time network slicing in a private 5G maritime use case.”
    Industry Partners Deploy Other NVIDIA-Powered Autonomous Network Technologies
    The AI Blueprint for telco network configuration is just one of many announcements at NVIDIA GTC Paris showcasing how the telecom industry is using agentic AI to make autonomous networks a reality.
    Beyond the blueprint, leading telecom companies and solutions providers are tapping into NVIDIA accelerated computing, software and microservices to provide breakthrough innovations poised to vastly improve networks and communications services — accelerating the progress to autonomous networks and improving customer experiences.
    NTT DATA is powering its agentic platform for telcos with NVIDIA accelerated compute and the NVIDIA AI Enterprise software platform. Its first agentic use case is focused on network alarms management, where NVIDIA NIM microservices help automate and power observability, troubleshooting, anomaly detection and resolution with closed loop ticketing.
    Tata Consultancy Services is delivering agentic AI solutions for telcos built on NVIDIA DGX Cloud and using NVIDIA AI Enterprise to develop, fine-tune and integrate large telco models into AI agent workflows. These range from billing and revenue assurance, autonomous network management to hybrid edge-cloud distributed inference.
    For example, the company’s anomaly management agentic AI model includes real-time detection and resolution of network anomalies and service performance optimization. This increases business agility and improves operational efficiencies by up to 40% by eliminating human intensive toils, overheads and cross-departmental silos.
    Prodapt has introduced an autonomous operations workflow for networks, powered by NVIDIA AI Enterprise, that offers agentic AI capabilities to support autonomous telecom networks. AI agents can autonomously monitor networks, detect anomalies in real time, initiate diagnostics, analyze root causes of issues using historical data and correlation techniques, automatically execute corrective actions, and generate, enrich and assign incident tickets through integrated ticketing systems.
    Accenture announced its new portfolio of agentic AI solutions for telecommunications through its AI Refinery platform, built on NVIDIA AI Enterprise software and accelerated computing.
    The first available solution, the NOC Agentic App, boosts network operations center tasks by using a generative AI-driven, nonlinear agentic framework to automate processes such as incident and fault management, root cause analysis and configuration planning. Using the Llama 3.1 70B NVIDIA NIM microservice and the AI Refinery Distiller Framework, the NOC Agentic App orchestrates networks of intelligent agents for faster, more efficient decision-making.
    Infosys is announcing its agentic autonomous operations platform, called Infosys Smart Network Assurance, designed to accelerate telecom operators’ journeys toward fully autonomous network operations.
    ISNA helps address long-standing operational challenges for telcos — such as limited automation and high average time to repair — with an integrated, AI-driven platform that reduces operational costs by up to 40% and shortens fault resolution times by up to 30%. NVIDIA NIM and NeMo microservices enhance the platform’s reasoning and hallucination-detection capabilities, reduce latency and increase accuracy.
    Get started with the new blueprint today.
    Learn more about the latest AI advancements for telecom and other industries at NVIDIA GTC Paris, running through Thursday, June 12, at VivaTech, including a keynote from NVIDIA founder and CEO Jensen Huang and a special address from Ronnie Vasishta, senior vice president of telecom at NVIDIA. Plus, hear from industry leaders in a panel session with Orange, Swisscom, Telenor and NVIDIA.
    #calling #llms #new #nvidia #blueprint
    Calling on LLMs: New NVIDIA AI Blueprint Helps Automate Telco Network Configuration
    Telecom companies last year spent nearly billion in capital expenditures and over trillion in operating expenditures. These large expenses are due in part to laborious manual processes that telcos face when operating networks that require continuous optimizations. For example, telcos must constantly tune network parameters for tasks — such as transferring calls from one network to another or distributing network traffic across multiple servers — based on the time of day, user behavior, mobility and traffic type. These factors directly affect network performance, user experience and energy consumption. To automate these optimization processes and save costs for telcos across the globe, NVIDIA today unveiled at GTC Paris its first AI Blueprint for telco network configuration. At the blueprint’s core are customized large language models trained specifically on telco network data — as well as the full technical and operational architecture for turning the LLMs into an autonomous, goal-driven AI agent for telcos. Automate Network Configuration With the AI Blueprint NVIDIA AI Blueprints — available on build.nvidia.com — are customizable AI workflow examples. They include reference code, documentation and deployment tools that show enterprise developers how to deliver business value with NVIDIA NIM microservices. The AI Blueprint for telco network configuration — built with BubbleRAN 5G solutions and datasets — enables developers, network engineers and telecom providers to automatically optimize the configuration of network parameters using agentic AI. This can streamline operations, reduce costs and significantly improve service quality by embedding continuous learning and adaptability directly into network infrastructures. Traditionally, network configurations required manual intervention or followed rigid rules to adapt to dynamic network conditions. These approaches limited adaptability and increased operational complexities, costs and inefficiencies. The new blueprint helps shift telco operations from relying on static, rules-based systems to operations based on dynamic, AI-driven automation. It enables developers to build advanced, telco-specific AI agents that make real-time, intelligent decisions and autonomously balance trade-offs — such as network speed versus interference, or energy savings versus utilization — without human input. Powered and Deployed by Industry Leaders Trained on 5G data generated by BubbleRAN, and deployed on the BubbleRAN 5G O-RAN platform, the blueprint provides telcos with insight on how to set various parameters to reach performance goals, like achieving a certain bitrate while choosing an acceptable signal-to-noise ratio — a measure that impacts voice quality and thus user experience. With the new AI Blueprint, network engineers can confidently set initial parameter values and update them as demanded by continuous network changes. Norway-based Telenor Group, which serves over 200 million customers globally, is the first telco to integrate the AI Blueprint for telco network configuration as part of its initiative to deploy intelligent, autonomous networks that meet the performance and agility demands of 5G and beyond. “The blueprint is helping us address configuration challenges and enhance quality of service during network installation,” said Knut Fjellheim, chief technology innovation officer at Telenor Maritime. “Implementing it is part of our push toward network automation and follows the successful deployment of agentic AI for real-time network slicing in a private 5G maritime use case.” Industry Partners Deploy Other NVIDIA-Powered Autonomous Network Technologies The AI Blueprint for telco network configuration is just one of many announcements at NVIDIA GTC Paris showcasing how the telecom industry is using agentic AI to make autonomous networks a reality. Beyond the blueprint, leading telecom companies and solutions providers are tapping into NVIDIA accelerated computing, software and microservices to provide breakthrough innovations poised to vastly improve networks and communications services — accelerating the progress to autonomous networks and improving customer experiences. NTT DATA is powering its agentic platform for telcos with NVIDIA accelerated compute and the NVIDIA AI Enterprise software platform. Its first agentic use case is focused on network alarms management, where NVIDIA NIM microservices help automate and power observability, troubleshooting, anomaly detection and resolution with closed loop ticketing. Tata Consultancy Services is delivering agentic AI solutions for telcos built on NVIDIA DGX Cloud and using NVIDIA AI Enterprise to develop, fine-tune and integrate large telco models into AI agent workflows. These range from billing and revenue assurance, autonomous network management to hybrid edge-cloud distributed inference. For example, the company’s anomaly management agentic AI model includes real-time detection and resolution of network anomalies and service performance optimization. This increases business agility and improves operational efficiencies by up to 40% by eliminating human intensive toils, overheads and cross-departmental silos. Prodapt has introduced an autonomous operations workflow for networks, powered by NVIDIA AI Enterprise, that offers agentic AI capabilities to support autonomous telecom networks. AI agents can autonomously monitor networks, detect anomalies in real time, initiate diagnostics, analyze root causes of issues using historical data and correlation techniques, automatically execute corrective actions, and generate, enrich and assign incident tickets through integrated ticketing systems. Accenture announced its new portfolio of agentic AI solutions for telecommunications through its AI Refinery platform, built on NVIDIA AI Enterprise software and accelerated computing. The first available solution, the NOC Agentic App, boosts network operations center tasks by using a generative AI-driven, nonlinear agentic framework to automate processes such as incident and fault management, root cause analysis and configuration planning. Using the Llama 3.1 70B NVIDIA NIM microservice and the AI Refinery Distiller Framework, the NOC Agentic App orchestrates networks of intelligent agents for faster, more efficient decision-making. Infosys is announcing its agentic autonomous operations platform, called Infosys Smart Network Assurance, designed to accelerate telecom operators’ journeys toward fully autonomous network operations. ISNA helps address long-standing operational challenges for telcos — such as limited automation and high average time to repair — with an integrated, AI-driven platform that reduces operational costs by up to 40% and shortens fault resolution times by up to 30%. NVIDIA NIM and NeMo microservices enhance the platform’s reasoning and hallucination-detection capabilities, reduce latency and increase accuracy. Get started with the new blueprint today. Learn more about the latest AI advancements for telecom and other industries at NVIDIA GTC Paris, running through Thursday, June 12, at VivaTech, including a keynote from NVIDIA founder and CEO Jensen Huang and a special address from Ronnie Vasishta, senior vice president of telecom at NVIDIA. Plus, hear from industry leaders in a panel session with Orange, Swisscom, Telenor and NVIDIA. #calling #llms #new #nvidia #blueprint
    BLOGS.NVIDIA.COM
    Calling on LLMs: New NVIDIA AI Blueprint Helps Automate Telco Network Configuration
    Telecom companies last year spent nearly $295 billion in capital expenditures and over $1 trillion in operating expenditures. These large expenses are due in part to laborious manual processes that telcos face when operating networks that require continuous optimizations. For example, telcos must constantly tune network parameters for tasks — such as transferring calls from one network to another or distributing network traffic across multiple servers — based on the time of day, user behavior, mobility and traffic type. These factors directly affect network performance, user experience and energy consumption. To automate these optimization processes and save costs for telcos across the globe, NVIDIA today unveiled at GTC Paris its first AI Blueprint for telco network configuration. At the blueprint’s core are customized large language models trained specifically on telco network data — as well as the full technical and operational architecture for turning the LLMs into an autonomous, goal-driven AI agent for telcos. Automate Network Configuration With the AI Blueprint NVIDIA AI Blueprints — available on build.nvidia.com — are customizable AI workflow examples. They include reference code, documentation and deployment tools that show enterprise developers how to deliver business value with NVIDIA NIM microservices. The AI Blueprint for telco network configuration — built with BubbleRAN 5G solutions and datasets — enables developers, network engineers and telecom providers to automatically optimize the configuration of network parameters using agentic AI. This can streamline operations, reduce costs and significantly improve service quality by embedding continuous learning and adaptability directly into network infrastructures. Traditionally, network configurations required manual intervention or followed rigid rules to adapt to dynamic network conditions. These approaches limited adaptability and increased operational complexities, costs and inefficiencies. The new blueprint helps shift telco operations from relying on static, rules-based systems to operations based on dynamic, AI-driven automation. It enables developers to build advanced, telco-specific AI agents that make real-time, intelligent decisions and autonomously balance trade-offs — such as network speed versus interference, or energy savings versus utilization — without human input. Powered and Deployed by Industry Leaders Trained on 5G data generated by BubbleRAN, and deployed on the BubbleRAN 5G O-RAN platform, the blueprint provides telcos with insight on how to set various parameters to reach performance goals, like achieving a certain bitrate while choosing an acceptable signal-to-noise ratio — a measure that impacts voice quality and thus user experience. With the new AI Blueprint, network engineers can confidently set initial parameter values and update them as demanded by continuous network changes. Norway-based Telenor Group, which serves over 200 million customers globally, is the first telco to integrate the AI Blueprint for telco network configuration as part of its initiative to deploy intelligent, autonomous networks that meet the performance and agility demands of 5G and beyond. “The blueprint is helping us address configuration challenges and enhance quality of service during network installation,” said Knut Fjellheim, chief technology innovation officer at Telenor Maritime. “Implementing it is part of our push toward network automation and follows the successful deployment of agentic AI for real-time network slicing in a private 5G maritime use case.” Industry Partners Deploy Other NVIDIA-Powered Autonomous Network Technologies The AI Blueprint for telco network configuration is just one of many announcements at NVIDIA GTC Paris showcasing how the telecom industry is using agentic AI to make autonomous networks a reality. Beyond the blueprint, leading telecom companies and solutions providers are tapping into NVIDIA accelerated computing, software and microservices to provide breakthrough innovations poised to vastly improve networks and communications services — accelerating the progress to autonomous networks and improving customer experiences. NTT DATA is powering its agentic platform for telcos with NVIDIA accelerated compute and the NVIDIA AI Enterprise software platform. Its first agentic use case is focused on network alarms management, where NVIDIA NIM microservices help automate and power observability, troubleshooting, anomaly detection and resolution with closed loop ticketing. Tata Consultancy Services is delivering agentic AI solutions for telcos built on NVIDIA DGX Cloud and using NVIDIA AI Enterprise to develop, fine-tune and integrate large telco models into AI agent workflows. These range from billing and revenue assurance, autonomous network management to hybrid edge-cloud distributed inference. For example, the company’s anomaly management agentic AI model includes real-time detection and resolution of network anomalies and service performance optimization. This increases business agility and improves operational efficiencies by up to 40% by eliminating human intensive toils, overheads and cross-departmental silos. Prodapt has introduced an autonomous operations workflow for networks, powered by NVIDIA AI Enterprise, that offers agentic AI capabilities to support autonomous telecom networks. AI agents can autonomously monitor networks, detect anomalies in real time, initiate diagnostics, analyze root causes of issues using historical data and correlation techniques, automatically execute corrective actions, and generate, enrich and assign incident tickets through integrated ticketing systems. Accenture announced its new portfolio of agentic AI solutions for telecommunications through its AI Refinery platform, built on NVIDIA AI Enterprise software and accelerated computing. The first available solution, the NOC Agentic App, boosts network operations center tasks by using a generative AI-driven, nonlinear agentic framework to automate processes such as incident and fault management, root cause analysis and configuration planning. Using the Llama 3.1 70B NVIDIA NIM microservice and the AI Refinery Distiller Framework, the NOC Agentic App orchestrates networks of intelligent agents for faster, more efficient decision-making. Infosys is announcing its agentic autonomous operations platform, called Infosys Smart Network Assurance (ISNA), designed to accelerate telecom operators’ journeys toward fully autonomous network operations. ISNA helps address long-standing operational challenges for telcos — such as limited automation and high average time to repair — with an integrated, AI-driven platform that reduces operational costs by up to 40% and shortens fault resolution times by up to 30%. NVIDIA NIM and NeMo microservices enhance the platform’s reasoning and hallucination-detection capabilities, reduce latency and increase accuracy. Get started with the new blueprint today. Learn more about the latest AI advancements for telecom and other industries at NVIDIA GTC Paris, running through Thursday, June 12, at VivaTech, including a keynote from NVIDIA founder and CEO Jensen Huang and a special address from Ronnie Vasishta, senior vice president of telecom at NVIDIA. Plus, hear from industry leaders in a panel session with Orange, Swisscom, Telenor and NVIDIA.
    Like
    Love
    Wow
    Sad
    Angry
    80
    0 Commentarios 0 Acciones
  • Plug and Play: Build a G-Assist Plug-In Today

    Project G-Assist — available through the NVIDIA App — is an experimental AI assistant that helps tune, control and optimize NVIDIA GeForce RTX systems.
    NVIDIA’s Plug and Play: Project G-Assist Plug-In Hackathon — running virtually through Wednesday, July 16 — invites the community to explore AI and build custom G-Assist plug-ins for a chance to win prizes and be featured on NVIDIA social media channels.

    G-Assist allows users to control their RTX GPU and other system settings using natural language, thanks to a small language model that runs on device. It can be used from the NVIDIA Overlay in the NVIDIA App without needing to tab out or switch programs. Users can expand its capabilities via plug-ins and even connect it to agentic frameworks such as Langflow.
    Below, find popular G-Assist plug-ins, hackathon details and tips to get started.
    Plug-In and Win
    Join the hackathon by registering and checking out the curated technical resources.
    G-Assist plug-ins can be built in several ways, including with Python for rapid development, with C++ for performance-critical apps and with custom system interactions for hardware and operating system automation.
    For those that prefer vibe coding, the G-Assist Plug-In Builder — a ChatGPT-based app that allows no-code or low-code development with natural language commands — makes it easy for enthusiasts to start creating plug-ins.
    To submit an entry, participants must provide a GitHub repository, including source code file, requirements.txt, manifest.json, config.json, a plug-in executable file and READme code.
    Then, submit a video — between 30 seconds and two minutes — showcasing the plug-in in action.
    Finally, hackathoners must promote their plug-in using #AIonRTXHackathon on a social media channel: Instagram, TikTok or X. Submit projects via this form by Wednesday, July 16.
    Judges will assess plug-ins based on three main criteria: 1) innovation and creativity, 2) technical execution and integration, reviewing technical depth, G-Assist integration and scalability, and 3) usability and community impact, aka how easy it is to use the plug-in.
    Winners will be selected on Wednesday, Aug. 20. First place will receive a GeForce RTX 5090 laptop, second place a GeForce RTX 5080 GPU and third a GeForce RTX 5070 GPU. These top three will also be featured on NVIDIA’s social media channels, get the opportunity to meet the NVIDIA G-Assist team and earn an NVIDIA Deep Learning Institute self-paced course credit.
    Project G-Assist requires a GeForce RTX 50, 40 or 30 Series Desktop GPU with at least 12GB of VRAM, Windows 11 or 10 operating system, a compatible CPU, specific disk space requirements and a recent GeForce Game Ready Driver or NVIDIA Studio Driver.
    Plug-InExplore open-source plug-in samples available on GitHub, which showcase the diverse ways on-device AI can enhance PC and gaming workflows.

    Popular plug-ins include:

    Google Gemini: Enables search-based queries using Google Search integration and large language model-based queries using Gemini capabilities in real time without needing to switch programs from the convenience of the NVIDIA App Overlay.
    Discord: Enables users to easily share game highlights or messages directly to Discord servers without disrupting gameplay.
    IFTTT: Lets users create automations across hundreds of compatible endpoints to trigger IoT routines — such as adjusting room lights and smart shades, or pushing the latest gaming news to a mobile device.
    Spotify: Lets users control Spotify using simple voice commands or the G-Assist interface to play favorite tracks and manage playlists.
    Twitch: Checks if any Twitch streamer is currently live and can access detailed stream information such as titles, games, view counts and more.

    Get G-Assist 
    Join the NVIDIA Developer Discord channel to collaborate, share creations and gain support from fellow AI enthusiasts and NVIDIA staff.
    the date for NVIDIA’s How to Build a G-Assist Plug-In webinar on Wednesday, July 9, from 10-11 a.m. PT, to learn more about Project G-Assist capabilities, discover the fundamentals of building, testing and deploying Project G-Assist plug-ins, and participate in a live Q&A session.
    Explore NVIDIA’s GitHub repository, which provides everything needed to get started developing with G-Assist, including sample plug-ins, step-by-step instructions and documentation for building custom functionalities.
    Learn more about the ChatGPT Plug-In Builder to transform ideas into functional G-Assist plug-ins with minimal coding. The tool uses OpenAI’s custom GPT builder to generate plug-in code and streamline the development process.
    NVIDIA’s technical blog walks through the architecture of a G-Assist plug-in, using a Twitch integration as an example. Discover how plug-ins work, how they communicate with G-Assist and how to build them from scratch.
    Each week, the RTX AI Garage blog series features community-driven AI innovations and content for those looking to learn more about NVIDIA NIM microservices and AI Blueprints, as well as building AI agents, creative workflows, digital humans, productivity apps and more on AI PCs and workstations. 
    Plug in to NVIDIA AI PC on Facebook, Instagram, TikTok and X — and stay informed by subscribing to the RTX AI PC newsletter.
    Follow NVIDIA Workstation on LinkedIn and X. 
    See notice regarding software product information.
    #plug #play #build #gassist #plugin
    Plug and Play: Build a G-Assist Plug-In Today
    Project G-Assist — available through the NVIDIA App — is an experimental AI assistant that helps tune, control and optimize NVIDIA GeForce RTX systems. NVIDIA’s Plug and Play: Project G-Assist Plug-In Hackathon — running virtually through Wednesday, July 16 — invites the community to explore AI and build custom G-Assist plug-ins for a chance to win prizes and be featured on NVIDIA social media channels. G-Assist allows users to control their RTX GPU and other system settings using natural language, thanks to a small language model that runs on device. It can be used from the NVIDIA Overlay in the NVIDIA App without needing to tab out or switch programs. Users can expand its capabilities via plug-ins and even connect it to agentic frameworks such as Langflow. Below, find popular G-Assist plug-ins, hackathon details and tips to get started. Plug-In and Win Join the hackathon by registering and checking out the curated technical resources. G-Assist plug-ins can be built in several ways, including with Python for rapid development, with C++ for performance-critical apps and with custom system interactions for hardware and operating system automation. For those that prefer vibe coding, the G-Assist Plug-In Builder — a ChatGPT-based app that allows no-code or low-code development with natural language commands — makes it easy for enthusiasts to start creating plug-ins. To submit an entry, participants must provide a GitHub repository, including source code file, requirements.txt, manifest.json, config.json, a plug-in executable file and READme code. Then, submit a video — between 30 seconds and two minutes — showcasing the plug-in in action. Finally, hackathoners must promote their plug-in using #AIonRTXHackathon on a social media channel: Instagram, TikTok or X. Submit projects via this form by Wednesday, July 16. Judges will assess plug-ins based on three main criteria: 1) innovation and creativity, 2) technical execution and integration, reviewing technical depth, G-Assist integration and scalability, and 3) usability and community impact, aka how easy it is to use the plug-in. Winners will be selected on Wednesday, Aug. 20. First place will receive a GeForce RTX 5090 laptop, second place a GeForce RTX 5080 GPU and third a GeForce RTX 5070 GPU. These top three will also be featured on NVIDIA’s social media channels, get the opportunity to meet the NVIDIA G-Assist team and earn an NVIDIA Deep Learning Institute self-paced course credit. Project G-Assist requires a GeForce RTX 50, 40 or 30 Series Desktop GPU with at least 12GB of VRAM, Windows 11 or 10 operating system, a compatible CPU, specific disk space requirements and a recent GeForce Game Ready Driver or NVIDIA Studio Driver. Plug-InExplore open-source plug-in samples available on GitHub, which showcase the diverse ways on-device AI can enhance PC and gaming workflows. Popular plug-ins include: Google Gemini: Enables search-based queries using Google Search integration and large language model-based queries using Gemini capabilities in real time without needing to switch programs from the convenience of the NVIDIA App Overlay. Discord: Enables users to easily share game highlights or messages directly to Discord servers without disrupting gameplay. IFTTT: Lets users create automations across hundreds of compatible endpoints to trigger IoT routines — such as adjusting room lights and smart shades, or pushing the latest gaming news to a mobile device. Spotify: Lets users control Spotify using simple voice commands or the G-Assist interface to play favorite tracks and manage playlists. Twitch: Checks if any Twitch streamer is currently live and can access detailed stream information such as titles, games, view counts and more. Get G-Assist  Join the NVIDIA Developer Discord channel to collaborate, share creations and gain support from fellow AI enthusiasts and NVIDIA staff. the date for NVIDIA’s How to Build a G-Assist Plug-In webinar on Wednesday, July 9, from 10-11 a.m. PT, to learn more about Project G-Assist capabilities, discover the fundamentals of building, testing and deploying Project G-Assist plug-ins, and participate in a live Q&A session. Explore NVIDIA’s GitHub repository, which provides everything needed to get started developing with G-Assist, including sample plug-ins, step-by-step instructions and documentation for building custom functionalities. Learn more about the ChatGPT Plug-In Builder to transform ideas into functional G-Assist plug-ins with minimal coding. The tool uses OpenAI’s custom GPT builder to generate plug-in code and streamline the development process. NVIDIA’s technical blog walks through the architecture of a G-Assist plug-in, using a Twitch integration as an example. Discover how plug-ins work, how they communicate with G-Assist and how to build them from scratch. Each week, the RTX AI Garage blog series features community-driven AI innovations and content for those looking to learn more about NVIDIA NIM microservices and AI Blueprints, as well as building AI agents, creative workflows, digital humans, productivity apps and more on AI PCs and workstations.  Plug in to NVIDIA AI PC on Facebook, Instagram, TikTok and X — and stay informed by subscribing to the RTX AI PC newsletter. Follow NVIDIA Workstation on LinkedIn and X.  See notice regarding software product information. #plug #play #build #gassist #plugin
    BLOGS.NVIDIA.COM
    Plug and Play: Build a G-Assist Plug-In Today
    Project G-Assist — available through the NVIDIA App — is an experimental AI assistant that helps tune, control and optimize NVIDIA GeForce RTX systems. NVIDIA’s Plug and Play: Project G-Assist Plug-In Hackathon — running virtually through Wednesday, July 16 — invites the community to explore AI and build custom G-Assist plug-ins for a chance to win prizes and be featured on NVIDIA social media channels. G-Assist allows users to control their RTX GPU and other system settings using natural language, thanks to a small language model that runs on device. It can be used from the NVIDIA Overlay in the NVIDIA App without needing to tab out or switch programs. Users can expand its capabilities via plug-ins and even connect it to agentic frameworks such as Langflow. Below, find popular G-Assist plug-ins, hackathon details and tips to get started. Plug-In and Win Join the hackathon by registering and checking out the curated technical resources. G-Assist plug-ins can be built in several ways, including with Python for rapid development, with C++ for performance-critical apps and with custom system interactions for hardware and operating system automation. For those that prefer vibe coding, the G-Assist Plug-In Builder — a ChatGPT-based app that allows no-code or low-code development with natural language commands — makes it easy for enthusiasts to start creating plug-ins. To submit an entry, participants must provide a GitHub repository, including source code file (plugin.py), requirements.txt, manifest.json, config.json (if applicable), a plug-in executable file and READme code. Then, submit a video — between 30 seconds and two minutes — showcasing the plug-in in action. Finally, hackathoners must promote their plug-in using #AIonRTXHackathon on a social media channel: Instagram, TikTok or X. Submit projects via this form by Wednesday, July 16. Judges will assess plug-ins based on three main criteria: 1) innovation and creativity, 2) technical execution and integration, reviewing technical depth, G-Assist integration and scalability, and 3) usability and community impact, aka how easy it is to use the plug-in. Winners will be selected on Wednesday, Aug. 20. First place will receive a GeForce RTX 5090 laptop, second place a GeForce RTX 5080 GPU and third a GeForce RTX 5070 GPU. These top three will also be featured on NVIDIA’s social media channels, get the opportunity to meet the NVIDIA G-Assist team and earn an NVIDIA Deep Learning Institute self-paced course credit. Project G-Assist requires a GeForce RTX 50, 40 or 30 Series Desktop GPU with at least 12GB of VRAM, Windows 11 or 10 operating system, a compatible CPU (Intel Pentium G Series, Core i3, i5, i7 or higher; AMD FX, Ryzen 3, 5, 7, 9, Threadripper or higher), specific disk space requirements and a recent GeForce Game Ready Driver or NVIDIA Studio Driver. Plug-In(spiration) Explore open-source plug-in samples available on GitHub, which showcase the diverse ways on-device AI can enhance PC and gaming workflows. Popular plug-ins include: Google Gemini: Enables search-based queries using Google Search integration and large language model-based queries using Gemini capabilities in real time without needing to switch programs from the convenience of the NVIDIA App Overlay. Discord: Enables users to easily share game highlights or messages directly to Discord servers without disrupting gameplay. IFTTT: Lets users create automations across hundreds of compatible endpoints to trigger IoT routines — such as adjusting room lights and smart shades, or pushing the latest gaming news to a mobile device. Spotify: Lets users control Spotify using simple voice commands or the G-Assist interface to play favorite tracks and manage playlists. Twitch: Checks if any Twitch streamer is currently live and can access detailed stream information such as titles, games, view counts and more. Get G-Assist(ance)  Join the NVIDIA Developer Discord channel to collaborate, share creations and gain support from fellow AI enthusiasts and NVIDIA staff. Save the date for NVIDIA’s How to Build a G-Assist Plug-In webinar on Wednesday, July 9, from 10-11 a.m. PT, to learn more about Project G-Assist capabilities, discover the fundamentals of building, testing and deploying Project G-Assist plug-ins, and participate in a live Q&A session. Explore NVIDIA’s GitHub repository, which provides everything needed to get started developing with G-Assist, including sample plug-ins, step-by-step instructions and documentation for building custom functionalities. Learn more about the ChatGPT Plug-In Builder to transform ideas into functional G-Assist plug-ins with minimal coding. The tool uses OpenAI’s custom GPT builder to generate plug-in code and streamline the development process. NVIDIA’s technical blog walks through the architecture of a G-Assist plug-in, using a Twitch integration as an example. Discover how plug-ins work, how they communicate with G-Assist and how to build them from scratch. Each week, the RTX AI Garage blog series features community-driven AI innovations and content for those looking to learn more about NVIDIA NIM microservices and AI Blueprints, as well as building AI agents, creative workflows, digital humans, productivity apps and more on AI PCs and workstations.  Plug in to NVIDIA AI PC on Facebook, Instagram, TikTok and X — and stay informed by subscribing to the RTX AI PC newsletter. Follow NVIDIA Workstation on LinkedIn and X.  See notice regarding software product information.
    Like
    Wow
    Love
    Sad
    25
    0 Commentarios 0 Acciones
  • In a world where creativity reigns supreme, Adobe has just gifted us with a shiny new toy: the Firefly Boards. Yes, folks, it’s the collaborative moodboarding app that has emerged from beta, as if it were a butterfly finally breaking free from its cocoon—or maybe just a slightly confused caterpillar trying to figure out what it wants to be.

    Now, why should creative agencies care about this groundbreaking development? Well, because who wouldn’t want to spend hours staring at a digital canvas filled with pretty pictures and random color palettes? Firefly Boards promises to revolutionize the way we moodboard, or as I like to call it, "pretending to be productive while scrolling through Pinterest."

    Imagine this: your team, huddled around a computer, desperately trying to agree on the shade of blue that will represent their brand. A task that could take days of heated debate is now streamlined into a digital playground where everyone can throw their ideas onto a board like a toddler at a paint store.

    But let's be real. Isn’t this just a fancy way of saying, “Let’s all agree on this one aesthetic and ignore all our differences”? Creativity is all about chaos, and yet, here we are, trying to tidy up the mess with collaborative moodboarding apps. What’s next? A group hug to decide on the font size?

    Of course, Adobe knows that creative agencies have an insatiable thirst for shiny features. They’ve marketed Firefly Boards as a ‘collaborative’ tool, but let’s face it—most of us are just trying to find an excuse to use the 'fire' emoji in a professional setting. It’s as if they’re saying, “Trust us, this will make your life easier!” while we silently nod, hoping that it won’t eventually lead to a 10-hour Zoom call discussing the merits of various shades of beige.

    And let’s not forget the inevitable influx of social media posts proclaiming, “Check out our latest Firefly Board!” — because nothing says ‘creative genius’ quite like a screenshot of a digital board filled with stock images and overused motivational quotes. Can’t wait to see how many ‘likes’ that garners!

    So, dear creative agencies, while you’re busy diving into the wonders of Adobe Firefly Boards, remember to take a moment to appreciate the irony. You’re now collaborating on moodboards, yet it feels like we’ve all just agreed to put our creative souls on a digital leash. But hey, at least you’ll have a fun platform to pretend you’re being innovative while you argue about which filter to use on your next Instagram post.

    #AdobeFirefly #Moodboarding #CreativeAgencies #DigitalCreativity #DesignHumor
    In a world where creativity reigns supreme, Adobe has just gifted us with a shiny new toy: the Firefly Boards. Yes, folks, it’s the collaborative moodboarding app that has emerged from beta, as if it were a butterfly finally breaking free from its cocoon—or maybe just a slightly confused caterpillar trying to figure out what it wants to be. Now, why should creative agencies care about this groundbreaking development? Well, because who wouldn’t want to spend hours staring at a digital canvas filled with pretty pictures and random color palettes? Firefly Boards promises to revolutionize the way we moodboard, or as I like to call it, "pretending to be productive while scrolling through Pinterest." Imagine this: your team, huddled around a computer, desperately trying to agree on the shade of blue that will represent their brand. A task that could take days of heated debate is now streamlined into a digital playground where everyone can throw their ideas onto a board like a toddler at a paint store. But let's be real. Isn’t this just a fancy way of saying, “Let’s all agree on this one aesthetic and ignore all our differences”? Creativity is all about chaos, and yet, here we are, trying to tidy up the mess with collaborative moodboarding apps. What’s next? A group hug to decide on the font size? Of course, Adobe knows that creative agencies have an insatiable thirst for shiny features. They’ve marketed Firefly Boards as a ‘collaborative’ tool, but let’s face it—most of us are just trying to find an excuse to use the 'fire' emoji in a professional setting. It’s as if they’re saying, “Trust us, this will make your life easier!” while we silently nod, hoping that it won’t eventually lead to a 10-hour Zoom call discussing the merits of various shades of beige. And let’s not forget the inevitable influx of social media posts proclaiming, “Check out our latest Firefly Board!” — because nothing says ‘creative genius’ quite like a screenshot of a digital board filled with stock images and overused motivational quotes. Can’t wait to see how many ‘likes’ that garners! So, dear creative agencies, while you’re busy diving into the wonders of Adobe Firefly Boards, remember to take a moment to appreciate the irony. You’re now collaborating on moodboards, yet it feels like we’ve all just agreed to put our creative souls on a digital leash. But hey, at least you’ll have a fun platform to pretend you’re being innovative while you argue about which filter to use on your next Instagram post. #AdobeFirefly #Moodboarding #CreativeAgencies #DigitalCreativity #DesignHumor
    Why creative agencies need to know about new Adobe Firefly Boards
    The collaborative moodboarding app is now out of beta.
    Like
    Love
    Wow
    Angry
    Sad
    512
    1 Commentarios 0 Acciones
  • Apple Games, game app, developers, iPhone, iPad, Mac, gaming hub, discover titles, gaming experience

    ## Introduction

    In an era where mobile gaming dominates the entertainment landscape, Apple has made a strategic move to streamline game discovery across its platforms. The new initiative, Apple Games, aims to create a unified hub where players can engage with and discover titles seamlessly across iPhone, iPad, and Mac. This article will delve into what Apple Games brings to the table and how de...
    Apple Games, game app, developers, iPhone, iPad, Mac, gaming hub, discover titles, gaming experience ## Introduction In an era where mobile gaming dominates the entertainment landscape, Apple has made a strategic move to streamline game discovery across its platforms. The new initiative, Apple Games, aims to create a unified hub where players can engage with and discover titles seamlessly across iPhone, iPad, and Mac. This article will delve into what Apple Games brings to the table and how de...
    A New Way to Get Discovered: Apple Pitches Its New Game App to Developers
    Apple Games, game app, developers, iPhone, iPad, Mac, gaming hub, discover titles, gaming experience ## Introduction In an era where mobile gaming dominates the entertainment landscape, Apple has made a strategic move to streamline game discovery across its platforms. The new initiative, Apple Games, aims to create a unified hub where players can engage with and discover titles seamlessly...
    Like
    Love
    Wow
    Sad
    Angry
    613
    1 Commentarios 0 Acciones
  • Inside the thinking behind Frontify Futures' standout brand identity

    Who knows where branding will go in the future? However, for many of us working in the creative industries, it's our job to know. So it's something we need to start talking about, and Frontify Futures wants to be the platform where that conversation unfolds.
    This ambitious new thought leadership initiative from Frontify brings together an extraordinary coalition of voices—CMOs who've scaled global brands, creative leaders reimagining possibilities, strategy directors pioneering new approaches, and cultural forecasters mapping emerging opportunities—to explore how effectiveness, innovation, and scale will shape tomorrow's brand-building landscape.
    But Frontify Futures isn't just another content platform. Excitingly, from a design perspective, it's also a living experiment in what brand identity can become when technology meets craft, when systems embrace chaos, and when the future itself becomes a design material.
    Endless variation
    What makes Frontify Futures' typography unique isn't just its custom foundation: it's how that foundation enables endless variation and evolution. This was primarily achieved, reveals developer and digital art director Daniel Powell, by building bespoke tools for the project.

    "Rather than rely solely on streamlined tools built for speed and production, we started building our own," he explains. "The first was a node-based design tool that takes our custom Frame and Hairline fonts as a base and uses them as the foundations for our type generator. With it, we can generate unique type variations for each content strand—each article, even—and create both static and animated type, exportable as video or rendered live in the browser."
    Each of these tools included what Daniel calls a "chaos element: a small but intentional glitch in the system. A microstatement about the nature of the future: that it can be anticipated but never fully known. It's our way of keeping gesture alive inside the system."
    One of the clearest examples of this is the colour palette generator. "It samples from a dynamic photo grid tied to a rotating colour wheel that completes one full revolution per year," Daniel explains. "But here's the twist: wind speed and direction in St. Gallen, Switzerland—Frontify's HQ—nudges the wheel unpredictably off-centre. It's a subtle, living mechanic; each article contains a log of the wind data in its code as a kind of Easter Egg."

    Another favourite of Daniel's—yet to be released—is an expanded version of Conway's Game of Life. "It's been running continuously for over a month now, evolving patterns used in one of the content strand headers," he reveals. "The designer becomes a kind of photographer, capturing moments from a petri dish of generative motion."
    Core Philosophy
    In developing this unique identity, two phrases stood out to Daniel as guiding lights from the outset. The first was, 'We will show, not tell.'
    "This became the foundation for how we approached the identity," recalls Daniel. "It had to feel like a playground: open, experimental, and fluid. Not overly precious or prescriptive. A system the Frontify team could truly own, shape, and evolve. A platform, not a final product. A foundation, just as the future is always built on the past."

    The second guiding phrase, pulled directly from Frontify's rebrand materials, felt like "a call to action," says Daniel. "'Gestural and geometric. Human and machine. Art and science.' It's a tension that feels especially relevant in the creative industries today. As technology accelerates, we ask ourselves: how do we still hold onto our craft? What does it mean to be expressive in an increasingly systemised world?"
    Stripped back and skeletal typography
    The identity that Daniel and his team created reflects these themes through typography that literally embodies the platform's core philosophy. It really started from this idea of the past being built upon the 'foundations' of the past," he explains. "At the time Frontify Futures was being created, Frontify itself was going through a rebrand. With that, they'd started using a new variable typeface called Cranny, a custom cut of Azurio by Narrow Type."
    Daniel's team took Cranny and "pushed it into a stripped-back and almost skeletal take". The result was Crany-Frame and Crany-Hairline. "These fonts then served as our base scaffolding," he continues. "They were never seen in design, but instead, we applied decoration them to produce new typefaces for each content strand, giving the identity the space to grow and allow new ideas and shapes to form."

    As Daniel saw it, the demands on the typeface were pretty simple. "It needed to set an atmosphere. We needed it needed to feel alive. We wanted it to be something shifting and repositioning. And so, while we have a bunch of static cuts of each base style, we rarely use them; the typefaces you see on the website and social only exist at the moment as a string of parameters to create a general style that we use to create live animating versions of the font generated on the fly."
    In addition to setting the atmosphere, it needed to be extremely flexible and feature live inputs, as a significant part of the branding is about the unpredictability of the future. "So Daniel's team built in those aforementioned "chaos moments where everything from user interaction to live windspeeds can affect the font."
    Design Process
    The process of creating the typefaces is a fascinating one. "We started by working with the custom cut of Azuriofrom Narrow Type. We then redrew it to take inspiration from how a frame and a hairline could be produced from this original cut. From there, we built a type generation tool that uses them as a base.
    "It's a custom node-based system that lets us really get in there and play with the overlays for everything from grid-sizing, shapes and timing for the animation," he outlines. "We used this tool to design the variants for different content strands. We weren't just designing letterforms; we were designing a comprehensive toolset that could evolve in tandem with the content.
    "That became a big part of the process: designing systems that designers could actually use, not just look at; again, it was a wider conversation and concept around the future and how designers and machines can work together."

    In short, the evolution of the typeface system reflects the platform's broader commitment to continuous growth and adaptation." The whole idea was to make something open enough to keep building on," Daniel stresses. "We've already got tools in place to generate new weights, shapes and animated variants, and the tool itself still has a ton of unused functionality.
    "I can see that growing as new content strands emerge; we'll keep adapting the type with them," he adds. "It's less about version numbers and more about ongoing movement. The system's alive; that's the point.
    A provocation for the industry
    In this context, the Frontify Futures identity represents more than smart visual branding; it's also a manifesto for how creative systems might evolve in an age of increasing automation and systematisation. By building unpredictability into their tools, embracing the tension between human craft and machine precision, and creating systems that grow and adapt rather than merely scale, Daniel and the Frontify team have created something that feels genuinely forward-looking.
    For creatives grappling with similar questions about the future of their craft, Frontify Futures offers both inspiration and practical demonstration. It shows how brands can remain human while embracing technological capability, how systems can be both consistent and surprising, and how the future itself can become a creative medium.
    This clever approach suggests that the future of branding lies not in choosing between human creativity and systematic efficiency but in finding new ways to make them work together, creating something neither could achieve alone.
    #inside #thinking #behind #frontify #futures039
    Inside the thinking behind Frontify Futures' standout brand identity
    Who knows where branding will go in the future? However, for many of us working in the creative industries, it's our job to know. So it's something we need to start talking about, and Frontify Futures wants to be the platform where that conversation unfolds. This ambitious new thought leadership initiative from Frontify brings together an extraordinary coalition of voices—CMOs who've scaled global brands, creative leaders reimagining possibilities, strategy directors pioneering new approaches, and cultural forecasters mapping emerging opportunities—to explore how effectiveness, innovation, and scale will shape tomorrow's brand-building landscape. But Frontify Futures isn't just another content platform. Excitingly, from a design perspective, it's also a living experiment in what brand identity can become when technology meets craft, when systems embrace chaos, and when the future itself becomes a design material. Endless variation What makes Frontify Futures' typography unique isn't just its custom foundation: it's how that foundation enables endless variation and evolution. This was primarily achieved, reveals developer and digital art director Daniel Powell, by building bespoke tools for the project. "Rather than rely solely on streamlined tools built for speed and production, we started building our own," he explains. "The first was a node-based design tool that takes our custom Frame and Hairline fonts as a base and uses them as the foundations for our type generator. With it, we can generate unique type variations for each content strand—each article, even—and create both static and animated type, exportable as video or rendered live in the browser." Each of these tools included what Daniel calls a "chaos element: a small but intentional glitch in the system. A microstatement about the nature of the future: that it can be anticipated but never fully known. It's our way of keeping gesture alive inside the system." One of the clearest examples of this is the colour palette generator. "It samples from a dynamic photo grid tied to a rotating colour wheel that completes one full revolution per year," Daniel explains. "But here's the twist: wind speed and direction in St. Gallen, Switzerland—Frontify's HQ—nudges the wheel unpredictably off-centre. It's a subtle, living mechanic; each article contains a log of the wind data in its code as a kind of Easter Egg." Another favourite of Daniel's—yet to be released—is an expanded version of Conway's Game of Life. "It's been running continuously for over a month now, evolving patterns used in one of the content strand headers," he reveals. "The designer becomes a kind of photographer, capturing moments from a petri dish of generative motion." Core Philosophy In developing this unique identity, two phrases stood out to Daniel as guiding lights from the outset. The first was, 'We will show, not tell.' "This became the foundation for how we approached the identity," recalls Daniel. "It had to feel like a playground: open, experimental, and fluid. Not overly precious or prescriptive. A system the Frontify team could truly own, shape, and evolve. A platform, not a final product. A foundation, just as the future is always built on the past." The second guiding phrase, pulled directly from Frontify's rebrand materials, felt like "a call to action," says Daniel. "'Gestural and geometric. Human and machine. Art and science.' It's a tension that feels especially relevant in the creative industries today. As technology accelerates, we ask ourselves: how do we still hold onto our craft? What does it mean to be expressive in an increasingly systemised world?" Stripped back and skeletal typography The identity that Daniel and his team created reflects these themes through typography that literally embodies the platform's core philosophy. It really started from this idea of the past being built upon the 'foundations' of the past," he explains. "At the time Frontify Futures was being created, Frontify itself was going through a rebrand. With that, they'd started using a new variable typeface called Cranny, a custom cut of Azurio by Narrow Type." Daniel's team took Cranny and "pushed it into a stripped-back and almost skeletal take". The result was Crany-Frame and Crany-Hairline. "These fonts then served as our base scaffolding," he continues. "They were never seen in design, but instead, we applied decoration them to produce new typefaces for each content strand, giving the identity the space to grow and allow new ideas and shapes to form." As Daniel saw it, the demands on the typeface were pretty simple. "It needed to set an atmosphere. We needed it needed to feel alive. We wanted it to be something shifting and repositioning. And so, while we have a bunch of static cuts of each base style, we rarely use them; the typefaces you see on the website and social only exist at the moment as a string of parameters to create a general style that we use to create live animating versions of the font generated on the fly." In addition to setting the atmosphere, it needed to be extremely flexible and feature live inputs, as a significant part of the branding is about the unpredictability of the future. "So Daniel's team built in those aforementioned "chaos moments where everything from user interaction to live windspeeds can affect the font." Design Process The process of creating the typefaces is a fascinating one. "We started by working with the custom cut of Azuriofrom Narrow Type. We then redrew it to take inspiration from how a frame and a hairline could be produced from this original cut. From there, we built a type generation tool that uses them as a base. "It's a custom node-based system that lets us really get in there and play with the overlays for everything from grid-sizing, shapes and timing for the animation," he outlines. "We used this tool to design the variants for different content strands. We weren't just designing letterforms; we were designing a comprehensive toolset that could evolve in tandem with the content. "That became a big part of the process: designing systems that designers could actually use, not just look at; again, it was a wider conversation and concept around the future and how designers and machines can work together." In short, the evolution of the typeface system reflects the platform's broader commitment to continuous growth and adaptation." The whole idea was to make something open enough to keep building on," Daniel stresses. "We've already got tools in place to generate new weights, shapes and animated variants, and the tool itself still has a ton of unused functionality. "I can see that growing as new content strands emerge; we'll keep adapting the type with them," he adds. "It's less about version numbers and more about ongoing movement. The system's alive; that's the point. A provocation for the industry In this context, the Frontify Futures identity represents more than smart visual branding; it's also a manifesto for how creative systems might evolve in an age of increasing automation and systematisation. By building unpredictability into their tools, embracing the tension between human craft and machine precision, and creating systems that grow and adapt rather than merely scale, Daniel and the Frontify team have created something that feels genuinely forward-looking. For creatives grappling with similar questions about the future of their craft, Frontify Futures offers both inspiration and practical demonstration. It shows how brands can remain human while embracing technological capability, how systems can be both consistent and surprising, and how the future itself can become a creative medium. This clever approach suggests that the future of branding lies not in choosing between human creativity and systematic efficiency but in finding new ways to make them work together, creating something neither could achieve alone. #inside #thinking #behind #frontify #futures039
    WWW.CREATIVEBOOM.COM
    Inside the thinking behind Frontify Futures' standout brand identity
    Who knows where branding will go in the future? However, for many of us working in the creative industries, it's our job to know. So it's something we need to start talking about, and Frontify Futures wants to be the platform where that conversation unfolds. This ambitious new thought leadership initiative from Frontify brings together an extraordinary coalition of voices—CMOs who've scaled global brands, creative leaders reimagining possibilities, strategy directors pioneering new approaches, and cultural forecasters mapping emerging opportunities—to explore how effectiveness, innovation, and scale will shape tomorrow's brand-building landscape. But Frontify Futures isn't just another content platform. Excitingly, from a design perspective, it's also a living experiment in what brand identity can become when technology meets craft, when systems embrace chaos, and when the future itself becomes a design material. Endless variation What makes Frontify Futures' typography unique isn't just its custom foundation: it's how that foundation enables endless variation and evolution. This was primarily achieved, reveals developer and digital art director Daniel Powell, by building bespoke tools for the project. "Rather than rely solely on streamlined tools built for speed and production, we started building our own," he explains. "The first was a node-based design tool that takes our custom Frame and Hairline fonts as a base and uses them as the foundations for our type generator. With it, we can generate unique type variations for each content strand—each article, even—and create both static and animated type, exportable as video or rendered live in the browser." Each of these tools included what Daniel calls a "chaos element: a small but intentional glitch in the system. A microstatement about the nature of the future: that it can be anticipated but never fully known. It's our way of keeping gesture alive inside the system." One of the clearest examples of this is the colour palette generator. "It samples from a dynamic photo grid tied to a rotating colour wheel that completes one full revolution per year," Daniel explains. "But here's the twist: wind speed and direction in St. Gallen, Switzerland—Frontify's HQ—nudges the wheel unpredictably off-centre. It's a subtle, living mechanic; each article contains a log of the wind data in its code as a kind of Easter Egg." Another favourite of Daniel's—yet to be released—is an expanded version of Conway's Game of Life. "It's been running continuously for over a month now, evolving patterns used in one of the content strand headers," he reveals. "The designer becomes a kind of photographer, capturing moments from a petri dish of generative motion." Core Philosophy In developing this unique identity, two phrases stood out to Daniel as guiding lights from the outset. The first was, 'We will show, not tell.' "This became the foundation for how we approached the identity," recalls Daniel. "It had to feel like a playground: open, experimental, and fluid. Not overly precious or prescriptive. A system the Frontify team could truly own, shape, and evolve. A platform, not a final product. A foundation, just as the future is always built on the past." The second guiding phrase, pulled directly from Frontify's rebrand materials, felt like "a call to action," says Daniel. "'Gestural and geometric. Human and machine. Art and science.' It's a tension that feels especially relevant in the creative industries today. As technology accelerates, we ask ourselves: how do we still hold onto our craft? What does it mean to be expressive in an increasingly systemised world?" Stripped back and skeletal typography The identity that Daniel and his team created reflects these themes through typography that literally embodies the platform's core philosophy. It really started from this idea of the past being built upon the 'foundations' of the past," he explains. "At the time Frontify Futures was being created, Frontify itself was going through a rebrand. With that, they'd started using a new variable typeface called Cranny, a custom cut of Azurio by Narrow Type." Daniel's team took Cranny and "pushed it into a stripped-back and almost skeletal take". The result was Crany-Frame and Crany-Hairline. "These fonts then served as our base scaffolding," he continues. "They were never seen in design, but instead, we applied decoration them to produce new typefaces for each content strand, giving the identity the space to grow and allow new ideas and shapes to form." As Daniel saw it, the demands on the typeface were pretty simple. "It needed to set an atmosphere. We needed it needed to feel alive. We wanted it to be something shifting and repositioning. And so, while we have a bunch of static cuts of each base style, we rarely use them; the typefaces you see on the website and social only exist at the moment as a string of parameters to create a general style that we use to create live animating versions of the font generated on the fly." In addition to setting the atmosphere, it needed to be extremely flexible and feature live inputs, as a significant part of the branding is about the unpredictability of the future. "So Daniel's team built in those aforementioned "chaos moments where everything from user interaction to live windspeeds can affect the font." Design Process The process of creating the typefaces is a fascinating one. "We started by working with the custom cut of Azurio (Cranny) from Narrow Type. We then redrew it to take inspiration from how a frame and a hairline could be produced from this original cut. From there, we built a type generation tool that uses them as a base. "It's a custom node-based system that lets us really get in there and play with the overlays for everything from grid-sizing, shapes and timing for the animation," he outlines. "We used this tool to design the variants for different content strands. We weren't just designing letterforms; we were designing a comprehensive toolset that could evolve in tandem with the content. "That became a big part of the process: designing systems that designers could actually use, not just look at; again, it was a wider conversation and concept around the future and how designers and machines can work together." In short, the evolution of the typeface system reflects the platform's broader commitment to continuous growth and adaptation." The whole idea was to make something open enough to keep building on," Daniel stresses. "We've already got tools in place to generate new weights, shapes and animated variants, and the tool itself still has a ton of unused functionality. "I can see that growing as new content strands emerge; we'll keep adapting the type with them," he adds. "It's less about version numbers and more about ongoing movement. The system's alive; that's the point. A provocation for the industry In this context, the Frontify Futures identity represents more than smart visual branding; it's also a manifesto for how creative systems might evolve in an age of increasing automation and systematisation. By building unpredictability into their tools, embracing the tension between human craft and machine precision, and creating systems that grow and adapt rather than merely scale, Daniel and the Frontify team have created something that feels genuinely forward-looking. For creatives grappling with similar questions about the future of their craft, Frontify Futures offers both inspiration and practical demonstration. It shows how brands can remain human while embracing technological capability, how systems can be both consistent and surprising, and how the future itself can become a creative medium. This clever approach suggests that the future of branding lies not in choosing between human creativity and systematic efficiency but in finding new ways to make them work together, creating something neither could achieve alone.
    0 Commentarios 0 Acciones
  • EasyDMARC Integrates With Pax8 Marketplace To Simplify Email Security For MSPs

    Originally published at EasyDMARC Integrates With Pax8 Marketplace To Simplify Email Security For MSPs by Anush Yolyan.

    The integration will deliver simple, accessible, and streamlined email security for vulnerable inboxes

    Global, 4 November 2024 – US-based email security firm EasyDMARC has today announced its integration with Pax8 Marketplace, the leading cloud commerce marketplace. As one of the first DMARC solution providers on the Pax8 Marketplace, EasyDMARC is expanding its mission to protect inboxes from the rising threat of phishing attacks with a rigorous, user-friendly DMARC solution.

    The integration comes as Google highlights the impressive results of recently implemented email authentication measures for bulk senders: a 65% reduction in unauthenticated messages to Gmail users, a 50% increase in bulk senders following best security practices, and 265 billion fewer unauthenticated messages sent in 2024. With email being such a crucial communication channel for businesses, email authentication measures are an essential part of any business’s cybersecurity offering. 

    Key features of the integration include:

    Centralized billing

    With centralized billing, customers can now streamline their cloud services under a single pane of glass, simplifying the management and billing of their EasyDMARC solution. This consolidated approach enables partners to reduce administrative complexity and manage all cloud expenses through one interface, providing a seamless billing and support experience.

    Automated provisioning 

    Through automated provisioning, Pax8’s automation capabilities make deploying DMARC across client accounts quick and hassle-free. By eliminating manual configurations, this integration ensures that customers can implement email security solutions rapidly, allowing them to safeguard client inboxes without delay.

    Bundled offerings

    The bundled offerings available through Pax8 allow partners to enhance their service portfolios by combining EasyDMARC with complementary security solutions. By creating all-in-one security packages, partners can offer their clients more robust protection, addressing a broader range of security needs from a single, trusted platform.

    Gerasim Hovhannisyan, Co-Founder and CEO of EasyDMARC, said:

    “We’re thrilled to be working with Pax8  to provide MSPs with a streamlined, effective way to deliver top-tier email security to their clients, all within a platform that equips them with everything needed to stay secure.  As phishing attacks grow in frequency and sophistication, businesses can no longer afford to overlook the importance of email security. Email authentication is a vital defense against the evolving threat of phishing and is crucial in preserving the integrity of email communication. This integration is designed to allow businesses of all sizes to benefit from DMARC’s extensive capabilities.”

    Ryan Burton, Vice President of Marketplace Vendor Strategy, at Pax8 said: 

    “We’re delighted to welcome EasyDMARC to the Pax8 Marketplace as an enterprise-class DMARC solution provider. This integration gives MSPs the tools they need to meet the growing demand for email security, with simplified deployment, billing, and bundling benefits. With EasyDMARC’s technical capabilities and intelligence, MSPs can deliver robust protection against phishing threats without the technical hassle that often holds businesses back.”

    About EasyDMARC

    EasyDMARC is a cloud-native B2B SaaS solution that addresses email security and deliverability problems with just a few clicks. For Managed Service Providers seeking to increase their revenue, EasyDMARC presents an ideal solution. The email authentication platform streamlines domain management, providing capabilities such as organizational control, domain grouping, and access management.

    Additionally, EasyDMARC offers a comprehensive sales and marketing enablement program designed to boost DMARC sales. All of these features are available for MSPs on a scalable platform with a flexible pay-as-you-go pricing model.

    For more information on the EasyDMARC, visit: /

    About Pax8 

    Pax8 is the technology marketplace of the future, linking partners, vendors, and small to midsized businessesthrough AI-powered insights and comprehensive product support. With a global partner ecosystem of over 38,000 managed service providers, Pax8 empowers SMBs worldwide by providing software and services that unlock their growth potential and enhance their security. Committed to innovating cloud commerce at scale, Pax8 drives customer acquisition and solution consumption across its entire ecosystem.

    Find out more: /

    The post EasyDMARC Integrates With Pax8 Marketplace To Simplify Email Security For MSPs appeared first on EasyDMARC.
    #easydmarc #integrates #with #pax8 #marketplace
    EasyDMARC Integrates With Pax8 Marketplace To Simplify Email Security For MSPs
    Originally published at EasyDMARC Integrates With Pax8 Marketplace To Simplify Email Security For MSPs by Anush Yolyan. The integration will deliver simple, accessible, and streamlined email security for vulnerable inboxes Global, 4 November 2024 – US-based email security firm EasyDMARC has today announced its integration with Pax8 Marketplace, the leading cloud commerce marketplace. As one of the first DMARC solution providers on the Pax8 Marketplace, EasyDMARC is expanding its mission to protect inboxes from the rising threat of phishing attacks with a rigorous, user-friendly DMARC solution. The integration comes as Google highlights the impressive results of recently implemented email authentication measures for bulk senders: a 65% reduction in unauthenticated messages to Gmail users, a 50% increase in bulk senders following best security practices, and 265 billion fewer unauthenticated messages sent in 2024. With email being such a crucial communication channel for businesses, email authentication measures are an essential part of any business’s cybersecurity offering.  Key features of the integration include: Centralized billing With centralized billing, customers can now streamline their cloud services under a single pane of glass, simplifying the management and billing of their EasyDMARC solution. This consolidated approach enables partners to reduce administrative complexity and manage all cloud expenses through one interface, providing a seamless billing and support experience. Automated provisioning  Through automated provisioning, Pax8’s automation capabilities make deploying DMARC across client accounts quick and hassle-free. By eliminating manual configurations, this integration ensures that customers can implement email security solutions rapidly, allowing them to safeguard client inboxes without delay. Bundled offerings The bundled offerings available through Pax8 allow partners to enhance their service portfolios by combining EasyDMARC with complementary security solutions. By creating all-in-one security packages, partners can offer their clients more robust protection, addressing a broader range of security needs from a single, trusted platform. Gerasim Hovhannisyan, Co-Founder and CEO of EasyDMARC, said: “We’re thrilled to be working with Pax8  to provide MSPs with a streamlined, effective way to deliver top-tier email security to their clients, all within a platform that equips them with everything needed to stay secure.  As phishing attacks grow in frequency and sophistication, businesses can no longer afford to overlook the importance of email security. Email authentication is a vital defense against the evolving threat of phishing and is crucial in preserving the integrity of email communication. This integration is designed to allow businesses of all sizes to benefit from DMARC’s extensive capabilities.” Ryan Burton, Vice President of Marketplace Vendor Strategy, at Pax8 said:  “We’re delighted to welcome EasyDMARC to the Pax8 Marketplace as an enterprise-class DMARC solution provider. This integration gives MSPs the tools they need to meet the growing demand for email security, with simplified deployment, billing, and bundling benefits. With EasyDMARC’s technical capabilities and intelligence, MSPs can deliver robust protection against phishing threats without the technical hassle that often holds businesses back.” About EasyDMARC EasyDMARC is a cloud-native B2B SaaS solution that addresses email security and deliverability problems with just a few clicks. For Managed Service Providers seeking to increase their revenue, EasyDMARC presents an ideal solution. The email authentication platform streamlines domain management, providing capabilities such as organizational control, domain grouping, and access management. Additionally, EasyDMARC offers a comprehensive sales and marketing enablement program designed to boost DMARC sales. All of these features are available for MSPs on a scalable platform with a flexible pay-as-you-go pricing model. For more information on the EasyDMARC, visit: / About Pax8  Pax8 is the technology marketplace of the future, linking partners, vendors, and small to midsized businessesthrough AI-powered insights and comprehensive product support. With a global partner ecosystem of over 38,000 managed service providers, Pax8 empowers SMBs worldwide by providing software and services that unlock their growth potential and enhance their security. Committed to innovating cloud commerce at scale, Pax8 drives customer acquisition and solution consumption across its entire ecosystem. Find out more: / The post EasyDMARC Integrates With Pax8 Marketplace To Simplify Email Security For MSPs appeared first on EasyDMARC. #easydmarc #integrates #with #pax8 #marketplace
    EASYDMARC.COM
    EasyDMARC Integrates With Pax8 Marketplace To Simplify Email Security For MSPs
    Originally published at EasyDMARC Integrates With Pax8 Marketplace To Simplify Email Security For MSPs by Anush Yolyan. The integration will deliver simple, accessible, and streamlined email security for vulnerable inboxes Global, 4 November 2024 – US-based email security firm EasyDMARC has today announced its integration with Pax8 Marketplace, the leading cloud commerce marketplace. As one of the first DMARC solution providers on the Pax8 Marketplace, EasyDMARC is expanding its mission to protect inboxes from the rising threat of phishing attacks with a rigorous, user-friendly DMARC solution. The integration comes as Google highlights the impressive results of recently implemented email authentication measures for bulk senders: a 65% reduction in unauthenticated messages to Gmail users, a 50% increase in bulk senders following best security practices, and 265 billion fewer unauthenticated messages sent in 2024. With email being such a crucial communication channel for businesses, email authentication measures are an essential part of any business’s cybersecurity offering.  Key features of the integration include: Centralized billing With centralized billing, customers can now streamline their cloud services under a single pane of glass, simplifying the management and billing of their EasyDMARC solution. This consolidated approach enables partners to reduce administrative complexity and manage all cloud expenses through one interface, providing a seamless billing and support experience. Automated provisioning  Through automated provisioning, Pax8’s automation capabilities make deploying DMARC across client accounts quick and hassle-free. By eliminating manual configurations, this integration ensures that customers can implement email security solutions rapidly, allowing them to safeguard client inboxes without delay. Bundled offerings The bundled offerings available through Pax8 allow partners to enhance their service portfolios by combining EasyDMARC with complementary security solutions. By creating all-in-one security packages, partners can offer their clients more robust protection, addressing a broader range of security needs from a single, trusted platform. Gerasim Hovhannisyan, Co-Founder and CEO of EasyDMARC, said: “We’re thrilled to be working with Pax8  to provide MSPs with a streamlined, effective way to deliver top-tier email security to their clients, all within a platform that equips them with everything needed to stay secure.  As phishing attacks grow in frequency and sophistication, businesses can no longer afford to overlook the importance of email security. Email authentication is a vital defense against the evolving threat of phishing and is crucial in preserving the integrity of email communication. This integration is designed to allow businesses of all sizes to benefit from DMARC’s extensive capabilities.” Ryan Burton, Vice President of Marketplace Vendor Strategy, at Pax8 said:  “We’re delighted to welcome EasyDMARC to the Pax8 Marketplace as an enterprise-class DMARC solution provider. This integration gives MSPs the tools they need to meet the growing demand for email security, with simplified deployment, billing, and bundling benefits. With EasyDMARC’s technical capabilities and intelligence, MSPs can deliver robust protection against phishing threats without the technical hassle that often holds businesses back.” About EasyDMARC EasyDMARC is a cloud-native B2B SaaS solution that addresses email security and deliverability problems with just a few clicks. For Managed Service Providers seeking to increase their revenue, EasyDMARC presents an ideal solution. The email authentication platform streamlines domain management, providing capabilities such as organizational control, domain grouping, and access management. Additionally, EasyDMARC offers a comprehensive sales and marketing enablement program designed to boost DMARC sales. All of these features are available for MSPs on a scalable platform with a flexible pay-as-you-go pricing model. For more information on the EasyDMARC, visit: https://easydmarc.com/ About Pax8  Pax8 is the technology marketplace of the future, linking partners, vendors, and small to midsized businesses (SMBs) through AI-powered insights and comprehensive product support. With a global partner ecosystem of over 38,000 managed service providers, Pax8 empowers SMBs worldwide by providing software and services that unlock their growth potential and enhance their security. Committed to innovating cloud commerce at scale, Pax8 drives customer acquisition and solution consumption across its entire ecosystem. Find out more: https://www.pax8.com/en-us/ The post EasyDMARC Integrates With Pax8 Marketplace To Simplify Email Security For MSPs appeared first on EasyDMARC.
    0 Commentarios 0 Acciones
  • MedTech AI, hardware, and clinical application programmes

    Modern healthcare innovations span AI, devices, software, images, and regulatory frameworks, all requiring stringent coordination. Generative AI arguably has the strongest transformative potential in healthcare technology programmes, with it already being applied across various domains, such as R&D, commercial operations, and supply chain management.Traditional models for medical appointments, like face-to-face appointments, and paper-based processes may not be sufficient to meet the fast-paced, data-driven medical landscape of today. Therefore, healthcare professionals and patients are seeking more convenient and efficient ways to access and share information, meeting the complex standards of modern medical science. According to McKinsey, Medtech companies are at the forefront of healthcare innovation, estimating they could capture between billion and billion annually in productivity gains. Through GenAI adoption, an additional billion plus in revenue is estimated from products and service innovations. A McKinsey 2024 survey revealed around two thirds of Medtech executives have already implemented Gen AI, with approximately 20% scaling their solutions up and reporting substantial benefits to productivity.  While advanced technology implementation is growing across the medical industry, challenges persist. Organisations face hurdles like data integration issues, decentralised strategies, and skill gaps. Together, these highlight a need for a more streamlined approach to Gen AI deployment. Of all the Medtech domains, R&D is leading the way in Gen AI adoption. Being the most comfortable with new technologies, R&D departments use Gen AI tools to streamline work processes, such as summarising research papers or scientific articles, highlighting a grassroots adoption trend. Individual researchers are using AI to enhance productivity, even when no formal company-wide strategies are in place.While AI tools automate and accelerate R&D tasks, human review is still required to ensure final submissions are correct and satisfactory. Gen AI is proving to reduce time spent on administrative tasks for teams and improve research accuracy and depth, with some companies experiencing 20% to 30% gains in research productivity. KPIs for success in healthcare product programmesMeasuring business performance is essential in the healthcare sector. The number one goal is, of course, to deliver high-quality care, yet simultaneously maintain efficient operations. By measuring and analysing KPIs, healthcare providers are in a better position to improve patient outcomes through their data-based considerations. KPIs can also improve resource allocation, and encourage continuous improvement in all areas of care. In terms of healthcare product programmes, these structured initiatives prioritise the development, delivery, and continual optimisation of medical products. But to be a success, they require cross-functional coordination of clinical, technical, regulatory, and business teams. Time to market is critical, ensuring a product moves from the concept stage to launch as quickly as possible.Of particular note is the emphasis needing to be placed on labelling and documentation. McKinsey notes that AI-assisted labelling has resulted in a 20%-30% improvement in operational efficiency. Resource utilisation rates are also important, showing how efficiently time, budget, and/or headcount are used during the developmental stage of products. In the healthcare sector, KPIs ought to focus on several factors, including operational efficiency, patient outcomes, financial health of the business, and patient satisfaction. To achieve a comprehensive view of performance, these can be categorised into financial, operational, clinical quality, and patient experience.Bridging user experience with technical precision – design awardsInnovation is no longer solely judged by technical performance with user experiencebeing equally important. Some of the latest innovations in healthcare are recognised at the UX Design Awards, products that exemplify the best in user experience as well as technical precision. Top products prioritise the needs and experiences of both patients and healthcare professionals, also ensuring each product meets the rigorous clinical and regulatory standards of the sector. One example is the CIARTIC Move by Siemens Healthineers, a self-driving 3D C-arm imaging system that lets surgeons operate, controlling the device wirelessly in a sterile field. Computer hardware company ASUS has also received accolades for its HealthConnect App and VivoWatch Series, showcasing the fusion of AIoT-driven smart healthcare solutions with user-friendly interfaces – sometimes in what are essentially consumer devices. This demonstrates how technical innovation is being made accessible and becoming increasingly intuitive as patients gain technical fluency.  Navigating regulatory and product development pathways simultaneously The establishing of clinical and regulatory paths is important, as this enables healthcare teams to feed a twin stream of findings back into development. Gen AI adoption has become a transformative approach, automating the production and refining of complex documents, mixed data sets, and structured and unstructured data. By integrating regulatory considerations early and adopting technologies like Gen AI as part of agile practices, healthcare product programmes help teams navigate a regulatory landscape that can often shift. Baking a regulatory mindset into a team early helps ensure compliance and continued innovation. Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is co-located with other leading events including Intelligent Automation Conference, BlockX, Digital Transformation Week, and Cyber Security & Cloud Expo.Explore other upcoming enterprise technology events and webinars powered by TechForge here.
    #medtech #hardware #clinical #application #programmes
    MedTech AI, hardware, and clinical application programmes
    Modern healthcare innovations span AI, devices, software, images, and regulatory frameworks, all requiring stringent coordination. Generative AI arguably has the strongest transformative potential in healthcare technology programmes, with it already being applied across various domains, such as R&D, commercial operations, and supply chain management.Traditional models for medical appointments, like face-to-face appointments, and paper-based processes may not be sufficient to meet the fast-paced, data-driven medical landscape of today. Therefore, healthcare professionals and patients are seeking more convenient and efficient ways to access and share information, meeting the complex standards of modern medical science. According to McKinsey, Medtech companies are at the forefront of healthcare innovation, estimating they could capture between billion and billion annually in productivity gains. Through GenAI adoption, an additional billion plus in revenue is estimated from products and service innovations. A McKinsey 2024 survey revealed around two thirds of Medtech executives have already implemented Gen AI, with approximately 20% scaling their solutions up and reporting substantial benefits to productivity.  While advanced technology implementation is growing across the medical industry, challenges persist. Organisations face hurdles like data integration issues, decentralised strategies, and skill gaps. Together, these highlight a need for a more streamlined approach to Gen AI deployment. Of all the Medtech domains, R&D is leading the way in Gen AI adoption. Being the most comfortable with new technologies, R&D departments use Gen AI tools to streamline work processes, such as summarising research papers or scientific articles, highlighting a grassroots adoption trend. Individual researchers are using AI to enhance productivity, even when no formal company-wide strategies are in place.While AI tools automate and accelerate R&D tasks, human review is still required to ensure final submissions are correct and satisfactory. Gen AI is proving to reduce time spent on administrative tasks for teams and improve research accuracy and depth, with some companies experiencing 20% to 30% gains in research productivity. KPIs for success in healthcare product programmesMeasuring business performance is essential in the healthcare sector. The number one goal is, of course, to deliver high-quality care, yet simultaneously maintain efficient operations. By measuring and analysing KPIs, healthcare providers are in a better position to improve patient outcomes through their data-based considerations. KPIs can also improve resource allocation, and encourage continuous improvement in all areas of care. In terms of healthcare product programmes, these structured initiatives prioritise the development, delivery, and continual optimisation of medical products. But to be a success, they require cross-functional coordination of clinical, technical, regulatory, and business teams. Time to market is critical, ensuring a product moves from the concept stage to launch as quickly as possible.Of particular note is the emphasis needing to be placed on labelling and documentation. McKinsey notes that AI-assisted labelling has resulted in a 20%-30% improvement in operational efficiency. Resource utilisation rates are also important, showing how efficiently time, budget, and/or headcount are used during the developmental stage of products. In the healthcare sector, KPIs ought to focus on several factors, including operational efficiency, patient outcomes, financial health of the business, and patient satisfaction. To achieve a comprehensive view of performance, these can be categorised into financial, operational, clinical quality, and patient experience.Bridging user experience with technical precision – design awardsInnovation is no longer solely judged by technical performance with user experiencebeing equally important. Some of the latest innovations in healthcare are recognised at the UX Design Awards, products that exemplify the best in user experience as well as technical precision. Top products prioritise the needs and experiences of both patients and healthcare professionals, also ensuring each product meets the rigorous clinical and regulatory standards of the sector. One example is the CIARTIC Move by Siemens Healthineers, a self-driving 3D C-arm imaging system that lets surgeons operate, controlling the device wirelessly in a sterile field. Computer hardware company ASUS has also received accolades for its HealthConnect App and VivoWatch Series, showcasing the fusion of AIoT-driven smart healthcare solutions with user-friendly interfaces – sometimes in what are essentially consumer devices. This demonstrates how technical innovation is being made accessible and becoming increasingly intuitive as patients gain technical fluency.  Navigating regulatory and product development pathways simultaneously The establishing of clinical and regulatory paths is important, as this enables healthcare teams to feed a twin stream of findings back into development. Gen AI adoption has become a transformative approach, automating the production and refining of complex documents, mixed data sets, and structured and unstructured data. By integrating regulatory considerations early and adopting technologies like Gen AI as part of agile practices, healthcare product programmes help teams navigate a regulatory landscape that can often shift. Baking a regulatory mindset into a team early helps ensure compliance and continued innovation. Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is co-located with other leading events including Intelligent Automation Conference, BlockX, Digital Transformation Week, and Cyber Security & Cloud Expo.Explore other upcoming enterprise technology events and webinars powered by TechForge here. #medtech #hardware #clinical #application #programmes
    WWW.ARTIFICIALINTELLIGENCE-NEWS.COM
    MedTech AI, hardware, and clinical application programmes
    Modern healthcare innovations span AI, devices, software, images, and regulatory frameworks, all requiring stringent coordination. Generative AI arguably has the strongest transformative potential in healthcare technology programmes, with it already being applied across various domains, such as R&D, commercial operations, and supply chain management.Traditional models for medical appointments, like face-to-face appointments, and paper-based processes may not be sufficient to meet the fast-paced, data-driven medical landscape of today. Therefore, healthcare professionals and patients are seeking more convenient and efficient ways to access and share information, meeting the complex standards of modern medical science. According to McKinsey, Medtech companies are at the forefront of healthcare innovation, estimating they could capture between $14 billion and $55 billion annually in productivity gains. Through GenAI adoption, an additional $50 billion plus in revenue is estimated from products and service innovations. A McKinsey 2024 survey revealed around two thirds of Medtech executives have already implemented Gen AI, with approximately 20% scaling their solutions up and reporting substantial benefits to productivity.  While advanced technology implementation is growing across the medical industry, challenges persist. Organisations face hurdles like data integration issues, decentralised strategies, and skill gaps. Together, these highlight a need for a more streamlined approach to Gen AI deployment. Of all the Medtech domains, R&D is leading the way in Gen AI adoption. Being the most comfortable with new technologies, R&D departments use Gen AI tools to streamline work processes, such as summarising research papers or scientific articles, highlighting a grassroots adoption trend. Individual researchers are using AI to enhance productivity, even when no formal company-wide strategies are in place.While AI tools automate and accelerate R&D tasks, human review is still required to ensure final submissions are correct and satisfactory. Gen AI is proving to reduce time spent on administrative tasks for teams and improve research accuracy and depth, with some companies experiencing 20% to 30% gains in research productivity. KPIs for success in healthcare product programmesMeasuring business performance is essential in the healthcare sector. The number one goal is, of course, to deliver high-quality care, yet simultaneously maintain efficient operations. By measuring and analysing KPIs, healthcare providers are in a better position to improve patient outcomes through their data-based considerations. KPIs can also improve resource allocation, and encourage continuous improvement in all areas of care. In terms of healthcare product programmes, these structured initiatives prioritise the development, delivery, and continual optimisation of medical products. But to be a success, they require cross-functional coordination of clinical, technical, regulatory, and business teams. Time to market is critical, ensuring a product moves from the concept stage to launch as quickly as possible.Of particular note is the emphasis needing to be placed on labelling and documentation. McKinsey notes that AI-assisted labelling has resulted in a 20%-30% improvement in operational efficiency. Resource utilisation rates are also important, showing how efficiently time, budget, and/or headcount are used during the developmental stage of products. In the healthcare sector, KPIs ought to focus on several factors, including operational efficiency, patient outcomes, financial health of the business, and patient satisfaction. To achieve a comprehensive view of performance, these can be categorised into financial, operational, clinical quality, and patient experience.Bridging user experience with technical precision – design awardsInnovation is no longer solely judged by technical performance with user experience (UX) being equally important. Some of the latest innovations in healthcare are recognised at the UX Design Awards, products that exemplify the best in user experience as well as technical precision. Top products prioritise the needs and experiences of both patients and healthcare professionals, also ensuring each product meets the rigorous clinical and regulatory standards of the sector. One example is the CIARTIC Move by Siemens Healthineers, a self-driving 3D C-arm imaging system that lets surgeons operate, controlling the device wirelessly in a sterile field. Computer hardware company ASUS has also received accolades for its HealthConnect App and VivoWatch Series, showcasing the fusion of AIoT-driven smart healthcare solutions with user-friendly interfaces – sometimes in what are essentially consumer devices. This demonstrates how technical innovation is being made accessible and becoming increasingly intuitive as patients gain technical fluency.  Navigating regulatory and product development pathways simultaneously The establishing of clinical and regulatory paths is important, as this enables healthcare teams to feed a twin stream of findings back into development. Gen AI adoption has become a transformative approach, automating the production and refining of complex documents, mixed data sets, and structured and unstructured data. By integrating regulatory considerations early and adopting technologies like Gen AI as part of agile practices, healthcare product programmes help teams navigate a regulatory landscape that can often shift. Baking a regulatory mindset into a team early helps ensure compliance and continued innovation. (Image source: “IBM Achieves New Deep Learning Breakthrough” by IBM Research is licensed under CC BY-ND 2.0.)Want to learn more about AI and big data from industry leaders? Check out AI & Big Data Expo taking place in Amsterdam, California, and London. The comprehensive event is co-located with other leading events including Intelligent Automation Conference, BlockX, Digital Transformation Week, and Cyber Security & Cloud Expo.Explore other upcoming enterprise technology events and webinars powered by TechForge here.
    0 Commentarios 0 Acciones
  • Mirela Cialai Q&A: Customer Engagement Book Interview

    Reading Time: 9 minutes
    In the ever-evolving landscape of customer engagement, staying ahead of the curve is not just advantageous, it’s essential.
    That’s why, for Chapter 7 of “The Customer Engagement Book: Adapt or Die,” we sat down with Mirela Cialai, a seasoned expert in CRM and Martech strategies at brands like Equinox. Mirela brings a wealth of knowledge in aligning technology roadmaps with business goals, shifting organizational focuses from acquisition to retention, and leveraging hyper-personalization to drive success.
    In this interview, Mirela dives deep into building robust customer engagement technology roadmaps. She unveils the “PAPER” framework—Plan, Audit, Prioritize, Execute, Refine—a simple yet effective strategy for marketers.
    You’ll gain insights into identifying gaps in your Martech stack, ensuring data accuracy, and prioritizing initiatives that deliver the greatest impact and ROI.
    Whether you’re navigating data silos, striving for cross-functional alignment, or aiming for seamless tech integration, Mirela’s expertise provides practical solutions and actionable takeaways.

     
    Mirela Cialai Q&A Interview
    1. How do you define the vision for a customer engagement platform roadmap in alignment with the broader business goals? Can you share any examples of successful visions from your experience?

    Defining the vision for the roadmap in alignment with the broader business goals involves creating a strategic framework that connects the team’s objectives with the organization’s overarching mission or primary objectives.

    This could be revenue growth, customer retention, market expansion, or operational efficiency.
    We then break down these goals into actionable areas where the team can contribute, such as improving engagement, increasing lifetime value, or driving acquisition.
    We articulate how the team will support business goals by defining the KPIs that link CRM outcomes — the team’s outcomes — to business goals.
    In a previous role, the CRM team I was leading faced significant challenges due to the lack of attribution capabilities and a reliance on surface-level metrics such as open rates and click-through rates to measure performance.
    This approach made it difficult to quantify the impact of our efforts on broader business objectives such as revenue growth.
    Recognizing this gap, I worked on defining a vision for the CRM team to address these shortcomings.
    Our vision was to drive measurable growth through enhanced data accuracy and improved attribution capabilities, which allowed us to deliver targeted, data-driven, and personalized customer experiences.
    To bring this vision to life, I developed a roadmap that focused on first improving data accuracy, building our attribution capabilities, and delivering personalization at scale.

    By aligning the vision with these strategic priorities, we were able to demonstrate the tangible impact of our efforts on the key business goals.

    2. What steps did you take to ensure data accuracy?
    The data team was very diligent in ensuring that our data warehouse had accurate data.
    So taking that as the source of truth, we started cleaning the data in all the other platforms that were integrated with our data warehouse — our CRM platform, our attribution analytics platform, etc.

    That’s where we started, looking at all the different integrations and ensuring that the data flows were correct and that we had all the right flows in place. And also validating and cleaning our email database — that helped, having more accurate data.

    3. How do you recommend shifting organizational focus from acquisition to retention within a customer engagement strategy?
    Shifting an organization’s focus from acquisition to retention requires a cultural and strategic shift, emphasizing the immense value that existing customers bring to long-term growth and profitability.
    I would start by quantifying the value of retention, showcasing how retaining customers is significantly more cost-effective than acquiring new ones. Research consistently shows that increasing retention rates by just 5% can boost profits by at least 25 to 95%.
    This data helps make a compelling case to stakeholders about the importance of prioritizing retention.
    Next, I would link retention to core business goals by demonstrating how enhancing customer lifetime value and loyalty can directly drive revenue growth.
    This involves shifting the organization’s focus to retention-specific metrics such as churn rate, repeat purchase rate, and customer LTV. These metrics provide actionable insights into customer behaviors and highlight the financial impact of retention initiatives, ensuring alignment with the broader company objectives.

    By framing retention as a driver of sustainable growth, the organization can see it not as a competing priority, but as a complementary strategy to acquisition, ultimately leading to a more balanced and effective customer engagement strategy.

    4. What are the key steps in analyzing a brand’s current Martech stack capabilities to identify gaps and opportunities for improvement?
    Developing a clear understanding of the Martech stack’s current state and ensuring it aligns with a brand’s strategic needs and future goals requires a structured and strategic approach.
    The process begins with defining what success looks like in terms of technology capabilities such as scalability, integration, automation, and data accessibility, and linking these capabilities directly to the brand’s broader business objectives.
    I start by doing an inventory of all tools currently in use, including their purpose, owner, and key functionalities, assessing if these tools are being used to their full potential or if there are features that remain unused, and reviewing how well tools integrate with one another and with our core systems, the data warehouse.
    Also, comparing the capabilities of each tool and results against industry standards and competitor practices and looking for missing functionalities such as personalization, omnichannel orchestration, or advanced analytics, and identifying overlapping tools that could be consolidated to save costs and streamline workflows.
    Finally, review the costs of the current tools against their impact on business outcomes and identify technologies that could reduce costs, increase efficiency, or deliver higher ROI through enhanced capabilities.

    Establish a regular review cycle for the Martech stack to ensure it evolves alongside the business and the technological landscape.

    5. How do you evaluate whether a company’s tech stack can support innovative customer-focused campaigns, and what red flags should marketers look out for?
    I recommend taking a structured approach and first ensure there is seamless integration across all tools to support a unified customer view and data sharing across the different channels.
    Determine if the stack can handle increasing data volumes, larger audiences, and additional channels as the campaigns grow, and check if it supports dynamic content, behavior-based triggers, and advanced segmentation and can process and act on data in real time through emerging technologies like AI/ML predictive analytics to enable marketers to launch responsive and timely campaigns.
    Most importantly, we need to ensure that the stack offers robust reporting tools that provide actionable insights, allowing teams to track performance and optimize campaigns.
    Some of the red flags are: data silos where customer data is fragmented across platforms and not easily accessible or integrated, inability to process or respond to customer behavior in real time, a reliance on manual intervention for tasks like segmentation, data extraction, campaign deployment, and poor scalability.

    If the stack struggles with growing data volumes or expanding to new channels, it won’t support the company’s evolving needs.

    6. What role do hyper-personalization and timely communication play in a successful customer engagement strategy? How do you ensure they’re built into the technology roadmap?
    Hyper-personalization and timely communication are essential components of a successful customer engagement strategy because they create meaningful, relevant, and impactful experiences that deepen the relationship with customers, enhance loyalty, and drive business outcomes.
    Hyper-personalization leverages data to deliver tailored content that resonates with each individual based on their preferences, behavior, or past interactions, and timely communication ensures these personalized interactions occur at the most relevant moments, which ultimately increases their impact.
    Customers are more likely to engage with messages that feel relevant and align with their needs, and real-time triggers such as cart abandonment or post-purchase upsells capitalize on moments when customers are most likely to convert.

    By embedding these capabilities into the roadmap through data integration, AI-driven insights, automation, and continuous optimization, we can deliver impactful, relevant, and timely experiences that foster deeper customer relationships and drive long-term success.

    7. What’s your approach to breaking down the customer engagement technology roadmap into manageable phases? How do you prioritize the initiatives?
    To create a manageable roadmap, we need to divide it into distinct phases, starting with building the foundation by addressing data cleanup, system integrations, and establishing metrics, which lays the groundwork for success.
    Next, we can focus on early wins and quick impact by launching behavior-based campaigns, automating workflows, and improving personalization to drive immediate value.
    Then we can move to optimization and expansion, incorporating predictive analytics, cross-channel orchestration, and refined attribution models to enhance our capabilities.
    Finally, prioritize innovation and scalability, leveraging AI/ML for hyper-personalization, scaling campaigns to new markets, and ensuring the system is equipped for future growth.
    By starting with foundational projects, delivering quick wins, and building towards scalable innovation, we can drive measurable outcomes while maintaining our agility to adapt to evolving needs.

    In terms of prioritizing initiatives effectively, I would focus on projects that deliver the greatest impact on business goals, on customer experience and ROI, while we consider feasibility, urgency, and resource availability.

    In the past, I’ve used frameworks like Impact Effort Matrix to identify the high-impact, low-effort initiatives and ensure that the most critical projects are addressed first.
    8. How do you ensure cross-functional alignment around this roadmap? What processes have worked best for you?
    Ensuring cross-functional alignment requires clear communication, collaborative planning, and shared accountability.
    We need to establish a shared understanding of the roadmap’s purpose and how it ties to the company’s overall goals by clearly articulating the “why” behind the roadmap and how each team can contribute to its success.
    To foster buy-in and ensure the roadmap reflects diverse perspectives and needs, we need to involve all stakeholders early on during the roadmap development and clearly outline each team’s role in executing the roadmap to ensure accountability across the different teams.

    To keep teams informed and aligned, we use meetings such as roadmap kickoff sessions and regular check-ins to share updates, address challenges collaboratively, and celebrate milestones together.

    9. If you were to outline a simple framework for marketers to follow when building a customer engagement technology roadmap, what would it look like?
    A simple framework for marketers to follow when building the roadmap can be summarized in five clear steps: Plan, Audit, Prioritize, Execute, and Refine.
    In one word: PAPER. Here’s how it breaks down.

    Plan: We lay the groundwork for the roadmap by defining the CRM strategy and aligning it with the business goals.
    Audit: We evaluate the current state of our CRM capabilities. We conduct a comprehensive assessment of our tools, our data, the processes, and team workflows to identify any potential gaps.
    Prioritize: initiatives based on impact, feasibility, and ROI potential.
    Execute: by implementing the roadmap in manageable phases.
    Refine: by continuously improving CRM performance and refining the roadmap.

    So the PAPER framework — Plan, Audit, Prioritize, Execute, and Refine — provides a structured, iterative approach allowing marketers to create a scalable and impactful customer engagement strategy.

    10. What are the most common challenges marketers face in creating or executing a customer engagement strategy, and how can they address these effectively?
    The most critical is when the customer data is siloed across different tools and platforms, making it very difficult to get a unified view of the customer. This limits the ability to deliver personalized and consistent experiences.

    The solution is to invest in tools that can centralize data from all touchpoints and ensure seamless integration between different platforms to create a single source of truth.

    Another challenge is the lack of clear metrics and ROI measurement and the inability to connect engagement efforts to tangible business outcomes, making it very hard to justify investment or optimize strategies.
    The solution for that is to define clear KPIs at the outset and use attribution models to link customer interactions to revenue and other key outcomes.
    Overcoming internal silos is another challenge where there is misalignment between teams, which can lead to inconsistent messaging and delayed execution.
    A solution to this is to foster cross-functional collaboration through shared goals, regular communication, and joint planning sessions.
    Besides these, other challenges marketers can face are delivering personalization at scale, keeping up with changing customer expectations, resource and budget constraints, resistance to change, and others.
    While creating and executing a customer engagement strategy can be challenging, these obstacles can be addressed through strategic planning, leveraging the right tools, fostering collaboration, and staying adaptable to customer needs and industry trends.

    By tackling these challenges proactively, marketers can deliver impactful customer-centric strategies that drive long-term success.

    11. What are the top takeaways or lessons that you’ve learned from building customer engagement technology roadmaps that others should keep in mind?
    I would say one of the most important takeaways is to ensure that the roadmap directly supports the company’s broader objectives.
    Whether the focus is on retention, customer lifetime value, or revenue growth, the roadmap must bridge the gap between high-level business goals and actionable initiatives.

    Another important lesson: The roadmap is only as effective as the data and systems it’s built upon.

    I’ve learned the importance of prioritizing foundational elements like data cleanup, integrations, and governance before tackling advanced initiatives like personalization or predictive analytics. Skipping this step can lead to inefficiencies or missed opportunities later on.
    A Customer Engagement Roadmap is a strategic tool that evolves alongside the business and its customers.

    So by aligning with business goals, building a solid foundation, focusing on impact, fostering collaboration, and remaining adaptable, you can create a roadmap that delivers measurable results and meaningful customer experiences.

     

     
    This interview Q&A was hosted with Mirela Cialai, Director of CRM & MarTech at Equinox, for Chapter 7 of The Customer Engagement Book: Adapt or Die.
    Download the PDF or request a physical copy of the book here.
    The post Mirela Cialai Q&A: Customer Engagement Book Interview appeared first on MoEngage.
    #mirela #cialai #qampampa #customer #engagement
    Mirela Cialai Q&A: Customer Engagement Book Interview
    Reading Time: 9 minutes In the ever-evolving landscape of customer engagement, staying ahead of the curve is not just advantageous, it’s essential. That’s why, for Chapter 7 of “The Customer Engagement Book: Adapt or Die,” we sat down with Mirela Cialai, a seasoned expert in CRM and Martech strategies at brands like Equinox. Mirela brings a wealth of knowledge in aligning technology roadmaps with business goals, shifting organizational focuses from acquisition to retention, and leveraging hyper-personalization to drive success. In this interview, Mirela dives deep into building robust customer engagement technology roadmaps. She unveils the “PAPER” framework—Plan, Audit, Prioritize, Execute, Refine—a simple yet effective strategy for marketers. You’ll gain insights into identifying gaps in your Martech stack, ensuring data accuracy, and prioritizing initiatives that deliver the greatest impact and ROI. Whether you’re navigating data silos, striving for cross-functional alignment, or aiming for seamless tech integration, Mirela’s expertise provides practical solutions and actionable takeaways.   Mirela Cialai Q&A Interview 1. How do you define the vision for a customer engagement platform roadmap in alignment with the broader business goals? Can you share any examples of successful visions from your experience? Defining the vision for the roadmap in alignment with the broader business goals involves creating a strategic framework that connects the team’s objectives with the organization’s overarching mission or primary objectives. This could be revenue growth, customer retention, market expansion, or operational efficiency. We then break down these goals into actionable areas where the team can contribute, such as improving engagement, increasing lifetime value, or driving acquisition. We articulate how the team will support business goals by defining the KPIs that link CRM outcomes — the team’s outcomes — to business goals. In a previous role, the CRM team I was leading faced significant challenges due to the lack of attribution capabilities and a reliance on surface-level metrics such as open rates and click-through rates to measure performance. This approach made it difficult to quantify the impact of our efforts on broader business objectives such as revenue growth. Recognizing this gap, I worked on defining a vision for the CRM team to address these shortcomings. Our vision was to drive measurable growth through enhanced data accuracy and improved attribution capabilities, which allowed us to deliver targeted, data-driven, and personalized customer experiences. To bring this vision to life, I developed a roadmap that focused on first improving data accuracy, building our attribution capabilities, and delivering personalization at scale. By aligning the vision with these strategic priorities, we were able to demonstrate the tangible impact of our efforts on the key business goals. 2. What steps did you take to ensure data accuracy? The data team was very diligent in ensuring that our data warehouse had accurate data. So taking that as the source of truth, we started cleaning the data in all the other platforms that were integrated with our data warehouse — our CRM platform, our attribution analytics platform, etc. That’s where we started, looking at all the different integrations and ensuring that the data flows were correct and that we had all the right flows in place. And also validating and cleaning our email database — that helped, having more accurate data. 3. How do you recommend shifting organizational focus from acquisition to retention within a customer engagement strategy? Shifting an organization’s focus from acquisition to retention requires a cultural and strategic shift, emphasizing the immense value that existing customers bring to long-term growth and profitability. I would start by quantifying the value of retention, showcasing how retaining customers is significantly more cost-effective than acquiring new ones. Research consistently shows that increasing retention rates by just 5% can boost profits by at least 25 to 95%. This data helps make a compelling case to stakeholders about the importance of prioritizing retention. Next, I would link retention to core business goals by demonstrating how enhancing customer lifetime value and loyalty can directly drive revenue growth. This involves shifting the organization’s focus to retention-specific metrics such as churn rate, repeat purchase rate, and customer LTV. These metrics provide actionable insights into customer behaviors and highlight the financial impact of retention initiatives, ensuring alignment with the broader company objectives. By framing retention as a driver of sustainable growth, the organization can see it not as a competing priority, but as a complementary strategy to acquisition, ultimately leading to a more balanced and effective customer engagement strategy. 4. What are the key steps in analyzing a brand’s current Martech stack capabilities to identify gaps and opportunities for improvement? Developing a clear understanding of the Martech stack’s current state and ensuring it aligns with a brand’s strategic needs and future goals requires a structured and strategic approach. The process begins with defining what success looks like in terms of technology capabilities such as scalability, integration, automation, and data accessibility, and linking these capabilities directly to the brand’s broader business objectives. I start by doing an inventory of all tools currently in use, including their purpose, owner, and key functionalities, assessing if these tools are being used to their full potential or if there are features that remain unused, and reviewing how well tools integrate with one another and with our core systems, the data warehouse. Also, comparing the capabilities of each tool and results against industry standards and competitor practices and looking for missing functionalities such as personalization, omnichannel orchestration, or advanced analytics, and identifying overlapping tools that could be consolidated to save costs and streamline workflows. Finally, review the costs of the current tools against their impact on business outcomes and identify technologies that could reduce costs, increase efficiency, or deliver higher ROI through enhanced capabilities. Establish a regular review cycle for the Martech stack to ensure it evolves alongside the business and the technological landscape. 5. How do you evaluate whether a company’s tech stack can support innovative customer-focused campaigns, and what red flags should marketers look out for? I recommend taking a structured approach and first ensure there is seamless integration across all tools to support a unified customer view and data sharing across the different channels. Determine if the stack can handle increasing data volumes, larger audiences, and additional channels as the campaigns grow, and check if it supports dynamic content, behavior-based triggers, and advanced segmentation and can process and act on data in real time through emerging technologies like AI/ML predictive analytics to enable marketers to launch responsive and timely campaigns. Most importantly, we need to ensure that the stack offers robust reporting tools that provide actionable insights, allowing teams to track performance and optimize campaigns. Some of the red flags are: data silos where customer data is fragmented across platforms and not easily accessible or integrated, inability to process or respond to customer behavior in real time, a reliance on manual intervention for tasks like segmentation, data extraction, campaign deployment, and poor scalability. If the stack struggles with growing data volumes or expanding to new channels, it won’t support the company’s evolving needs. 6. What role do hyper-personalization and timely communication play in a successful customer engagement strategy? How do you ensure they’re built into the technology roadmap? Hyper-personalization and timely communication are essential components of a successful customer engagement strategy because they create meaningful, relevant, and impactful experiences that deepen the relationship with customers, enhance loyalty, and drive business outcomes. Hyper-personalization leverages data to deliver tailored content that resonates with each individual based on their preferences, behavior, or past interactions, and timely communication ensures these personalized interactions occur at the most relevant moments, which ultimately increases their impact. Customers are more likely to engage with messages that feel relevant and align with their needs, and real-time triggers such as cart abandonment or post-purchase upsells capitalize on moments when customers are most likely to convert. By embedding these capabilities into the roadmap through data integration, AI-driven insights, automation, and continuous optimization, we can deliver impactful, relevant, and timely experiences that foster deeper customer relationships and drive long-term success. 7. What’s your approach to breaking down the customer engagement technology roadmap into manageable phases? How do you prioritize the initiatives? To create a manageable roadmap, we need to divide it into distinct phases, starting with building the foundation by addressing data cleanup, system integrations, and establishing metrics, which lays the groundwork for success. Next, we can focus on early wins and quick impact by launching behavior-based campaigns, automating workflows, and improving personalization to drive immediate value. Then we can move to optimization and expansion, incorporating predictive analytics, cross-channel orchestration, and refined attribution models to enhance our capabilities. Finally, prioritize innovation and scalability, leveraging AI/ML for hyper-personalization, scaling campaigns to new markets, and ensuring the system is equipped for future growth. By starting with foundational projects, delivering quick wins, and building towards scalable innovation, we can drive measurable outcomes while maintaining our agility to adapt to evolving needs. In terms of prioritizing initiatives effectively, I would focus on projects that deliver the greatest impact on business goals, on customer experience and ROI, while we consider feasibility, urgency, and resource availability. In the past, I’ve used frameworks like Impact Effort Matrix to identify the high-impact, low-effort initiatives and ensure that the most critical projects are addressed first. 8. How do you ensure cross-functional alignment around this roadmap? What processes have worked best for you? Ensuring cross-functional alignment requires clear communication, collaborative planning, and shared accountability. We need to establish a shared understanding of the roadmap’s purpose and how it ties to the company’s overall goals by clearly articulating the “why” behind the roadmap and how each team can contribute to its success. To foster buy-in and ensure the roadmap reflects diverse perspectives and needs, we need to involve all stakeholders early on during the roadmap development and clearly outline each team’s role in executing the roadmap to ensure accountability across the different teams. To keep teams informed and aligned, we use meetings such as roadmap kickoff sessions and regular check-ins to share updates, address challenges collaboratively, and celebrate milestones together. 9. If you were to outline a simple framework for marketers to follow when building a customer engagement technology roadmap, what would it look like? A simple framework for marketers to follow when building the roadmap can be summarized in five clear steps: Plan, Audit, Prioritize, Execute, and Refine. In one word: PAPER. Here’s how it breaks down. Plan: We lay the groundwork for the roadmap by defining the CRM strategy and aligning it with the business goals. Audit: We evaluate the current state of our CRM capabilities. We conduct a comprehensive assessment of our tools, our data, the processes, and team workflows to identify any potential gaps. Prioritize: initiatives based on impact, feasibility, and ROI potential. Execute: by implementing the roadmap in manageable phases. Refine: by continuously improving CRM performance and refining the roadmap. So the PAPER framework — Plan, Audit, Prioritize, Execute, and Refine — provides a structured, iterative approach allowing marketers to create a scalable and impactful customer engagement strategy. 10. What are the most common challenges marketers face in creating or executing a customer engagement strategy, and how can they address these effectively? The most critical is when the customer data is siloed across different tools and platforms, making it very difficult to get a unified view of the customer. This limits the ability to deliver personalized and consistent experiences. The solution is to invest in tools that can centralize data from all touchpoints and ensure seamless integration between different platforms to create a single source of truth. Another challenge is the lack of clear metrics and ROI measurement and the inability to connect engagement efforts to tangible business outcomes, making it very hard to justify investment or optimize strategies. The solution for that is to define clear KPIs at the outset and use attribution models to link customer interactions to revenue and other key outcomes. Overcoming internal silos is another challenge where there is misalignment between teams, which can lead to inconsistent messaging and delayed execution. A solution to this is to foster cross-functional collaboration through shared goals, regular communication, and joint planning sessions. Besides these, other challenges marketers can face are delivering personalization at scale, keeping up with changing customer expectations, resource and budget constraints, resistance to change, and others. While creating and executing a customer engagement strategy can be challenging, these obstacles can be addressed through strategic planning, leveraging the right tools, fostering collaboration, and staying adaptable to customer needs and industry trends. By tackling these challenges proactively, marketers can deliver impactful customer-centric strategies that drive long-term success. 11. What are the top takeaways or lessons that you’ve learned from building customer engagement technology roadmaps that others should keep in mind? I would say one of the most important takeaways is to ensure that the roadmap directly supports the company’s broader objectives. Whether the focus is on retention, customer lifetime value, or revenue growth, the roadmap must bridge the gap between high-level business goals and actionable initiatives. Another important lesson: The roadmap is only as effective as the data and systems it’s built upon. I’ve learned the importance of prioritizing foundational elements like data cleanup, integrations, and governance before tackling advanced initiatives like personalization or predictive analytics. Skipping this step can lead to inefficiencies or missed opportunities later on. A Customer Engagement Roadmap is a strategic tool that evolves alongside the business and its customers. So by aligning with business goals, building a solid foundation, focusing on impact, fostering collaboration, and remaining adaptable, you can create a roadmap that delivers measurable results and meaningful customer experiences.     This interview Q&A was hosted with Mirela Cialai, Director of CRM & MarTech at Equinox, for Chapter 7 of The Customer Engagement Book: Adapt or Die. Download the PDF or request a physical copy of the book here. The post Mirela Cialai Q&A: Customer Engagement Book Interview appeared first on MoEngage. #mirela #cialai #qampampa #customer #engagement
    WWW.MOENGAGE.COM
    Mirela Cialai Q&A: Customer Engagement Book Interview
    Reading Time: 9 minutes In the ever-evolving landscape of customer engagement, staying ahead of the curve is not just advantageous, it’s essential. That’s why, for Chapter 7 of “The Customer Engagement Book: Adapt or Die,” we sat down with Mirela Cialai, a seasoned expert in CRM and Martech strategies at brands like Equinox. Mirela brings a wealth of knowledge in aligning technology roadmaps with business goals, shifting organizational focuses from acquisition to retention, and leveraging hyper-personalization to drive success. In this interview, Mirela dives deep into building robust customer engagement technology roadmaps. She unveils the “PAPER” framework—Plan, Audit, Prioritize, Execute, Refine—a simple yet effective strategy for marketers. You’ll gain insights into identifying gaps in your Martech stack, ensuring data accuracy, and prioritizing initiatives that deliver the greatest impact and ROI. Whether you’re navigating data silos, striving for cross-functional alignment, or aiming for seamless tech integration, Mirela’s expertise provides practical solutions and actionable takeaways.   Mirela Cialai Q&A Interview 1. How do you define the vision for a customer engagement platform roadmap in alignment with the broader business goals? Can you share any examples of successful visions from your experience? Defining the vision for the roadmap in alignment with the broader business goals involves creating a strategic framework that connects the team’s objectives with the organization’s overarching mission or primary objectives. This could be revenue growth, customer retention, market expansion, or operational efficiency. We then break down these goals into actionable areas where the team can contribute, such as improving engagement, increasing lifetime value, or driving acquisition. We articulate how the team will support business goals by defining the KPIs that link CRM outcomes — the team’s outcomes — to business goals. In a previous role, the CRM team I was leading faced significant challenges due to the lack of attribution capabilities and a reliance on surface-level metrics such as open rates and click-through rates to measure performance. This approach made it difficult to quantify the impact of our efforts on broader business objectives such as revenue growth. Recognizing this gap, I worked on defining a vision for the CRM team to address these shortcomings. Our vision was to drive measurable growth through enhanced data accuracy and improved attribution capabilities, which allowed us to deliver targeted, data-driven, and personalized customer experiences. To bring this vision to life, I developed a roadmap that focused on first improving data accuracy, building our attribution capabilities, and delivering personalization at scale. By aligning the vision with these strategic priorities, we were able to demonstrate the tangible impact of our efforts on the key business goals. 2. What steps did you take to ensure data accuracy? The data team was very diligent in ensuring that our data warehouse had accurate data. So taking that as the source of truth, we started cleaning the data in all the other platforms that were integrated with our data warehouse — our CRM platform, our attribution analytics platform, etc. That’s where we started, looking at all the different integrations and ensuring that the data flows were correct and that we had all the right flows in place. And also validating and cleaning our email database — that helped, having more accurate data. 3. How do you recommend shifting organizational focus from acquisition to retention within a customer engagement strategy? Shifting an organization’s focus from acquisition to retention requires a cultural and strategic shift, emphasizing the immense value that existing customers bring to long-term growth and profitability. I would start by quantifying the value of retention, showcasing how retaining customers is significantly more cost-effective than acquiring new ones. Research consistently shows that increasing retention rates by just 5% can boost profits by at least 25 to 95%. This data helps make a compelling case to stakeholders about the importance of prioritizing retention. Next, I would link retention to core business goals by demonstrating how enhancing customer lifetime value and loyalty can directly drive revenue growth. This involves shifting the organization’s focus to retention-specific metrics such as churn rate, repeat purchase rate, and customer LTV. These metrics provide actionable insights into customer behaviors and highlight the financial impact of retention initiatives, ensuring alignment with the broader company objectives. By framing retention as a driver of sustainable growth, the organization can see it not as a competing priority, but as a complementary strategy to acquisition, ultimately leading to a more balanced and effective customer engagement strategy. 4. What are the key steps in analyzing a brand’s current Martech stack capabilities to identify gaps and opportunities for improvement? Developing a clear understanding of the Martech stack’s current state and ensuring it aligns with a brand’s strategic needs and future goals requires a structured and strategic approach. The process begins with defining what success looks like in terms of technology capabilities such as scalability, integration, automation, and data accessibility, and linking these capabilities directly to the brand’s broader business objectives. I start by doing an inventory of all tools currently in use, including their purpose, owner, and key functionalities, assessing if these tools are being used to their full potential or if there are features that remain unused, and reviewing how well tools integrate with one another and with our core systems, the data warehouse. Also, comparing the capabilities of each tool and results against industry standards and competitor practices and looking for missing functionalities such as personalization, omnichannel orchestration, or advanced analytics, and identifying overlapping tools that could be consolidated to save costs and streamline workflows. Finally, review the costs of the current tools against their impact on business outcomes and identify technologies that could reduce costs, increase efficiency, or deliver higher ROI through enhanced capabilities. Establish a regular review cycle for the Martech stack to ensure it evolves alongside the business and the technological landscape. 5. How do you evaluate whether a company’s tech stack can support innovative customer-focused campaigns, and what red flags should marketers look out for? I recommend taking a structured approach and first ensure there is seamless integration across all tools to support a unified customer view and data sharing across the different channels. Determine if the stack can handle increasing data volumes, larger audiences, and additional channels as the campaigns grow, and check if it supports dynamic content, behavior-based triggers, and advanced segmentation and can process and act on data in real time through emerging technologies like AI/ML predictive analytics to enable marketers to launch responsive and timely campaigns. Most importantly, we need to ensure that the stack offers robust reporting tools that provide actionable insights, allowing teams to track performance and optimize campaigns. Some of the red flags are: data silos where customer data is fragmented across platforms and not easily accessible or integrated, inability to process or respond to customer behavior in real time, a reliance on manual intervention for tasks like segmentation, data extraction, campaign deployment, and poor scalability. If the stack struggles with growing data volumes or expanding to new channels, it won’t support the company’s evolving needs. 6. What role do hyper-personalization and timely communication play in a successful customer engagement strategy? How do you ensure they’re built into the technology roadmap? Hyper-personalization and timely communication are essential components of a successful customer engagement strategy because they create meaningful, relevant, and impactful experiences that deepen the relationship with customers, enhance loyalty, and drive business outcomes. Hyper-personalization leverages data to deliver tailored content that resonates with each individual based on their preferences, behavior, or past interactions, and timely communication ensures these personalized interactions occur at the most relevant moments, which ultimately increases their impact. Customers are more likely to engage with messages that feel relevant and align with their needs, and real-time triggers such as cart abandonment or post-purchase upsells capitalize on moments when customers are most likely to convert. By embedding these capabilities into the roadmap through data integration, AI-driven insights, automation, and continuous optimization, we can deliver impactful, relevant, and timely experiences that foster deeper customer relationships and drive long-term success. 7. What’s your approach to breaking down the customer engagement technology roadmap into manageable phases? How do you prioritize the initiatives? To create a manageable roadmap, we need to divide it into distinct phases, starting with building the foundation by addressing data cleanup, system integrations, and establishing metrics, which lays the groundwork for success. Next, we can focus on early wins and quick impact by launching behavior-based campaigns, automating workflows, and improving personalization to drive immediate value. Then we can move to optimization and expansion, incorporating predictive analytics, cross-channel orchestration, and refined attribution models to enhance our capabilities. Finally, prioritize innovation and scalability, leveraging AI/ML for hyper-personalization, scaling campaigns to new markets, and ensuring the system is equipped for future growth. By starting with foundational projects, delivering quick wins, and building towards scalable innovation, we can drive measurable outcomes while maintaining our agility to adapt to evolving needs. In terms of prioritizing initiatives effectively, I would focus on projects that deliver the greatest impact on business goals, on customer experience and ROI, while we consider feasibility, urgency, and resource availability. In the past, I’ve used frameworks like Impact Effort Matrix to identify the high-impact, low-effort initiatives and ensure that the most critical projects are addressed first. 8. How do you ensure cross-functional alignment around this roadmap? What processes have worked best for you? Ensuring cross-functional alignment requires clear communication, collaborative planning, and shared accountability. We need to establish a shared understanding of the roadmap’s purpose and how it ties to the company’s overall goals by clearly articulating the “why” behind the roadmap and how each team can contribute to its success. To foster buy-in and ensure the roadmap reflects diverse perspectives and needs, we need to involve all stakeholders early on during the roadmap development and clearly outline each team’s role in executing the roadmap to ensure accountability across the different teams. To keep teams informed and aligned, we use meetings such as roadmap kickoff sessions and regular check-ins to share updates, address challenges collaboratively, and celebrate milestones together. 9. If you were to outline a simple framework for marketers to follow when building a customer engagement technology roadmap, what would it look like? A simple framework for marketers to follow when building the roadmap can be summarized in five clear steps: Plan, Audit, Prioritize, Execute, and Refine. In one word: PAPER. Here’s how it breaks down. Plan: We lay the groundwork for the roadmap by defining the CRM strategy and aligning it with the business goals. Audit: We evaluate the current state of our CRM capabilities. We conduct a comprehensive assessment of our tools, our data, the processes, and team workflows to identify any potential gaps. Prioritize: initiatives based on impact, feasibility, and ROI potential. Execute: by implementing the roadmap in manageable phases. Refine: by continuously improving CRM performance and refining the roadmap. So the PAPER framework — Plan, Audit, Prioritize, Execute, and Refine — provides a structured, iterative approach allowing marketers to create a scalable and impactful customer engagement strategy. 10. What are the most common challenges marketers face in creating or executing a customer engagement strategy, and how can they address these effectively? The most critical is when the customer data is siloed across different tools and platforms, making it very difficult to get a unified view of the customer. This limits the ability to deliver personalized and consistent experiences. The solution is to invest in tools that can centralize data from all touchpoints and ensure seamless integration between different platforms to create a single source of truth. Another challenge is the lack of clear metrics and ROI measurement and the inability to connect engagement efforts to tangible business outcomes, making it very hard to justify investment or optimize strategies. The solution for that is to define clear KPIs at the outset and use attribution models to link customer interactions to revenue and other key outcomes. Overcoming internal silos is another challenge where there is misalignment between teams, which can lead to inconsistent messaging and delayed execution. A solution to this is to foster cross-functional collaboration through shared goals, regular communication, and joint planning sessions. Besides these, other challenges marketers can face are delivering personalization at scale, keeping up with changing customer expectations, resource and budget constraints, resistance to change, and others. While creating and executing a customer engagement strategy can be challenging, these obstacles can be addressed through strategic planning, leveraging the right tools, fostering collaboration, and staying adaptable to customer needs and industry trends. By tackling these challenges proactively, marketers can deliver impactful customer-centric strategies that drive long-term success. 11. What are the top takeaways or lessons that you’ve learned from building customer engagement technology roadmaps that others should keep in mind? I would say one of the most important takeaways is to ensure that the roadmap directly supports the company’s broader objectives. Whether the focus is on retention, customer lifetime value, or revenue growth, the roadmap must bridge the gap between high-level business goals and actionable initiatives. Another important lesson: The roadmap is only as effective as the data and systems it’s built upon. I’ve learned the importance of prioritizing foundational elements like data cleanup, integrations, and governance before tackling advanced initiatives like personalization or predictive analytics. Skipping this step can lead to inefficiencies or missed opportunities later on. A Customer Engagement Roadmap is a strategic tool that evolves alongside the business and its customers. So by aligning with business goals, building a solid foundation, focusing on impact, fostering collaboration, and remaining adaptable, you can create a roadmap that delivers measurable results and meaningful customer experiences.     This interview Q&A was hosted with Mirela Cialai, Director of CRM & MarTech at Equinox, for Chapter 7 of The Customer Engagement Book: Adapt or Die. Download the PDF or request a physical copy of the book here. The post Mirela Cialai Q&A: Customer Engagement Book Interview appeared first on MoEngage.
    0 Commentarios 0 Acciones
  • New Zealand’s Email Security Requirements for Government Organizations: What You Need to Know

    The Secure Government EmailCommon Implementation Framework
    New Zealand’s government is introducing a comprehensive email security framework designed to protect official communications from phishing and domain spoofing. This new framework, which will be mandatory for all government agencies by October 2025, establishes clear technical standards to enhance email security and retire the outdated SEEMail service. 
    Key Takeaways

    All NZ government agencies must comply with new email security requirements by October 2025.
    The new framework strengthens trust and security in government communications by preventing spoofing and phishing.
    The framework mandates TLS 1.2+, SPF, DKIM, DMARC with p=reject, MTA-STS, and DLP controls.
    EasyDMARC simplifies compliance with our guided setup, monitoring, and automated reporting.

    Start a Free Trial

    What is the Secure Government Email Common Implementation Framework?
    The Secure Government EmailCommon Implementation Framework is a new government-led initiative in New Zealand designed to standardize email security across all government agencies. Its main goal is to secure external email communication, reduce domain spoofing in phishing attacks, and replace the legacy SEEMail service.
    Why is New Zealand Implementing New Government Email Security Standards?
    The framework was developed by New Zealand’s Department of Internal Affairsas part of its role in managing ICT Common Capabilities. It leverages modern email security controls via the Domain Name Systemto enable the retirement of the legacy SEEMail service and provide:

    Encryption for transmission security
    Digital signing for message integrity
    Basic non-repudiationDomain spoofing protection

    These improvements apply to all emails, not just those routed through SEEMail, offering broader protection across agency communications.
    What Email Security Technologies Are Required by the New NZ SGE Framework?
    The SGE Framework outlines the following key technologies that agencies must implement:

    TLS 1.2 or higher with implicit TLS enforced
    TLS-RPTSPFDKIMDMARCwith reporting
    MTA-STSData Loss Prevention controls

    These technologies work together to ensure encrypted email transmission, validate sender identity, prevent unauthorized use of domains, and reduce the risk of sensitive data leaks.

    Get in touch

    When Do NZ Government Agencies Need to Comply with this Framework?
    All New Zealand government agencies are expected to fully implement the Secure Government EmailCommon Implementation Framework by October 2025. Agencies should begin their planning and deployment now to ensure full compliance by the deadline.
    The All of Government Secure Email Common Implementation Framework v1.0
    What are the Mandated Requirements for Domains?
    Below are the exact requirements for all email-enabled domains under the new framework.
    ControlExact RequirementTLSMinimum TLS 1.2. TLS 1.1, 1.0, SSL, or clear-text not permitted.TLS-RPTAll email-sending domains must have TLS reporting enabled.SPFMust exist and end with -all.DKIMAll outbound email from every sending service must be DKIM-signed at the final hop.DMARCPolicy of p=reject on all email-enabled domains. adkim=s is recommended when not bulk-sending.MTA-STSEnabled and set to enforce.Implicit TLSMust be configured and enforced for every connection.Data Loss PreventionEnforce in line with the New Zealand Information Security Manualand Protective Security Requirements.
    Compliance Monitoring and Reporting
    The All of Government Service Deliveryteam will be monitoring compliance with the framework. Monitoring will initially cover SPF, DMARC, and MTA-STS settings and will be expanded to include DKIM. Changes to these settings will be monitored, enabling reporting on email security compliance across all government agencies. Ongoing monitoring will highlight changes to domains, ensure new domains are set up with security in place, and monitor the implementation of future email security technologies. 
    Should compliance changes occur, such as an agency’s SPF record being changed from -all to ~all, this will be captured so that the AoGSD Security Team can investigate. They will then communicate directly with the agency to determine if an issue exists or if an error has occurred, reviewing each case individually.
    Deployment Checklist for NZ Government Compliance

    Enforce TLS 1.2 minimum, implicit TLS, MTA-STS & TLS-RPT
    SPF with -all
    DKIM on all outbound email
    DMARC p=reject 
    adkim=s where suitable
    For non-email/parked domains: SPF -all, empty DKIM, DMARC reject strict
    Compliance dashboard
    Inbound DMARC evaluation enforced
    DLP aligned with NZISM

    Start a Free Trial

    How EasyDMARC Can Help Government Agencies Comply
    EasyDMARC provides a comprehensive email security solution that simplifies the deployment and ongoing management of DNS-based email security protocols like SPF, DKIM, and DMARC with reporting. Our platform offers automated checks, real-time monitoring, and a guided setup to help government organizations quickly reach compliance.
    1. TLS-RPT / MTA-STS audit
    EasyDMARC enables you to enable the Managed MTA-STS and TLS-RPT option with a single click. We provide the required DNS records and continuously monitor them for issues, delivering reports on TLS negotiation problems. This helps agencies ensure secure email transmission and quickly detect delivery or encryption failures.

    Note: In this screenshot, you can see how to deploy MTA-STS and TLS Reporting by adding just three CNAME records provided by EasyDMARC. It’s recommended to start in “testing” mode, evaluate the TLS-RPT reports, and then gradually switch your MTA-STS policy to “enforce”. The process is simple and takes just a few clicks.

    As shown above, EasyDMARC parses incoming TLS reports into a centralized dashboard, giving you clear visibility into delivery and encryption issues across all sending sources.
    2. SPF with “-all”In the EasyDARC platform, you can run the SPF Record Generator to create a compliant record. Publish your v=spf1 record with “-all” to enforce a hard fail for unauthorized senders and prevent spoofed emails from passing SPF checks. This strengthens your domain’s protection against impersonation.

    Note: It is highly recommended to start adjusting your SPF record only after you begin receiving DMARC reports and identifying your legitimate email sources. As we’ll explain in more detail below, both SPF and DKIM should be adjusted after you gain visibility through reports.
    Making changes without proper visibility can lead to false positives, misconfigurations, and potential loss of legitimate emails. That’s why the first step should always be setting DMARC to p=none, receiving reports, analyzing them, and then gradually fixing any SPF or DKIM issues.
    3. DKIM on all outbound email
    DKIM must be configured for all email sources sending emails on behalf of your domain. This is critical, as DKIM plays a bigger role than SPF when it comes to building domain reputation, surviving auto-forwarding, mailing lists, and other edge cases.
    As mentioned above, DMARC reports provide visibility into your email sources, allowing you to implement DKIM accordingly. If you’re using third-party services like Google Workspace, Microsoft 365, or Mimecast, you’ll need to retrieve the public DKIM key from your provider’s admin interface.
    EasyDMARC maintains a backend directory of over 1,400 email sources. We also give you detailed guidance on how to configure SPF and DKIM correctly for major ESPs. 
    Note: At the end of this article, you’ll find configuration links for well-known ESPs like Google Workspace, Microsoft 365, Zoho Mail, Amazon SES, and SendGrid – helping you avoid common misconfigurations and get aligned with SGE requirements.
    If you’re using a dedicated MTA, DKIM must be implemented manually. EasyDMARC’s DKIM Record Generator lets you generate both public and private keys for your server. The private key is stored on your MTA, while the public key must be published in your DNS.

    4. DMARC p=reject rollout
    As mentioned in previous points, DMARC reporting is the first and most important step on your DMARC enforcement journey. Always start with a p=none policy and configure RUA reports to be sent to EasyDMARC. Use the report insights to identify and fix SPF and DKIM alignment issues, then gradually move to p=quarantine and finally p=reject once all legitimate email sources have been authenticated. 
    This phased approach ensures full protection against domain spoofing without risking legitimate email delivery.

    5. adkim Strict Alignment Check
    This strict alignment check is not always applicable, especially if you’re using third-party bulk ESPs, such as Sendgrid, that require you to set DKIM on a subdomain level. You can set adkim=s in your DMARC TXT record, or simply enable strict mode in EasyDMARC’s Managed DMARC settings. This ensures that only emails with a DKIM signature that exactly match your domain pass alignment, adding an extra layer of protection against domain spoofing. But only do this if you are NOT a bulk sender.

    6. Securing Non-Email Enabled Domains
    The purpose of deploying email security to non-email-enabled domains, or parked domains, is to prevent messages being spoofed from that domain. This requirement remains even if the root-level domain has SP=reject set within its DMARC record.
    Under this new framework, you must bulk import and mark parked domains as “Parked.” Crucially, this requires adjusting SPF settings to an empty record, setting DMARC to p=reject, and ensuring an empty DKIM record is in place: • SPF record: “v=spf1 -all”.
    • Wildcard DKIM record with empty public key.• DMARC record: “v=DMARC1;p=reject;adkim=s;aspf=s;rua=mailto:…”.
    EasyDMARC allows you to add and label parked domains for free. This is important because it helps you monitor any activity from these domains and ensure they remain protected with a strict DMARC policy of p=reject.
    7. Compliance Dashboard
    Use EasyDMARC’s Domain Scanner to assess the security posture of each domain with a clear compliance score and risk level. The dashboard highlights configuration gaps and guides remediation steps, helping government agencies stay on track toward full compliance with the SGE Framework.

    8. Inbound DMARC Evaluation Enforced
    You don’t need to apply any changes if you’re using Google Workspace, Microsoft 365, or other major mailbox providers. Most of them already enforce DMARC evaluation on incoming emails.
    However, some legacy Microsoft 365 setups may still quarantine emails that fail DMARC checks, even when the sending domain has a p=reject policy, instead of rejecting them. This behavior can be adjusted directly from your Microsoft Defender portal. about this in our step-by-step guide on how to set up SPF, DKIM, and DMARC from Microsoft Defender.
    If you’re using a third-party mail provider that doesn’t enforce having a DMARC policy for incoming emails, which is rare, you’ll need to contact their support to request a configuration change.
    9. Data Loss Prevention Aligned with NZISM
    The New Zealand Information Security Manualis the New Zealand Government’s manual on information assurance and information systems security. It includes guidance on data loss prevention, which must be followed to be aligned with the SEG.
    Need Help Setting up SPF and DKIM for your Email Provider?
    Setting up SPF and DKIM for different ESPs often requires specific configurations. Some providers require you to publish SPF and DKIM on a subdomain, while others only require DKIM, or have different formatting rules. We’ve simplified all these steps to help you avoid misconfigurations that could delay your DMARC enforcement, or worse, block legitimate emails from reaching your recipients.
    Below you’ll find comprehensive setup guides for Google Workspace, Microsoft 365, Zoho Mail, Amazon SES, and SendGrid. You can also explore our full blog section that covers setup instructions for many other well-known ESPs.
    Remember, all this information is reflected in your DMARC aggregate reports. These reports give you live visibility into your outgoing email ecosystem, helping you analyze and fix any issues specific to a given provider.
    Here are our step-by-step guides for the most common platforms:

    Google Workspace

    Microsoft 365

    These guides will help ensure your DNS records are configured correctly as part of the Secure Government EmailFramework rollout.
    Meet New Government Email Security Standards With EasyDMARC
    New Zealand’s SEG Framework sets a clear path for government agencies to enhance their email security by October 2025. With EasyDMARC, you can meet these technical requirements efficiently and with confidence. From protocol setup to continuous monitoring and compliance tracking, EasyDMARC streamlines the entire process, ensuring strong protection against spoofing, phishing, and data loss while simplifying your transition from SEEMail.
    #new #zealands #email #security #requirements
    New Zealand’s Email Security Requirements for Government Organizations: What You Need to Know
    The Secure Government EmailCommon Implementation Framework New Zealand’s government is introducing a comprehensive email security framework designed to protect official communications from phishing and domain spoofing. This new framework, which will be mandatory for all government agencies by October 2025, establishes clear technical standards to enhance email security and retire the outdated SEEMail service.  Key Takeaways All NZ government agencies must comply with new email security requirements by October 2025. The new framework strengthens trust and security in government communications by preventing spoofing and phishing. The framework mandates TLS 1.2+, SPF, DKIM, DMARC with p=reject, MTA-STS, and DLP controls. EasyDMARC simplifies compliance with our guided setup, monitoring, and automated reporting. Start a Free Trial What is the Secure Government Email Common Implementation Framework? The Secure Government EmailCommon Implementation Framework is a new government-led initiative in New Zealand designed to standardize email security across all government agencies. Its main goal is to secure external email communication, reduce domain spoofing in phishing attacks, and replace the legacy SEEMail service. Why is New Zealand Implementing New Government Email Security Standards? The framework was developed by New Zealand’s Department of Internal Affairsas part of its role in managing ICT Common Capabilities. It leverages modern email security controls via the Domain Name Systemto enable the retirement of the legacy SEEMail service and provide: Encryption for transmission security Digital signing for message integrity Basic non-repudiationDomain spoofing protection These improvements apply to all emails, not just those routed through SEEMail, offering broader protection across agency communications. What Email Security Technologies Are Required by the New NZ SGE Framework? The SGE Framework outlines the following key technologies that agencies must implement: TLS 1.2 or higher with implicit TLS enforced TLS-RPTSPFDKIMDMARCwith reporting MTA-STSData Loss Prevention controls These technologies work together to ensure encrypted email transmission, validate sender identity, prevent unauthorized use of domains, and reduce the risk of sensitive data leaks. Get in touch When Do NZ Government Agencies Need to Comply with this Framework? All New Zealand government agencies are expected to fully implement the Secure Government EmailCommon Implementation Framework by October 2025. Agencies should begin their planning and deployment now to ensure full compliance by the deadline. The All of Government Secure Email Common Implementation Framework v1.0 What are the Mandated Requirements for Domains? Below are the exact requirements for all email-enabled domains under the new framework. ControlExact RequirementTLSMinimum TLS 1.2. TLS 1.1, 1.0, SSL, or clear-text not permitted.TLS-RPTAll email-sending domains must have TLS reporting enabled.SPFMust exist and end with -all.DKIMAll outbound email from every sending service must be DKIM-signed at the final hop.DMARCPolicy of p=reject on all email-enabled domains. adkim=s is recommended when not bulk-sending.MTA-STSEnabled and set to enforce.Implicit TLSMust be configured and enforced for every connection.Data Loss PreventionEnforce in line with the New Zealand Information Security Manualand Protective Security Requirements. Compliance Monitoring and Reporting The All of Government Service Deliveryteam will be monitoring compliance with the framework. Monitoring will initially cover SPF, DMARC, and MTA-STS settings and will be expanded to include DKIM. Changes to these settings will be monitored, enabling reporting on email security compliance across all government agencies. Ongoing monitoring will highlight changes to domains, ensure new domains are set up with security in place, and monitor the implementation of future email security technologies.  Should compliance changes occur, such as an agency’s SPF record being changed from -all to ~all, this will be captured so that the AoGSD Security Team can investigate. They will then communicate directly with the agency to determine if an issue exists or if an error has occurred, reviewing each case individually. Deployment Checklist for NZ Government Compliance Enforce TLS 1.2 minimum, implicit TLS, MTA-STS & TLS-RPT SPF with -all DKIM on all outbound email DMARC p=reject  adkim=s where suitable For non-email/parked domains: SPF -all, empty DKIM, DMARC reject strict Compliance dashboard Inbound DMARC evaluation enforced DLP aligned with NZISM Start a Free Trial How EasyDMARC Can Help Government Agencies Comply EasyDMARC provides a comprehensive email security solution that simplifies the deployment and ongoing management of DNS-based email security protocols like SPF, DKIM, and DMARC with reporting. Our platform offers automated checks, real-time monitoring, and a guided setup to help government organizations quickly reach compliance. 1. TLS-RPT / MTA-STS audit EasyDMARC enables you to enable the Managed MTA-STS and TLS-RPT option with a single click. We provide the required DNS records and continuously monitor them for issues, delivering reports on TLS negotiation problems. This helps agencies ensure secure email transmission and quickly detect delivery or encryption failures. Note: In this screenshot, you can see how to deploy MTA-STS and TLS Reporting by adding just three CNAME records provided by EasyDMARC. It’s recommended to start in “testing” mode, evaluate the TLS-RPT reports, and then gradually switch your MTA-STS policy to “enforce”. The process is simple and takes just a few clicks. As shown above, EasyDMARC parses incoming TLS reports into a centralized dashboard, giving you clear visibility into delivery and encryption issues across all sending sources. 2. SPF with “-all”In the EasyDARC platform, you can run the SPF Record Generator to create a compliant record. Publish your v=spf1 record with “-all” to enforce a hard fail for unauthorized senders and prevent spoofed emails from passing SPF checks. This strengthens your domain’s protection against impersonation. Note: It is highly recommended to start adjusting your SPF record only after you begin receiving DMARC reports and identifying your legitimate email sources. As we’ll explain in more detail below, both SPF and DKIM should be adjusted after you gain visibility through reports. Making changes without proper visibility can lead to false positives, misconfigurations, and potential loss of legitimate emails. That’s why the first step should always be setting DMARC to p=none, receiving reports, analyzing them, and then gradually fixing any SPF or DKIM issues. 3. DKIM on all outbound email DKIM must be configured for all email sources sending emails on behalf of your domain. This is critical, as DKIM plays a bigger role than SPF when it comes to building domain reputation, surviving auto-forwarding, mailing lists, and other edge cases. As mentioned above, DMARC reports provide visibility into your email sources, allowing you to implement DKIM accordingly. If you’re using third-party services like Google Workspace, Microsoft 365, or Mimecast, you’ll need to retrieve the public DKIM key from your provider’s admin interface. EasyDMARC maintains a backend directory of over 1,400 email sources. We also give you detailed guidance on how to configure SPF and DKIM correctly for major ESPs.  Note: At the end of this article, you’ll find configuration links for well-known ESPs like Google Workspace, Microsoft 365, Zoho Mail, Amazon SES, and SendGrid – helping you avoid common misconfigurations and get aligned with SGE requirements. If you’re using a dedicated MTA, DKIM must be implemented manually. EasyDMARC’s DKIM Record Generator lets you generate both public and private keys for your server. The private key is stored on your MTA, while the public key must be published in your DNS. 4. DMARC p=reject rollout As mentioned in previous points, DMARC reporting is the first and most important step on your DMARC enforcement journey. Always start with a p=none policy and configure RUA reports to be sent to EasyDMARC. Use the report insights to identify and fix SPF and DKIM alignment issues, then gradually move to p=quarantine and finally p=reject once all legitimate email sources have been authenticated.  This phased approach ensures full protection against domain spoofing without risking legitimate email delivery. 5. adkim Strict Alignment Check This strict alignment check is not always applicable, especially if you’re using third-party bulk ESPs, such as Sendgrid, that require you to set DKIM on a subdomain level. You can set adkim=s in your DMARC TXT record, or simply enable strict mode in EasyDMARC’s Managed DMARC settings. This ensures that only emails with a DKIM signature that exactly match your domain pass alignment, adding an extra layer of protection against domain spoofing. But only do this if you are NOT a bulk sender. 6. Securing Non-Email Enabled Domains The purpose of deploying email security to non-email-enabled domains, or parked domains, is to prevent messages being spoofed from that domain. This requirement remains even if the root-level domain has SP=reject set within its DMARC record. Under this new framework, you must bulk import and mark parked domains as “Parked.” Crucially, this requires adjusting SPF settings to an empty record, setting DMARC to p=reject, and ensuring an empty DKIM record is in place: • SPF record: “v=spf1 -all”. • Wildcard DKIM record with empty public key.• DMARC record: “v=DMARC1;p=reject;adkim=s;aspf=s;rua=mailto:…”. EasyDMARC allows you to add and label parked domains for free. This is important because it helps you monitor any activity from these domains and ensure they remain protected with a strict DMARC policy of p=reject. 7. Compliance Dashboard Use EasyDMARC’s Domain Scanner to assess the security posture of each domain with a clear compliance score and risk level. The dashboard highlights configuration gaps and guides remediation steps, helping government agencies stay on track toward full compliance with the SGE Framework. 8. Inbound DMARC Evaluation Enforced You don’t need to apply any changes if you’re using Google Workspace, Microsoft 365, or other major mailbox providers. Most of them already enforce DMARC evaluation on incoming emails. However, some legacy Microsoft 365 setups may still quarantine emails that fail DMARC checks, even when the sending domain has a p=reject policy, instead of rejecting them. This behavior can be adjusted directly from your Microsoft Defender portal. about this in our step-by-step guide on how to set up SPF, DKIM, and DMARC from Microsoft Defender. If you’re using a third-party mail provider that doesn’t enforce having a DMARC policy for incoming emails, which is rare, you’ll need to contact their support to request a configuration change. 9. Data Loss Prevention Aligned with NZISM The New Zealand Information Security Manualis the New Zealand Government’s manual on information assurance and information systems security. It includes guidance on data loss prevention, which must be followed to be aligned with the SEG. Need Help Setting up SPF and DKIM for your Email Provider? Setting up SPF and DKIM for different ESPs often requires specific configurations. Some providers require you to publish SPF and DKIM on a subdomain, while others only require DKIM, or have different formatting rules. We’ve simplified all these steps to help you avoid misconfigurations that could delay your DMARC enforcement, or worse, block legitimate emails from reaching your recipients. Below you’ll find comprehensive setup guides for Google Workspace, Microsoft 365, Zoho Mail, Amazon SES, and SendGrid. You can also explore our full blog section that covers setup instructions for many other well-known ESPs. Remember, all this information is reflected in your DMARC aggregate reports. These reports give you live visibility into your outgoing email ecosystem, helping you analyze and fix any issues specific to a given provider. Here are our step-by-step guides for the most common platforms: Google Workspace Microsoft 365 These guides will help ensure your DNS records are configured correctly as part of the Secure Government EmailFramework rollout. Meet New Government Email Security Standards With EasyDMARC New Zealand’s SEG Framework sets a clear path for government agencies to enhance their email security by October 2025. With EasyDMARC, you can meet these technical requirements efficiently and with confidence. From protocol setup to continuous monitoring and compliance tracking, EasyDMARC streamlines the entire process, ensuring strong protection against spoofing, phishing, and data loss while simplifying your transition from SEEMail. #new #zealands #email #security #requirements
    EASYDMARC.COM
    New Zealand’s Email Security Requirements for Government Organizations: What You Need to Know
    The Secure Government Email (SGE) Common Implementation Framework New Zealand’s government is introducing a comprehensive email security framework designed to protect official communications from phishing and domain spoofing. This new framework, which will be mandatory for all government agencies by October 2025, establishes clear technical standards to enhance email security and retire the outdated SEEMail service.  Key Takeaways All NZ government agencies must comply with new email security requirements by October 2025. The new framework strengthens trust and security in government communications by preventing spoofing and phishing. The framework mandates TLS 1.2+, SPF, DKIM, DMARC with p=reject, MTA-STS, and DLP controls. EasyDMARC simplifies compliance with our guided setup, monitoring, and automated reporting. Start a Free Trial What is the Secure Government Email Common Implementation Framework? The Secure Government Email (SGE) Common Implementation Framework is a new government-led initiative in New Zealand designed to standardize email security across all government agencies. Its main goal is to secure external email communication, reduce domain spoofing in phishing attacks, and replace the legacy SEEMail service. Why is New Zealand Implementing New Government Email Security Standards? The framework was developed by New Zealand’s Department of Internal Affairs (DIA) as part of its role in managing ICT Common Capabilities. It leverages modern email security controls via the Domain Name System (DNS) to enable the retirement of the legacy SEEMail service and provide: Encryption for transmission security Digital signing for message integrity Basic non-repudiation (by allowing only authorized senders) Domain spoofing protection These improvements apply to all emails, not just those routed through SEEMail, offering broader protection across agency communications. What Email Security Technologies Are Required by the New NZ SGE Framework? The SGE Framework outlines the following key technologies that agencies must implement: TLS 1.2 or higher with implicit TLS enforced TLS-RPT (TLS Reporting) SPF (Sender Policy Framework) DKIM (DomainKeys Identified Mail) DMARC (Domain-based Message Authentication, Reporting, and Conformance) with reporting MTA-STS (Mail Transfer Agent Strict Transport Security) Data Loss Prevention controls These technologies work together to ensure encrypted email transmission, validate sender identity, prevent unauthorized use of domains, and reduce the risk of sensitive data leaks. Get in touch When Do NZ Government Agencies Need to Comply with this Framework? All New Zealand government agencies are expected to fully implement the Secure Government Email (SGE) Common Implementation Framework by October 2025. Agencies should begin their planning and deployment now to ensure full compliance by the deadline. The All of Government Secure Email Common Implementation Framework v1.0 What are the Mandated Requirements for Domains? Below are the exact requirements for all email-enabled domains under the new framework. ControlExact RequirementTLSMinimum TLS 1.2. TLS 1.1, 1.0, SSL, or clear-text not permitted.TLS-RPTAll email-sending domains must have TLS reporting enabled.SPFMust exist and end with -all.DKIMAll outbound email from every sending service must be DKIM-signed at the final hop.DMARCPolicy of p=reject on all email-enabled domains. adkim=s is recommended when not bulk-sending.MTA-STSEnabled and set to enforce.Implicit TLSMust be configured and enforced for every connection.Data Loss PreventionEnforce in line with the New Zealand Information Security Manual (NZISM) and Protective Security Requirements (PSR). Compliance Monitoring and Reporting The All of Government Service Delivery (AoGSD) team will be monitoring compliance with the framework. Monitoring will initially cover SPF, DMARC, and MTA-STS settings and will be expanded to include DKIM. Changes to these settings will be monitored, enabling reporting on email security compliance across all government agencies. Ongoing monitoring will highlight changes to domains, ensure new domains are set up with security in place, and monitor the implementation of future email security technologies.  Should compliance changes occur, such as an agency’s SPF record being changed from -all to ~all, this will be captured so that the AoGSD Security Team can investigate. They will then communicate directly with the agency to determine if an issue exists or if an error has occurred, reviewing each case individually. Deployment Checklist for NZ Government Compliance Enforce TLS 1.2 minimum, implicit TLS, MTA-STS & TLS-RPT SPF with -all DKIM on all outbound email DMARC p=reject  adkim=s where suitable For non-email/parked domains: SPF -all, empty DKIM, DMARC reject strict Compliance dashboard Inbound DMARC evaluation enforced DLP aligned with NZISM Start a Free Trial How EasyDMARC Can Help Government Agencies Comply EasyDMARC provides a comprehensive email security solution that simplifies the deployment and ongoing management of DNS-based email security protocols like SPF, DKIM, and DMARC with reporting. Our platform offers automated checks, real-time monitoring, and a guided setup to help government organizations quickly reach compliance. 1. TLS-RPT / MTA-STS audit EasyDMARC enables you to enable the Managed MTA-STS and TLS-RPT option with a single click. We provide the required DNS records and continuously monitor them for issues, delivering reports on TLS negotiation problems. This helps agencies ensure secure email transmission and quickly detect delivery or encryption failures. Note: In this screenshot, you can see how to deploy MTA-STS and TLS Reporting by adding just three CNAME records provided by EasyDMARC. It’s recommended to start in “testing” mode, evaluate the TLS-RPT reports, and then gradually switch your MTA-STS policy to “enforce”. The process is simple and takes just a few clicks. As shown above, EasyDMARC parses incoming TLS reports into a centralized dashboard, giving you clear visibility into delivery and encryption issues across all sending sources. 2. SPF with “-all”In the EasyDARC platform, you can run the SPF Record Generator to create a compliant record. Publish your v=spf1 record with “-all” to enforce a hard fail for unauthorized senders and prevent spoofed emails from passing SPF checks. This strengthens your domain’s protection against impersonation. Note: It is highly recommended to start adjusting your SPF record only after you begin receiving DMARC reports and identifying your legitimate email sources. As we’ll explain in more detail below, both SPF and DKIM should be adjusted after you gain visibility through reports. Making changes without proper visibility can lead to false positives, misconfigurations, and potential loss of legitimate emails. That’s why the first step should always be setting DMARC to p=none, receiving reports, analyzing them, and then gradually fixing any SPF or DKIM issues. 3. DKIM on all outbound email DKIM must be configured for all email sources sending emails on behalf of your domain. This is critical, as DKIM plays a bigger role than SPF when it comes to building domain reputation, surviving auto-forwarding, mailing lists, and other edge cases. As mentioned above, DMARC reports provide visibility into your email sources, allowing you to implement DKIM accordingly (see first screenshot). If you’re using third-party services like Google Workspace, Microsoft 365, or Mimecast, you’ll need to retrieve the public DKIM key from your provider’s admin interface (see second screenshot). EasyDMARC maintains a backend directory of over 1,400 email sources. We also give you detailed guidance on how to configure SPF and DKIM correctly for major ESPs.  Note: At the end of this article, you’ll find configuration links for well-known ESPs like Google Workspace, Microsoft 365, Zoho Mail, Amazon SES, and SendGrid – helping you avoid common misconfigurations and get aligned with SGE requirements. If you’re using a dedicated MTA (e.g., Postfix), DKIM must be implemented manually. EasyDMARC’s DKIM Record Generator lets you generate both public and private keys for your server. The private key is stored on your MTA, while the public key must be published in your DNS (see third and fourth screenshots). 4. DMARC p=reject rollout As mentioned in previous points, DMARC reporting is the first and most important step on your DMARC enforcement journey. Always start with a p=none policy and configure RUA reports to be sent to EasyDMARC. Use the report insights to identify and fix SPF and DKIM alignment issues, then gradually move to p=quarantine and finally p=reject once all legitimate email sources have been authenticated.  This phased approach ensures full protection against domain spoofing without risking legitimate email delivery. 5. adkim Strict Alignment Check This strict alignment check is not always applicable, especially if you’re using third-party bulk ESPs, such as Sendgrid, that require you to set DKIM on a subdomain level. You can set adkim=s in your DMARC TXT record, or simply enable strict mode in EasyDMARC’s Managed DMARC settings. This ensures that only emails with a DKIM signature that exactly match your domain pass alignment, adding an extra layer of protection against domain spoofing. But only do this if you are NOT a bulk sender. 6. Securing Non-Email Enabled Domains The purpose of deploying email security to non-email-enabled domains, or parked domains, is to prevent messages being spoofed from that domain. This requirement remains even if the root-level domain has SP=reject set within its DMARC record. Under this new framework, you must bulk import and mark parked domains as “Parked.” Crucially, this requires adjusting SPF settings to an empty record, setting DMARC to p=reject, and ensuring an empty DKIM record is in place: • SPF record: “v=spf1 -all”. • Wildcard DKIM record with empty public key.• DMARC record: “v=DMARC1;p=reject;adkim=s;aspf=s;rua=mailto:…”. EasyDMARC allows you to add and label parked domains for free. This is important because it helps you monitor any activity from these domains and ensure they remain protected with a strict DMARC policy of p=reject. 7. Compliance Dashboard Use EasyDMARC’s Domain Scanner to assess the security posture of each domain with a clear compliance score and risk level. The dashboard highlights configuration gaps and guides remediation steps, helping government agencies stay on track toward full compliance with the SGE Framework. 8. Inbound DMARC Evaluation Enforced You don’t need to apply any changes if you’re using Google Workspace, Microsoft 365, or other major mailbox providers. Most of them already enforce DMARC evaluation on incoming emails. However, some legacy Microsoft 365 setups may still quarantine emails that fail DMARC checks, even when the sending domain has a p=reject policy, instead of rejecting them. This behavior can be adjusted directly from your Microsoft Defender portal. Read more about this in our step-by-step guide on how to set up SPF, DKIM, and DMARC from Microsoft Defender. If you’re using a third-party mail provider that doesn’t enforce having a DMARC policy for incoming emails, which is rare, you’ll need to contact their support to request a configuration change. 9. Data Loss Prevention Aligned with NZISM The New Zealand Information Security Manual (NZISM) is the New Zealand Government’s manual on information assurance and information systems security. It includes guidance on data loss prevention (DLP), which must be followed to be aligned with the SEG. Need Help Setting up SPF and DKIM for your Email Provider? Setting up SPF and DKIM for different ESPs often requires specific configurations. Some providers require you to publish SPF and DKIM on a subdomain, while others only require DKIM, or have different formatting rules. We’ve simplified all these steps to help you avoid misconfigurations that could delay your DMARC enforcement, or worse, block legitimate emails from reaching your recipients. Below you’ll find comprehensive setup guides for Google Workspace, Microsoft 365, Zoho Mail, Amazon SES, and SendGrid. You can also explore our full blog section that covers setup instructions for many other well-known ESPs. Remember, all this information is reflected in your DMARC aggregate reports. These reports give you live visibility into your outgoing email ecosystem, helping you analyze and fix any issues specific to a given provider. Here are our step-by-step guides for the most common platforms: Google Workspace Microsoft 365 These guides will help ensure your DNS records are configured correctly as part of the Secure Government Email (SGE) Framework rollout. Meet New Government Email Security Standards With EasyDMARC New Zealand’s SEG Framework sets a clear path for government agencies to enhance their email security by October 2025. With EasyDMARC, you can meet these technical requirements efficiently and with confidence. From protocol setup to continuous monitoring and compliance tracking, EasyDMARC streamlines the entire process, ensuring strong protection against spoofing, phishing, and data loss while simplifying your transition from SEEMail.
    0 Commentarios 0 Acciones
  • How AI is reshaping the future of healthcare and medical research

    Transcript       
    PETER LEE: “In ‘The Little Black Bag,’ a classic science fiction story, a high-tech doctor’s kit of the future is accidentally transported back to the 1950s, into the shaky hands of a washed-up, alcoholic doctor. The ultimate medical tool, it redeems the doctor wielding it, allowing him to practice gratifyingly heroic medicine. … The tale ends badly for the doctor and his treacherous assistant, but it offered a picture of how advanced technology could transform medicine—powerful when it was written nearly 75 years ago and still so today. What would be the Al equivalent of that little black bag? At this moment when new capabilities are emerging, how do we imagine them into medicine?”          
    This is The AI Revolution in Medicine, Revisited. I’m your host, Peter Lee.   
    Shortly after OpenAI’s GPT-4 was publicly released, Carey Goldberg, Dr. Zak Kohane, and I published The AI Revolution in Medicine to help educate the world of healthcare and medical research about the transformative impact this new generative AI technology could have. But because we wrote the book when GPT-4 was still a secret, we had to speculate. Now, two years later, what did we get right, and what did we get wrong?    
    In this series, we’ll talk to clinicians, patients, hospital administrators, and others to understand the reality of AI in the field and where we go from here.  The book passage I read at the top is from “Chapter 10: The Big Black Bag.” 
    In imagining AI in medicine, Carey, Zak, and I included in our book two fictional accounts. In the first, a medical resident consults GPT-4 on her personal phone as the patient in front of her crashes. Within seconds, it offers an alternate response based on recent literature. In the second account, a 90-year-old woman with several chronic conditions is living independently and receiving near-constant medical support from an AI aide.   
    In our conversations with the guests we’ve spoken to so far, we’ve caught a glimpse of these predicted futures, seeing how clinicians and patients are actually using AI today and how developers are leveraging the technology in the healthcare products and services they’re creating. In fact, that first fictional account isn’t so fictional after all, as most of the doctors in the real world actually appear to be using AI at least occasionally—and sometimes much more than occasionally—to help in their daily clinical work. And as for the second fictional account, which is more of a science fiction account, it seems we are indeed on the verge of a new way of delivering and receiving healthcare, though the future is still very much open. 
    As we continue to examine the current state of AI in healthcare and its potential to transform the field, I’m pleased to welcome Bill Gates and Sébastien Bubeck.  
    Bill may be best known as the co-founder of Microsoft, having created the company with his childhood friend Paul Allen in 1975. He’s now the founder of Breakthrough Energy, which aims to advance clean energy innovation, and TerraPower, a company developing groundbreaking nuclear energy and science technologies. He also chairs the world’s largest philanthropic organization, the Gates Foundation, and focuses on solving a variety of health challenges around the globe and here at home. 
    Sébastien is a research lead at OpenAI. He was previously a distinguished scientist, vice president of AI, and a colleague of mine here at Microsoft, where his work included spearheading the development of the family of small language models known as Phi. While at Microsoft, he also coauthored the discussion-provoking 2023 paper “Sparks of Artificial General Intelligence,” which presented the results of early experiments with GPT-4 conducted by a small team from Microsoft Research.     
    Here’s my conversation with Bill Gates and Sébastien Bubeck. 
    LEE: Bill, welcome. 
    BILL GATES: Thank you. 
    LEE: Seb … 
    SÉBASTIEN BUBECK: Yeah. Hi, hi, Peter. Nice to be here. 
    LEE: You know, one of the things that I’ve been doing just to get the conversation warmed up is to talk about origin stories, and what I mean about origin stories is, you know, what was the first contact that you had with large language models or the concept of generative AI that convinced you or made you think that something really important was happening? 
    And so, Bill, I think I’ve heard the story about, you know, the time when the OpenAI folks—Sam Altman, Greg Brockman, and others—showed you something, but could we hear from you what those early encounters were like and what was going through your mind?  
    GATES: Well, I’d been visiting OpenAI soon after it was created to see things like GPT-2 and to see the little arm they had that was trying to match human manipulation and, you know, looking at their games like Dota that they were trying to get as good as human play. And honestly, I didn’t think the language model stuff they were doing, even when they got to GPT-3, would show the ability to learn, you know, in the same sense that a human reads a biology book and is able to take that knowledge and access it not only to pass a test but also to create new medicines. 
    And so my challenge to them was that if their LLM could get a five on the advanced placement biology test, then I would say, OK, it took biologic knowledge and encoded it in an accessible way and that I didn’t expect them to do that very quickly but it would be profound.  
    And it was only about six months after I challenged them to do that, that an early version of GPT-4 they brought up to a dinner at my house, and in fact, it answered most of the questions that night very well. The one it got totally wrong, we were … because it was so good, we kept thinking, Oh, we must be wrong. It turned out it was a math weaknessthat, you know, we later understood that that was an area of, weirdly, of incredible weakness of those early models. But, you know, that was when I realized, OK, the age of cheap intelligence was at its beginning. 
    LEE: Yeah. So I guess it seems like you had something similar to me in that my first encounters, I actually harbored some skepticism. Is it fair to say you were skeptical before that? 
    GATES: Well, the idea that we’ve figured out how to encode and access knowledge in this very deep sense without even understanding the nature of the encoding, … 
    LEE: Right.  
    GATES: … that is a bit weird.  
    LEE: Yeah. 
    GATES: We have an algorithm that creates the computation, but even say, OK, where is the president’s birthday stored in there? Where is this fact stored in there? The fact that even now when we’re playing around, getting a little bit more sense of it, it’s opaque to us what the semantic encoding is, it’s, kind of, amazing to me. I thought the invention of knowledge storage would be an explicit way of encoding knowledge, not an implicit statistical training. 
    LEE: Yeah, yeah. All right. So, Seb, you know, on this same topic, you know, I got—as we say at Microsoft—I got pulled into the tent. 
    BUBECK: Yes.  
    LEE: Because this was a very secret project. And then, um, I had the opportunity to select a small number of researchers in MSRto join and start investigating this thing seriously. And the first person I pulled in was you. 
    BUBECK: Yeah. 
    LEE: And so what were your first encounters? Because I actually don’t remember what happened then. 
    BUBECK: Oh, I remember it very well.My first encounter with GPT-4 was in a meeting with the two of you, actually. But my kind of first contact, the first moment where I realized that something was happening with generative AI, was before that. And I agree with Bill that I also wasn’t too impressed by GPT-3. 
    I though that it was kind of, you know, very naturally mimicking the web, sort of parroting what was written there in a nice way. Still in a way which seemed very impressive. But it wasn’t really intelligent in any way. But shortly after GPT-3, there was a model before GPT-4 that really shocked me, and this was the first image generation model, DALL-E 1. 
    So that was in 2021. And I will forever remember the press release of OpenAI where they had this prompt of an avocado chair and then you had this image of the avocado chair.And what really shocked me is that clearly the model kind of “understood” what is a chair, what is an avocado, and was able to merge those concepts. 
    So this was really, to me, the first moment where I saw some understanding in those models.  
    LEE: So this was, just to get the timing right, that was before I pulled you into the tent. 
    BUBECK: That was before. That was like a year before. 
    LEE: Right.  
    BUBECK: And now I will tell you how, you know, we went from that moment to the meeting with the two of you and GPT-4. 
    So once I saw this kind of understanding, I thought, OK, fine. It understands concept, but it’s still not able to reason. It cannot—as, you know, Bill was saying—it cannot learn from your document. It cannot reason.  
    So I set out to try to prove that. You know, this is what I was in the business of at the time, trying to prove things in mathematics. So I was trying to prove that basically autoregressive transformers could never reason. So I was trying to prove this. And after a year of work, I had something reasonable to show. And so I had the meeting with the two of you, and I had this example where I wanted to say, there is no way that an LLM is going to be able to do x. 
    And then as soon as I … I don’t know if you remember, Bill. But as soon as I said that, you said, oh, but wait a second. I had, you know, the OpenAI crew at my house recently, and they showed me a new model. Why don’t we ask this new model this question?  
    LEE: Yeah.
    BUBECK: And we did, and it solved it on the spot. And that really, honestly, just changed my life. Like, you know, I had been working for a year trying to say that this was impossible. And just right there, it was shown to be possible.  
    LEE:One of the very first things I got interested in—because I was really thinking a lot about healthcare—was healthcare and medicine. 
    And I don’t know if the two of you remember, but I ended up doing a lot of tests. I ran through, you know, step one and step two of the US Medical Licensing Exam. Did a whole bunch of other things. I wrote this big report. It was, you know, I can’t remember … a couple hundred pages.  
    And I needed to share this with someone. I didn’t … there weren’t too many people I could share it with. So I sent, I think, a copy to you, Bill. Sent a copy to you, Seb.  
    I hardly slept for about a week putting that report together. And, yeah, and I kept working on it. But I was far from alone. I think everyone who was in the tent, so to speak, in those early days was going through something pretty similar. All right. So I think … of course, a lot of what I put in the report also ended up being examples that made it into the book. 
    But the main purpose of this conversation isn’t to reminisce aboutor indulge in those reminiscences but to talk about what’s happening in healthcare and medicine. And, you know, as I said, we wrote this book. We did it very, very quickly. Seb, you helped. Bill, you know, you provided a review and some endorsements. 
    But, you know, honestly, we didn’t know what we were talking about because no one had access to this thing. And so we just made a bunch of guesses. So really, the whole thing I wanted to probe with the two of you is, now with two years of experience out in the world, what, you know, what do we think is happening today? 
    You know, is AI actually having an impact, positive or negative, on healthcare and medicine? And what do we now think is going to happen in the next two years, five years, or 10 years? And so I realize it’s a little bit too abstract to just ask it that way. So let me just try to narrow the discussion and guide us a little bit.  
    Um, the kind of administrative and clerical work, paperwork, around healthcare—and we made a lot of guesses about that—that appears to be going well, but, you know, Bill, I know we’ve discussed that sometimes that you think there ought to be a lot more going on. Do you have a viewpoint on how AI is actually finding its way into reducing paperwork? 
    GATES: Well, I’m stunned … I don’t think there should be a patient-doctor meeting where the AI is not sitting in and both transcribing, offering to help with the paperwork, and even making suggestions, although the doctor will be the one, you know, who makes the final decision about the diagnosis and whatever prescription gets done.  
    It’s so helpful. You know, when that patient goes home and their, you know, son who wants to understand what happened has some questions, that AI should be available to continue that conversation. And the way you can improve that experience and streamline things and, you know, involve the people who advise you. I don’t understand why that’s not more adopted, because there you still have the human in the loop making that final decision. 
    But even for, like, follow-up calls to make sure the patient did things, to understand if they have concerns and knowing when to escalate back to the doctor, the benefit is incredible. And, you know, that thing is ready for prime time. That paradigm is ready for prime time, in my view. 
    LEE: Yeah, there are some good products, but it seems like the number one use right now—and we kind of got this from some of the previous guests in previous episodes—is the use of AI just to respond to emails from patients.Does that make sense to you? 
    BUBECK: Yeah. So maybe I want to second what Bill was saying but maybe take a step back first. You know, two years ago, like, the concept of clinical scribes, which is one of the things that we’re talking about right now, it would have sounded, in fact, it sounded two years ago, borderline dangerous. Because everybody was worried about hallucinations. What happened if you have this AI listening in and then it transcribes, you know, something wrong? 
    Now, two years later, I think it’s mostly working. And in fact, it is not yet, you know, fully adopted. You’re right. But it is in production. It is used, you know, in many, many places. So this rate of progress is astounding because it wasn’t obvious that we would be able to overcome those obstacles of hallucination. It’s not to say that hallucinations are fully solved. In the case of the closed system, they are.  
    Now, I think more generally what’s going on in the background is that there is something that we, that certainly I, underestimated, which is this management overhead. So I think the reason why this is not adopted everywhere is really a training and teaching aspect. People need to be taught, like, those systems, how to interact with them. 
    And one example that I really like, a study that recently appeared where they tried to use ChatGPT for diagnosis and they were comparing doctors without and with ChatGPT. And the amazing thing … so this was a set of cases where the accuracy of the doctors alone was around 75%. ChatGPT alone was 90%. So that’s already kind of mind blowing. But then the kicker is that doctors with ChatGPT was 80%.  
    Intelligence alone is not enough. It’s also how it’s presented, how you interact with it. And ChatGPT, it’s an amazing tool. Obviously, I absolutely love it. But it’s not … you don’t want a doctor to have to type in, you know, prompts and use it that way. 
    It should be, as Bill was saying, kind of running continuously in the background, sending you notifications. And you have to be really careful of the rate at which those notifications are being sent. Because if they are too frequent, then the doctor will learn to ignore them. So you have to … all of those things matter, in fact, at least as much as the level of intelligence of the machine. 
    LEE: One of the things I think about, Bill, in that scenario that you described, doctors do some thinking about the patient when they write the note. So, you know, I’m always a little uncertain whether it’s actually … you know, you wouldn’t necessarily want to fully automate this, I don’t think. Or at least there needs to be some prompt to the doctor to make sure that the doctor puts some thought into what happened in the encounter with the patient. Does that make sense to you at all? 
    GATES: At this stage, you know, I’d still put the onus on the doctor to write the conclusions and the summary and not delegate that. 
    The tradeoffs you make a little bit are somewhat dependent on the situation you’re in. If you’re in Africa,
    So, yes, the doctor’s still going to have to do a lot of work, but just the quality of letting the patient and the people around them interact and ask questions and have things explained, that alone is such a quality improvement. It’s mind blowing.  
    LEE: So since you mentioned, you know, Africa—and, of course, this touches on the mission and some of the priorities of the Gates Foundation and this idea of democratization of access to expert medical care—what’s the most interesting stuff going on right now? Are there people and organizations or technologies that are impressing you or that you’re tracking? 
    GATES: Yeah. So the Gates Foundation has given out a lot of grants to people in Africa doing education, agriculture but more healthcare examples than anything. And the way these things start off, they often start out either being patient-centric in a narrow situation, like, OK, I’m a pregnant woman; talk to me. Or, I have infectious disease symptoms; talk to me. Or they’re connected to a health worker where they’re helping that worker get their job done. And we have lots of pilots out, you know, in both of those cases.  
    The dream would be eventually to have the thing the patient consults be so broad that it’s like having a doctor available who understands the local things.  
    LEE: Right.  
    GATES: We’re not there yet. But over the next two or three years, you know, particularly given the worsening financial constraints against African health systems, where the withdrawal of money has been dramatic, you know, figuring out how to take this—what I sometimes call “free intelligence”—and build a quality health system around that, we will have to be more radical in low-income countries than any rich country is ever going to be.  
    LEE: Also, there’s maybe a different regulatory environment, so some of those things maybe are easier? Because right now, I think the world hasn’t figured out how to and whether to regulate, let’s say, an AI that might give a medical diagnosis or write a prescription for a medication. 
    BUBECK: Yeah. I think one issue with this, and it’s also slowing down the deployment of AI in healthcare more generally, is a lack of proper benchmark. Because, you know, you were mentioning the USMLE, for example. That’s a great test to test human beings and their knowledge of healthcare and medicine. But it’s not a great test to give to an AI. 
    It’s not asking the right questions. So finding what are the right questions to test whether an AI system is ready to give diagnosis in a constrained setting, that’s a very, very important direction, which to my surprise, is not yet accelerating at the rate that I was hoping for. 
    LEE: OK, so that gives me an excuse to get more now into the core AI tech because something I’ve discussed with both of you is this issue of what are the right tests. And you both know the very first test I give to any new spin of an LLM is I present a patient, the results—a mythical patient—the results of my physical exam, my mythical physical exam. Maybe some results of some initial labs. And then I present or propose a differential diagnosis. And if you’re not in medicine, a differential diagnosis you can just think of as a prioritized list of the possible diagnoses that fit with all that data. And in that proposed differential, I always intentionally make two mistakes. 
    I make a textbook technical error in one of the possible elements of the differential diagnosis, and I have an error of omission. And, you know, I just want to know, does the LLM understand what I’m talking about? And all the good ones out there do now. But then I want to know, can it spot the errors? And then most importantly, is it willing to tell me I’m wrong, that I’ve made a mistake?  
    That last piece seems really hard for AI today. And so let me ask you first, Seb, because at the time of this taping, of course, there was a new spin of GPT-4o last week that became overly sycophantic. In other words, it was actually prone in that test of mine not only to not tell me I’m wrong, but it actually praised me for the creativity of my differential.What’s up with that? 
    BUBECK: Yeah, I guess it’s a testament to the fact that training those models is still more of an art than a science. So it’s a difficult job. Just to be clear with the audience, we have rolled back thatversion of GPT-4o, so now we don’t have the sycophant version out there. 
    Yeah, no, it’s a really difficult question. It has to do … as you said, it’s very technical. It has to do with the post-training and how, like, where do you nudge the model? So, you know, there is this very classical by now technique called RLHF, where you push the model in the direction of a certain reward model. So the reward model is just telling the model, you know, what behavior is good, what behavior is bad. 
    But this reward model is itself an LLM, and, you know, Bill was saying at the very beginning of the conversation that we don’t really understand how those LLMs deal with concepts like, you know, where is the capital of France located? Things like that. It is the same thing for this reward model. We don’t know why it says that it prefers one output to another, and whether this is correlated with some sycophancy is, you know, something that we discovered basically just now. That if you push too hard in optimization on this reward model, you will get a sycophant model. 
    So it’s kind of … what I’m trying to say is we became too good at what we were doing, and we ended up, in fact, in a trap of the reward model. 
    LEE: I mean, you do want … it’s a difficult balance because you do want models to follow your desires and … 
    BUBECK: It’s a very difficult, very difficult balance. 
    LEE: So this brings up then the following question for me, which is the extent to which we think we’ll need to have specially trained models for things. So let me start with you, Bill. Do you have a point of view on whether we will need to, you know, quote-unquote take AI models to med school? Have them specially trained? Like, if you were going to deploy something to give medical care in underserved parts of the world, do we need to do something special to create those models? 
    GATES: We certainly need to teach them the African languages and the unique dialects so that the multimedia interactions are very high quality. We certainly need to teach them the disease prevalence and unique disease patterns like, you know, neglected tropical diseases and malaria. So we need to gather a set of facts that somebody trying to go for a US customer base, you know, wouldn’t necessarily have that in there. 
    Those two things are actually very straightforward because the additional training time is small. I’d say for the next few years, we’ll also need to do reinforcement learning about the context of being a doctor and how important certain behaviors are. Humans learn over the course of their life to some degree that, I’m in a different context and the way I behave in terms of being willing to criticize or be nice, you know, how important is it? Who’s here? What’s my relationship to them?  
    Right now, these machines don’t have that broad social experience. And so if you know it’s going to be used for health things, a lot of reinforcement learning of the very best humans in that context would still be valuable. Eventually, the models will, having read all the literature of the world about good doctors, bad doctors, it’ll understand as soon as you say, “I want you to be a doctor diagnosing somebody.” All of the implicit reinforcement that fits that situation, you know, will be there.
    LEE: Yeah.
    GATES: And so I hope three years from now, we don’t have to do that reinforcement learning. But today, for any medical context, you would want a lot of data to reinforce tone, willingness to say things when, you know, there might be something significant at stake. 
    LEE: Yeah. So, you know, something Bill said, kind of, reminds me of another thing that I think we missed, which is, the context also … and the specialization also pertains to different, I guess, what we still call “modes,” although I don’t know if the idea of multimodal is the same as it was two years ago. But, you know, what do you make of all of the hubbub around—in fact, within Microsoft Research, this is a big deal, but I think we’re far from alone—you know, medical images and vision, video, proteins and molecules, cell, you know, cellular data and so on. 
    BUBECK: Yeah. OK. So there is a lot to say to everything … to the last, you know, couple of minutes. Maybe on the specialization aspect, you know, I think there is, hiding behind this, a really fundamental scientific question of whether eventually we have a singular AGIthat kind of knows everything and you can just put, you know, explain your own context and it will just get it and understand everything. 
    That’s one vision. I have to say, I don’t particularly believe in this vision. In fact, we humans are not like that at all. I think, hopefully, we are general intelligences, yet we have to specialize a lot. And, you know, I did myself a lot of RL, reinforcement learning, on mathematics. Like, that’s what I did, you know, spent a lot of time doing that. And I didn’t improve on other aspects. You know, in fact, I probably degraded in other aspects.So it’s … I think it’s an important example to have in mind. 
    LEE: I think I might disagree with you on that, though, because, like, doesn’t a model have to see both good science and bad science in order to be able to gain the ability to discern between the two? 
    BUBECK: Yeah, no, that absolutely. I think there is value in seeing the generality, in having a very broad base. But then you, kind of, specialize on verticals. And this is where also, you know, open-weights model, which we haven’t talked about yet, are really important because they allow you to provide this broad base to everyone. And then you can specialize on top of it. 
    LEE: So we have about three hours of stuff to talk about, but our time is actually running low.
    BUBECK: Yes, yes, yes.  
    LEE: So I think I want … there’s a more provocative question. It’s almost a silly question, but I need to ask it of the two of you, which is, is there a future, you know, where AI replaces doctors or replaces, you know, medical specialties that we have today? So what does the world look like, say, five years from now? 
    GATES: Well, it’s important to distinguish healthcare discovery activity from healthcare delivery activity. We focused mostly on delivery. I think it’s very much within the realm of possibility that the AI is not only accelerating healthcare discovery but substituting for a lot of the roles of, you know, I’m an organic chemist, or I run various types of assays. I can see those, which are, you know, testable-output-type jobs but with still very high value, I can see, you know, some replacement in those areas before the doctor.  
    The doctor, still understanding the human condition and long-term dialogues, you know, they’ve had a lifetime of reinforcement of that, particularly when you get into areas like mental health. So I wouldn’t say in five years, either people will choose to adopt it, but it will be profound that there’ll be this nearly free intelligence that can do follow-up, that can help you, you know, make sure you went through different possibilities. 
    And so I’d say, yes, we’ll have doctors, but I’d say healthcare will be massively transformed in its quality and in efficiency by AI in that time period. 
    LEE: Is there a comparison, useful comparison, say, between doctors and, say, programmers, computer programmers, or doctors and, I don’t know, lawyers? 
    GATES: Programming is another one that has, kind of, a mathematical correctness to it, you know, and so the objective function that you’re trying to reinforce to, as soon as you can understand the state machines, you can have something that’s “checkable”; that’s correct. So I think programming, you know, which is weird to say, that the machine will beat us at most programming tasks before we let it take over roles that have deep empathy, you know, physical presence and social understanding in them. 
    LEE: Yeah. By the way, you know, I fully expect in five years that AI will produce mathematical proofs that are checkable for validity, easily checkable, because they’ll be written in a proof-checking language like Lean or something but will be so complex that no human mathematician can understand them. I expect that to happen.  
    I can imagine in some fields, like cellular biology, we could have the same situation in the future because the molecular pathways, the chemistry, biochemistry of human cells or living cells is as complex as any mathematics, and so it seems possible that we may be in a state where in wet lab, we see, Oh yeah, this actually works, but no one can understand why. 
    BUBECK: Yeah, absolutely. I mean, I think I really agree with Bill’s distinction of the discovery and the delivery, and indeed, the discovery’s when you can check things, and at the end, there is an artifact that you can verify. You know, you can run the protocol in the wet lab and seeproduced what you wanted. So I absolutely agree with that.  
    And in fact, you know, we don’t have to talk five years from now. I don’t know if you know, but just recently, there was a paper that was published on a scientific discovery using o3- mini. So this is really amazing. And, you know, just very quickly, just so people know, it was about this statistical physics model, the frustrated Potts model, which has to do with coloring, and basically, the case of three colors, like, more than two colors was open for a long time, and o3 was able to reduce the case of three colors to two colors.  
    LEE: Yeah. 
    BUBECK: Which is just, like, astounding. And this is not … this is now. This is happening right now. So this is something that I personally didn’t expect it would happen so quickly, and it’s due to those reasoning models.  
    Now, on the delivery side, I would add something more to it for the reason why doctors and, in fact, lawyers and coders will remain for a long time, and it’s because we still don’t understand how those models generalize. Like, at the end of the day, we are not able to tell you when they are confronted with a really new, novel situation, whether they will work or not. 
    Nobody is able to give you that guarantee. And I think until we understand this generalization better, we’re not going to be willing to just let the system in the wild without human supervision. 
    LEE: But don’t human doctors, human specialists … so, for example, a cardiologist sees a patient in a certain way that a nephrologist … 
    BUBECK: Yeah.
    LEE: … or an endocrinologist might not.
    BUBECK: That’s right. But another cardiologist will understand and, kind of, expect a certain level of generalization from their peer. And this, we just don’t have it with AI models. Now, of course, you’re exactly right. That generalization is also hard for humans. Like, if you have a human trained for one task and you put them into another task, then you don’t … you often don’t know.
    LEE: OK. You know, the podcast is focused on what’s happened over the last two years. But now, I’d like one provocative prediction about what you think the world of AI and medicine is going to be at some point in the future. You pick your timeframe. I don’t care if it’s two years or 20 years from now, but, you know, what do you think will be different about AI in medicine in that future than today? 
    BUBECK: Yeah, I think the deployment is going to accelerate soon. Like, we’re really not missing very much. There is this enormous capability overhang. Like, even if progress completely stopped, with current systems, we can do a lot more than what we’re doing right now. So I think this will … this has to be realized, you know, sooner rather than later. 
    And I think it’s probably dependent on these benchmarks and proper evaluation and tying this with regulation. So these are things that take time in human society and for good reason. But now we already are at two years; you know, give it another two years and it should be really …  
    LEE: Will AI prescribe your medicines? Write your prescriptions? 
    BUBECK: I think yes. I think yes. 
    LEE: OK. Bill? 
    GATES: Well, I think the next two years, we’ll have massive pilots, and so the amount of use of the AI, still in a copilot-type mode, you know, we should get millions of patient visits, you know, both in general medicine and in the mental health side, as well. And I think that’s going to build up both the data and the confidence to give the AI some additional autonomy. You know, are you going to let it talk to you at night when you’re panicked about your mental health with some ability to escalate?
    And, you know, I’ve gone so far as to tell politicians with national health systems that if they deploy AI appropriately, that the quality of care, the overload of the doctors, the improvement in the economics will be enough that their voters will be stunned because they just don’t expect this, and, you know, they could be reelectedjust on this one thing of fixing what is a very overloaded and economically challenged health system in these rich countries. 
    You know, my personal role is going to be to make sure that in the poorer countries, there isn’t some lag; in fact, in many cases, that we’ll be more aggressive because, you know, we’re comparing to having no access to doctors at all. And, you know, so I think whether it’s India or Africa, there’ll be lessons that are globally valuable because we need medical intelligence. And, you know, thank god AI is going to provide a lot of that. 
    LEE: Well, on that optimistic note, I think that’s a good way to end. Bill, Seb, really appreciate all of this.  
    I think the most fundamental prediction we made in the book is that AI would actually find its way into the practice of medicine, and I think that that at least has come true, maybe in different ways than we expected, but it’s come true, and I think it’ll only accelerate from here. So thanks again, both of you.  
    GATES: Yeah. Thanks, you guys. 
    BUBECK: Thank you, Peter. Thanks, Bill. 
    LEE: I just always feel such a sense of privilege to have a chance to interact and actually work with people like Bill and Sébastien.   
    With Bill, I’m always amazed at how practically minded he is. He’s really thinking about the nuts and bolts of what AI might be able to do for people, and his thoughts about underserved parts of the world, the idea that we might actually be able to empower people with access to expert medical knowledge, I think is both inspiring and amazing.  
    And then, Seb, Sébastien Bubeck, he’s just absolutely a brilliant mind. He has a really firm grip on the deep mathematics of artificial intelligence and brings that to bear in his research and development work. And where that mathematics takes him isn’t just into the nuts and bolts of algorithms but into philosophical questions about the nature of intelligence.  
    One of the things that Sébastien brought up was the state of evaluation of AI systems. And indeed, he was fairly critical in our conversation. But of course, the world of AI research and development is just moving so fast, and indeed, since we recorded our conversation, OpenAI, in fact, released a new evaluation metric that is directly relevant to medical applications, and that is something called HealthBench. And Microsoft Research also released a new evaluation approach or process called ADeLe.  
    HealthBench and ADeLe are examples of new approaches to evaluating AI models that are less about testing their knowledge and ability to pass multiple-choice exams and instead are evaluation approaches designed to assess how well AI models are able to complete tasks that actually arise every day in typical healthcare or biomedical research settings. These are examples of really important good work that speak to how well AI models work in the real world of healthcare and biomedical research and how well they can collaborate with human beings in those settings. 
    You know, I asked Bill and Seb to make some predictions about the future. You know, my own answer, I expect that we’re going to be able to use AI to change how we diagnose patients, change how we decide treatment options.  
    If you’re a doctor or a nurse and you encounter a patient, you’ll ask questions, do a physical exam, you know, call out for labs just like you do today, but then you’ll be able to engage with AI based on all of that data and just ask, you know, based on all the other people who have gone through the same experience, who have similar data, how were they diagnosed? How were they treated? What were their outcomes? And what does that mean for the patient I have right now? Some people call it the “patients like me” paradigm. And I think that’s going to become real because of AI within our lifetimes. That idea of really grounding the delivery in healthcare and medical practice through data and intelligence, I actually now don’t see any barriers to that future becoming real.  
    I’d like to extend another big thank you to Bill and Sébastien for their time. And to our listeners, as always, it’s a pleasure to have you along for the ride. I hope you’ll join us for our remaining conversations, as well as a second coauthor roundtable with Carey and Zak.  
    Until next time.  
    #how #reshaping #future #healthcare #medical
    How AI is reshaping the future of healthcare and medical research
    Transcript        PETER LEE: “In ‘The Little Black Bag,’ a classic science fiction story, a high-tech doctor’s kit of the future is accidentally transported back to the 1950s, into the shaky hands of a washed-up, alcoholic doctor. The ultimate medical tool, it redeems the doctor wielding it, allowing him to practice gratifyingly heroic medicine. … The tale ends badly for the doctor and his treacherous assistant, but it offered a picture of how advanced technology could transform medicine—powerful when it was written nearly 75 years ago and still so today. What would be the Al equivalent of that little black bag? At this moment when new capabilities are emerging, how do we imagine them into medicine?”           This is The AI Revolution in Medicine, Revisited. I’m your host, Peter Lee.    Shortly after OpenAI’s GPT-4 was publicly released, Carey Goldberg, Dr. Zak Kohane, and I published The AI Revolution in Medicine to help educate the world of healthcare and medical research about the transformative impact this new generative AI technology could have. But because we wrote the book when GPT-4 was still a secret, we had to speculate. Now, two years later, what did we get right, and what did we get wrong?     In this series, we’ll talk to clinicians, patients, hospital administrators, and others to understand the reality of AI in the field and where we go from here.  The book passage I read at the top is from “Chapter 10: The Big Black Bag.”  In imagining AI in medicine, Carey, Zak, and I included in our book two fictional accounts. In the first, a medical resident consults GPT-4 on her personal phone as the patient in front of her crashes. Within seconds, it offers an alternate response based on recent literature. In the second account, a 90-year-old woman with several chronic conditions is living independently and receiving near-constant medical support from an AI aide.    In our conversations with the guests we’ve spoken to so far, we’ve caught a glimpse of these predicted futures, seeing how clinicians and patients are actually using AI today and how developers are leveraging the technology in the healthcare products and services they’re creating. In fact, that first fictional account isn’t so fictional after all, as most of the doctors in the real world actually appear to be using AI at least occasionally—and sometimes much more than occasionally—to help in their daily clinical work. And as for the second fictional account, which is more of a science fiction account, it seems we are indeed on the verge of a new way of delivering and receiving healthcare, though the future is still very much open.  As we continue to examine the current state of AI in healthcare and its potential to transform the field, I’m pleased to welcome Bill Gates and Sébastien Bubeck.   Bill may be best known as the co-founder of Microsoft, having created the company with his childhood friend Paul Allen in 1975. He’s now the founder of Breakthrough Energy, which aims to advance clean energy innovation, and TerraPower, a company developing groundbreaking nuclear energy and science technologies. He also chairs the world’s largest philanthropic organization, the Gates Foundation, and focuses on solving a variety of health challenges around the globe and here at home.  Sébastien is a research lead at OpenAI. He was previously a distinguished scientist, vice president of AI, and a colleague of mine here at Microsoft, where his work included spearheading the development of the family of small language models known as Phi. While at Microsoft, he also coauthored the discussion-provoking 2023 paper “Sparks of Artificial General Intelligence,” which presented the results of early experiments with GPT-4 conducted by a small team from Microsoft Research.      Here’s my conversation with Bill Gates and Sébastien Bubeck.  LEE: Bill, welcome.  BILL GATES: Thank you.  LEE: Seb …  SÉBASTIEN BUBECK: Yeah. Hi, hi, Peter. Nice to be here.  LEE: You know, one of the things that I’ve been doing just to get the conversation warmed up is to talk about origin stories, and what I mean about origin stories is, you know, what was the first contact that you had with large language models or the concept of generative AI that convinced you or made you think that something really important was happening?  And so, Bill, I think I’ve heard the story about, you know, the time when the OpenAI folks—Sam Altman, Greg Brockman, and others—showed you something, but could we hear from you what those early encounters were like and what was going through your mind?   GATES: Well, I’d been visiting OpenAI soon after it was created to see things like GPT-2 and to see the little arm they had that was trying to match human manipulation and, you know, looking at their games like Dota that they were trying to get as good as human play. And honestly, I didn’t think the language model stuff they were doing, even when they got to GPT-3, would show the ability to learn, you know, in the same sense that a human reads a biology book and is able to take that knowledge and access it not only to pass a test but also to create new medicines.  And so my challenge to them was that if their LLM could get a five on the advanced placement biology test, then I would say, OK, it took biologic knowledge and encoded it in an accessible way and that I didn’t expect them to do that very quickly but it would be profound.   And it was only about six months after I challenged them to do that, that an early version of GPT-4 they brought up to a dinner at my house, and in fact, it answered most of the questions that night very well. The one it got totally wrong, we were … because it was so good, we kept thinking, Oh, we must be wrong. It turned out it was a math weaknessthat, you know, we later understood that that was an area of, weirdly, of incredible weakness of those early models. But, you know, that was when I realized, OK, the age of cheap intelligence was at its beginning.  LEE: Yeah. So I guess it seems like you had something similar to me in that my first encounters, I actually harbored some skepticism. Is it fair to say you were skeptical before that?  GATES: Well, the idea that we’ve figured out how to encode and access knowledge in this very deep sense without even understanding the nature of the encoding, …  LEE: Right.   GATES: … that is a bit weird.   LEE: Yeah.  GATES: We have an algorithm that creates the computation, but even say, OK, where is the president’s birthday stored in there? Where is this fact stored in there? The fact that even now when we’re playing around, getting a little bit more sense of it, it’s opaque to us what the semantic encoding is, it’s, kind of, amazing to me. I thought the invention of knowledge storage would be an explicit way of encoding knowledge, not an implicit statistical training.  LEE: Yeah, yeah. All right. So, Seb, you know, on this same topic, you know, I got—as we say at Microsoft—I got pulled into the tent.  BUBECK: Yes.   LEE: Because this was a very secret project. And then, um, I had the opportunity to select a small number of researchers in MSRto join and start investigating this thing seriously. And the first person I pulled in was you.  BUBECK: Yeah.  LEE: And so what were your first encounters? Because I actually don’t remember what happened then.  BUBECK: Oh, I remember it very well.My first encounter with GPT-4 was in a meeting with the two of you, actually. But my kind of first contact, the first moment where I realized that something was happening with generative AI, was before that. And I agree with Bill that I also wasn’t too impressed by GPT-3.  I though that it was kind of, you know, very naturally mimicking the web, sort of parroting what was written there in a nice way. Still in a way which seemed very impressive. But it wasn’t really intelligent in any way. But shortly after GPT-3, there was a model before GPT-4 that really shocked me, and this was the first image generation model, DALL-E 1.  So that was in 2021. And I will forever remember the press release of OpenAI where they had this prompt of an avocado chair and then you had this image of the avocado chair.And what really shocked me is that clearly the model kind of “understood” what is a chair, what is an avocado, and was able to merge those concepts.  So this was really, to me, the first moment where I saw some understanding in those models.   LEE: So this was, just to get the timing right, that was before I pulled you into the tent.  BUBECK: That was before. That was like a year before.  LEE: Right.   BUBECK: And now I will tell you how, you know, we went from that moment to the meeting with the two of you and GPT-4.  So once I saw this kind of understanding, I thought, OK, fine. It understands concept, but it’s still not able to reason. It cannot—as, you know, Bill was saying—it cannot learn from your document. It cannot reason.   So I set out to try to prove that. You know, this is what I was in the business of at the time, trying to prove things in mathematics. So I was trying to prove that basically autoregressive transformers could never reason. So I was trying to prove this. And after a year of work, I had something reasonable to show. And so I had the meeting with the two of you, and I had this example where I wanted to say, there is no way that an LLM is going to be able to do x.  And then as soon as I … I don’t know if you remember, Bill. But as soon as I said that, you said, oh, but wait a second. I had, you know, the OpenAI crew at my house recently, and they showed me a new model. Why don’t we ask this new model this question?   LEE: Yeah. BUBECK: And we did, and it solved it on the spot. And that really, honestly, just changed my life. Like, you know, I had been working for a year trying to say that this was impossible. And just right there, it was shown to be possible.   LEE:One of the very first things I got interested in—because I was really thinking a lot about healthcare—was healthcare and medicine.  And I don’t know if the two of you remember, but I ended up doing a lot of tests. I ran through, you know, step one and step two of the US Medical Licensing Exam. Did a whole bunch of other things. I wrote this big report. It was, you know, I can’t remember … a couple hundred pages.   And I needed to share this with someone. I didn’t … there weren’t too many people I could share it with. So I sent, I think, a copy to you, Bill. Sent a copy to you, Seb.   I hardly slept for about a week putting that report together. And, yeah, and I kept working on it. But I was far from alone. I think everyone who was in the tent, so to speak, in those early days was going through something pretty similar. All right. So I think … of course, a lot of what I put in the report also ended up being examples that made it into the book.  But the main purpose of this conversation isn’t to reminisce aboutor indulge in those reminiscences but to talk about what’s happening in healthcare and medicine. And, you know, as I said, we wrote this book. We did it very, very quickly. Seb, you helped. Bill, you know, you provided a review and some endorsements.  But, you know, honestly, we didn’t know what we were talking about because no one had access to this thing. And so we just made a bunch of guesses. So really, the whole thing I wanted to probe with the two of you is, now with two years of experience out in the world, what, you know, what do we think is happening today?  You know, is AI actually having an impact, positive or negative, on healthcare and medicine? And what do we now think is going to happen in the next two years, five years, or 10 years? And so I realize it’s a little bit too abstract to just ask it that way. So let me just try to narrow the discussion and guide us a little bit.   Um, the kind of administrative and clerical work, paperwork, around healthcare—and we made a lot of guesses about that—that appears to be going well, but, you know, Bill, I know we’ve discussed that sometimes that you think there ought to be a lot more going on. Do you have a viewpoint on how AI is actually finding its way into reducing paperwork?  GATES: Well, I’m stunned … I don’t think there should be a patient-doctor meeting where the AI is not sitting in and both transcribing, offering to help with the paperwork, and even making suggestions, although the doctor will be the one, you know, who makes the final decision about the diagnosis and whatever prescription gets done.   It’s so helpful. You know, when that patient goes home and their, you know, son who wants to understand what happened has some questions, that AI should be available to continue that conversation. And the way you can improve that experience and streamline things and, you know, involve the people who advise you. I don’t understand why that’s not more adopted, because there you still have the human in the loop making that final decision.  But even for, like, follow-up calls to make sure the patient did things, to understand if they have concerns and knowing when to escalate back to the doctor, the benefit is incredible. And, you know, that thing is ready for prime time. That paradigm is ready for prime time, in my view.  LEE: Yeah, there are some good products, but it seems like the number one use right now—and we kind of got this from some of the previous guests in previous episodes—is the use of AI just to respond to emails from patients.Does that make sense to you?  BUBECK: Yeah. So maybe I want to second what Bill was saying but maybe take a step back first. You know, two years ago, like, the concept of clinical scribes, which is one of the things that we’re talking about right now, it would have sounded, in fact, it sounded two years ago, borderline dangerous. Because everybody was worried about hallucinations. What happened if you have this AI listening in and then it transcribes, you know, something wrong?  Now, two years later, I think it’s mostly working. And in fact, it is not yet, you know, fully adopted. You’re right. But it is in production. It is used, you know, in many, many places. So this rate of progress is astounding because it wasn’t obvious that we would be able to overcome those obstacles of hallucination. It’s not to say that hallucinations are fully solved. In the case of the closed system, they are.   Now, I think more generally what’s going on in the background is that there is something that we, that certainly I, underestimated, which is this management overhead. So I think the reason why this is not adopted everywhere is really a training and teaching aspect. People need to be taught, like, those systems, how to interact with them.  And one example that I really like, a study that recently appeared where they tried to use ChatGPT for diagnosis and they were comparing doctors without and with ChatGPT. And the amazing thing … so this was a set of cases where the accuracy of the doctors alone was around 75%. ChatGPT alone was 90%. So that’s already kind of mind blowing. But then the kicker is that doctors with ChatGPT was 80%.   Intelligence alone is not enough. It’s also how it’s presented, how you interact with it. And ChatGPT, it’s an amazing tool. Obviously, I absolutely love it. But it’s not … you don’t want a doctor to have to type in, you know, prompts and use it that way.  It should be, as Bill was saying, kind of running continuously in the background, sending you notifications. And you have to be really careful of the rate at which those notifications are being sent. Because if they are too frequent, then the doctor will learn to ignore them. So you have to … all of those things matter, in fact, at least as much as the level of intelligence of the machine.  LEE: One of the things I think about, Bill, in that scenario that you described, doctors do some thinking about the patient when they write the note. So, you know, I’m always a little uncertain whether it’s actually … you know, you wouldn’t necessarily want to fully automate this, I don’t think. Or at least there needs to be some prompt to the doctor to make sure that the doctor puts some thought into what happened in the encounter with the patient. Does that make sense to you at all?  GATES: At this stage, you know, I’d still put the onus on the doctor to write the conclusions and the summary and not delegate that.  The tradeoffs you make a little bit are somewhat dependent on the situation you’re in. If you’re in Africa, So, yes, the doctor’s still going to have to do a lot of work, but just the quality of letting the patient and the people around them interact and ask questions and have things explained, that alone is such a quality improvement. It’s mind blowing.   LEE: So since you mentioned, you know, Africa—and, of course, this touches on the mission and some of the priorities of the Gates Foundation and this idea of democratization of access to expert medical care—what’s the most interesting stuff going on right now? Are there people and organizations or technologies that are impressing you or that you’re tracking?  GATES: Yeah. So the Gates Foundation has given out a lot of grants to people in Africa doing education, agriculture but more healthcare examples than anything. And the way these things start off, they often start out either being patient-centric in a narrow situation, like, OK, I’m a pregnant woman; talk to me. Or, I have infectious disease symptoms; talk to me. Or they’re connected to a health worker where they’re helping that worker get their job done. And we have lots of pilots out, you know, in both of those cases.   The dream would be eventually to have the thing the patient consults be so broad that it’s like having a doctor available who understands the local things.   LEE: Right.   GATES: We’re not there yet. But over the next two or three years, you know, particularly given the worsening financial constraints against African health systems, where the withdrawal of money has been dramatic, you know, figuring out how to take this—what I sometimes call “free intelligence”—and build a quality health system around that, we will have to be more radical in low-income countries than any rich country is ever going to be.   LEE: Also, there’s maybe a different regulatory environment, so some of those things maybe are easier? Because right now, I think the world hasn’t figured out how to and whether to regulate, let’s say, an AI that might give a medical diagnosis or write a prescription for a medication.  BUBECK: Yeah. I think one issue with this, and it’s also slowing down the deployment of AI in healthcare more generally, is a lack of proper benchmark. Because, you know, you were mentioning the USMLE, for example. That’s a great test to test human beings and their knowledge of healthcare and medicine. But it’s not a great test to give to an AI.  It’s not asking the right questions. So finding what are the right questions to test whether an AI system is ready to give diagnosis in a constrained setting, that’s a very, very important direction, which to my surprise, is not yet accelerating at the rate that I was hoping for.  LEE: OK, so that gives me an excuse to get more now into the core AI tech because something I’ve discussed with both of you is this issue of what are the right tests. And you both know the very first test I give to any new spin of an LLM is I present a patient, the results—a mythical patient—the results of my physical exam, my mythical physical exam. Maybe some results of some initial labs. And then I present or propose a differential diagnosis. And if you’re not in medicine, a differential diagnosis you can just think of as a prioritized list of the possible diagnoses that fit with all that data. And in that proposed differential, I always intentionally make two mistakes.  I make a textbook technical error in one of the possible elements of the differential diagnosis, and I have an error of omission. And, you know, I just want to know, does the LLM understand what I’m talking about? And all the good ones out there do now. But then I want to know, can it spot the errors? And then most importantly, is it willing to tell me I’m wrong, that I’ve made a mistake?   That last piece seems really hard for AI today. And so let me ask you first, Seb, because at the time of this taping, of course, there was a new spin of GPT-4o last week that became overly sycophantic. In other words, it was actually prone in that test of mine not only to not tell me I’m wrong, but it actually praised me for the creativity of my differential.What’s up with that?  BUBECK: Yeah, I guess it’s a testament to the fact that training those models is still more of an art than a science. So it’s a difficult job. Just to be clear with the audience, we have rolled back thatversion of GPT-4o, so now we don’t have the sycophant version out there.  Yeah, no, it’s a really difficult question. It has to do … as you said, it’s very technical. It has to do with the post-training and how, like, where do you nudge the model? So, you know, there is this very classical by now technique called RLHF, where you push the model in the direction of a certain reward model. So the reward model is just telling the model, you know, what behavior is good, what behavior is bad.  But this reward model is itself an LLM, and, you know, Bill was saying at the very beginning of the conversation that we don’t really understand how those LLMs deal with concepts like, you know, where is the capital of France located? Things like that. It is the same thing for this reward model. We don’t know why it says that it prefers one output to another, and whether this is correlated with some sycophancy is, you know, something that we discovered basically just now. That if you push too hard in optimization on this reward model, you will get a sycophant model.  So it’s kind of … what I’m trying to say is we became too good at what we were doing, and we ended up, in fact, in a trap of the reward model.  LEE: I mean, you do want … it’s a difficult balance because you do want models to follow your desires and …  BUBECK: It’s a very difficult, very difficult balance.  LEE: So this brings up then the following question for me, which is the extent to which we think we’ll need to have specially trained models for things. So let me start with you, Bill. Do you have a point of view on whether we will need to, you know, quote-unquote take AI models to med school? Have them specially trained? Like, if you were going to deploy something to give medical care in underserved parts of the world, do we need to do something special to create those models?  GATES: We certainly need to teach them the African languages and the unique dialects so that the multimedia interactions are very high quality. We certainly need to teach them the disease prevalence and unique disease patterns like, you know, neglected tropical diseases and malaria. So we need to gather a set of facts that somebody trying to go for a US customer base, you know, wouldn’t necessarily have that in there.  Those two things are actually very straightforward because the additional training time is small. I’d say for the next few years, we’ll also need to do reinforcement learning about the context of being a doctor and how important certain behaviors are. Humans learn over the course of their life to some degree that, I’m in a different context and the way I behave in terms of being willing to criticize or be nice, you know, how important is it? Who’s here? What’s my relationship to them?   Right now, these machines don’t have that broad social experience. And so if you know it’s going to be used for health things, a lot of reinforcement learning of the very best humans in that context would still be valuable. Eventually, the models will, having read all the literature of the world about good doctors, bad doctors, it’ll understand as soon as you say, “I want you to be a doctor diagnosing somebody.” All of the implicit reinforcement that fits that situation, you know, will be there. LEE: Yeah. GATES: And so I hope three years from now, we don’t have to do that reinforcement learning. But today, for any medical context, you would want a lot of data to reinforce tone, willingness to say things when, you know, there might be something significant at stake.  LEE: Yeah. So, you know, something Bill said, kind of, reminds me of another thing that I think we missed, which is, the context also … and the specialization also pertains to different, I guess, what we still call “modes,” although I don’t know if the idea of multimodal is the same as it was two years ago. But, you know, what do you make of all of the hubbub around—in fact, within Microsoft Research, this is a big deal, but I think we’re far from alone—you know, medical images and vision, video, proteins and molecules, cell, you know, cellular data and so on.  BUBECK: Yeah. OK. So there is a lot to say to everything … to the last, you know, couple of minutes. Maybe on the specialization aspect, you know, I think there is, hiding behind this, a really fundamental scientific question of whether eventually we have a singular AGIthat kind of knows everything and you can just put, you know, explain your own context and it will just get it and understand everything.  That’s one vision. I have to say, I don’t particularly believe in this vision. In fact, we humans are not like that at all. I think, hopefully, we are general intelligences, yet we have to specialize a lot. And, you know, I did myself a lot of RL, reinforcement learning, on mathematics. Like, that’s what I did, you know, spent a lot of time doing that. And I didn’t improve on other aspects. You know, in fact, I probably degraded in other aspects.So it’s … I think it’s an important example to have in mind.  LEE: I think I might disagree with you on that, though, because, like, doesn’t a model have to see both good science and bad science in order to be able to gain the ability to discern between the two?  BUBECK: Yeah, no, that absolutely. I think there is value in seeing the generality, in having a very broad base. But then you, kind of, specialize on verticals. And this is where also, you know, open-weights model, which we haven’t talked about yet, are really important because they allow you to provide this broad base to everyone. And then you can specialize on top of it.  LEE: So we have about three hours of stuff to talk about, but our time is actually running low. BUBECK: Yes, yes, yes.   LEE: So I think I want … there’s a more provocative question. It’s almost a silly question, but I need to ask it of the two of you, which is, is there a future, you know, where AI replaces doctors or replaces, you know, medical specialties that we have today? So what does the world look like, say, five years from now?  GATES: Well, it’s important to distinguish healthcare discovery activity from healthcare delivery activity. We focused mostly on delivery. I think it’s very much within the realm of possibility that the AI is not only accelerating healthcare discovery but substituting for a lot of the roles of, you know, I’m an organic chemist, or I run various types of assays. I can see those, which are, you know, testable-output-type jobs but with still very high value, I can see, you know, some replacement in those areas before the doctor.   The doctor, still understanding the human condition and long-term dialogues, you know, they’ve had a lifetime of reinforcement of that, particularly when you get into areas like mental health. So I wouldn’t say in five years, either people will choose to adopt it, but it will be profound that there’ll be this nearly free intelligence that can do follow-up, that can help you, you know, make sure you went through different possibilities.  And so I’d say, yes, we’ll have doctors, but I’d say healthcare will be massively transformed in its quality and in efficiency by AI in that time period.  LEE: Is there a comparison, useful comparison, say, between doctors and, say, programmers, computer programmers, or doctors and, I don’t know, lawyers?  GATES: Programming is another one that has, kind of, a mathematical correctness to it, you know, and so the objective function that you’re trying to reinforce to, as soon as you can understand the state machines, you can have something that’s “checkable”; that’s correct. So I think programming, you know, which is weird to say, that the machine will beat us at most programming tasks before we let it take over roles that have deep empathy, you know, physical presence and social understanding in them.  LEE: Yeah. By the way, you know, I fully expect in five years that AI will produce mathematical proofs that are checkable for validity, easily checkable, because they’ll be written in a proof-checking language like Lean or something but will be so complex that no human mathematician can understand them. I expect that to happen.   I can imagine in some fields, like cellular biology, we could have the same situation in the future because the molecular pathways, the chemistry, biochemistry of human cells or living cells is as complex as any mathematics, and so it seems possible that we may be in a state where in wet lab, we see, Oh yeah, this actually works, but no one can understand why.  BUBECK: Yeah, absolutely. I mean, I think I really agree with Bill’s distinction of the discovery and the delivery, and indeed, the discovery’s when you can check things, and at the end, there is an artifact that you can verify. You know, you can run the protocol in the wet lab and seeproduced what you wanted. So I absolutely agree with that.   And in fact, you know, we don’t have to talk five years from now. I don’t know if you know, but just recently, there was a paper that was published on a scientific discovery using o3- mini. So this is really amazing. And, you know, just very quickly, just so people know, it was about this statistical physics model, the frustrated Potts model, which has to do with coloring, and basically, the case of three colors, like, more than two colors was open for a long time, and o3 was able to reduce the case of three colors to two colors.   LEE: Yeah.  BUBECK: Which is just, like, astounding. And this is not … this is now. This is happening right now. So this is something that I personally didn’t expect it would happen so quickly, and it’s due to those reasoning models.   Now, on the delivery side, I would add something more to it for the reason why doctors and, in fact, lawyers and coders will remain for a long time, and it’s because we still don’t understand how those models generalize. Like, at the end of the day, we are not able to tell you when they are confronted with a really new, novel situation, whether they will work or not.  Nobody is able to give you that guarantee. And I think until we understand this generalization better, we’re not going to be willing to just let the system in the wild without human supervision.  LEE: But don’t human doctors, human specialists … so, for example, a cardiologist sees a patient in a certain way that a nephrologist …  BUBECK: Yeah. LEE: … or an endocrinologist might not. BUBECK: That’s right. But another cardiologist will understand and, kind of, expect a certain level of generalization from their peer. And this, we just don’t have it with AI models. Now, of course, you’re exactly right. That generalization is also hard for humans. Like, if you have a human trained for one task and you put them into another task, then you don’t … you often don’t know. LEE: OK. You know, the podcast is focused on what’s happened over the last two years. But now, I’d like one provocative prediction about what you think the world of AI and medicine is going to be at some point in the future. You pick your timeframe. I don’t care if it’s two years or 20 years from now, but, you know, what do you think will be different about AI in medicine in that future than today?  BUBECK: Yeah, I think the deployment is going to accelerate soon. Like, we’re really not missing very much. There is this enormous capability overhang. Like, even if progress completely stopped, with current systems, we can do a lot more than what we’re doing right now. So I think this will … this has to be realized, you know, sooner rather than later.  And I think it’s probably dependent on these benchmarks and proper evaluation and tying this with regulation. So these are things that take time in human society and for good reason. But now we already are at two years; you know, give it another two years and it should be really …   LEE: Will AI prescribe your medicines? Write your prescriptions?  BUBECK: I think yes. I think yes.  LEE: OK. Bill?  GATES: Well, I think the next two years, we’ll have massive pilots, and so the amount of use of the AI, still in a copilot-type mode, you know, we should get millions of patient visits, you know, both in general medicine and in the mental health side, as well. And I think that’s going to build up both the data and the confidence to give the AI some additional autonomy. You know, are you going to let it talk to you at night when you’re panicked about your mental health with some ability to escalate? And, you know, I’ve gone so far as to tell politicians with national health systems that if they deploy AI appropriately, that the quality of care, the overload of the doctors, the improvement in the economics will be enough that their voters will be stunned because they just don’t expect this, and, you know, they could be reelectedjust on this one thing of fixing what is a very overloaded and economically challenged health system in these rich countries.  You know, my personal role is going to be to make sure that in the poorer countries, there isn’t some lag; in fact, in many cases, that we’ll be more aggressive because, you know, we’re comparing to having no access to doctors at all. And, you know, so I think whether it’s India or Africa, there’ll be lessons that are globally valuable because we need medical intelligence. And, you know, thank god AI is going to provide a lot of that.  LEE: Well, on that optimistic note, I think that’s a good way to end. Bill, Seb, really appreciate all of this.   I think the most fundamental prediction we made in the book is that AI would actually find its way into the practice of medicine, and I think that that at least has come true, maybe in different ways than we expected, but it’s come true, and I think it’ll only accelerate from here. So thanks again, both of you.   GATES: Yeah. Thanks, you guys.  BUBECK: Thank you, Peter. Thanks, Bill.  LEE: I just always feel such a sense of privilege to have a chance to interact and actually work with people like Bill and Sébastien.    With Bill, I’m always amazed at how practically minded he is. He’s really thinking about the nuts and bolts of what AI might be able to do for people, and his thoughts about underserved parts of the world, the idea that we might actually be able to empower people with access to expert medical knowledge, I think is both inspiring and amazing.   And then, Seb, Sébastien Bubeck, he’s just absolutely a brilliant mind. He has a really firm grip on the deep mathematics of artificial intelligence and brings that to bear in his research and development work. And where that mathematics takes him isn’t just into the nuts and bolts of algorithms but into philosophical questions about the nature of intelligence.   One of the things that Sébastien brought up was the state of evaluation of AI systems. And indeed, he was fairly critical in our conversation. But of course, the world of AI research and development is just moving so fast, and indeed, since we recorded our conversation, OpenAI, in fact, released a new evaluation metric that is directly relevant to medical applications, and that is something called HealthBench. And Microsoft Research also released a new evaluation approach or process called ADeLe.   HealthBench and ADeLe are examples of new approaches to evaluating AI models that are less about testing their knowledge and ability to pass multiple-choice exams and instead are evaluation approaches designed to assess how well AI models are able to complete tasks that actually arise every day in typical healthcare or biomedical research settings. These are examples of really important good work that speak to how well AI models work in the real world of healthcare and biomedical research and how well they can collaborate with human beings in those settings.  You know, I asked Bill and Seb to make some predictions about the future. You know, my own answer, I expect that we’re going to be able to use AI to change how we diagnose patients, change how we decide treatment options.   If you’re a doctor or a nurse and you encounter a patient, you’ll ask questions, do a physical exam, you know, call out for labs just like you do today, but then you’ll be able to engage with AI based on all of that data and just ask, you know, based on all the other people who have gone through the same experience, who have similar data, how were they diagnosed? How were they treated? What were their outcomes? And what does that mean for the patient I have right now? Some people call it the “patients like me” paradigm. And I think that’s going to become real because of AI within our lifetimes. That idea of really grounding the delivery in healthcare and medical practice through data and intelligence, I actually now don’t see any barriers to that future becoming real.   I’d like to extend another big thank you to Bill and Sébastien for their time. And to our listeners, as always, it’s a pleasure to have you along for the ride. I hope you’ll join us for our remaining conversations, as well as a second coauthor roundtable with Carey and Zak.   Until next time.   #how #reshaping #future #healthcare #medical
    WWW.MICROSOFT.COM
    How AI is reshaping the future of healthcare and medical research
    Transcript [MUSIC]      [BOOK PASSAGE]   PETER LEE: “In ‘The Little Black Bag,’ a classic science fiction story, a high-tech doctor’s kit of the future is accidentally transported back to the 1950s, into the shaky hands of a washed-up, alcoholic doctor. The ultimate medical tool, it redeems the doctor wielding it, allowing him to practice gratifyingly heroic medicine. … The tale ends badly for the doctor and his treacherous assistant, but it offered a picture of how advanced technology could transform medicine—powerful when it was written nearly 75 years ago and still so today. What would be the Al equivalent of that little black bag? At this moment when new capabilities are emerging, how do we imagine them into medicine?”   [END OF BOOK PASSAGE]     [THEME MUSIC]     This is The AI Revolution in Medicine, Revisited. I’m your host, Peter Lee.    Shortly after OpenAI’s GPT-4 was publicly released, Carey Goldberg, Dr. Zak Kohane, and I published The AI Revolution in Medicine to help educate the world of healthcare and medical research about the transformative impact this new generative AI technology could have. But because we wrote the book when GPT-4 was still a secret, we had to speculate. Now, two years later, what did we get right, and what did we get wrong?     In this series, we’ll talk to clinicians, patients, hospital administrators, and others to understand the reality of AI in the field and where we go from here.   [THEME MUSIC FADES] The book passage I read at the top is from “Chapter 10: The Big Black Bag.”  In imagining AI in medicine, Carey, Zak, and I included in our book two fictional accounts. In the first, a medical resident consults GPT-4 on her personal phone as the patient in front of her crashes. Within seconds, it offers an alternate response based on recent literature. In the second account, a 90-year-old woman with several chronic conditions is living independently and receiving near-constant medical support from an AI aide.    In our conversations with the guests we’ve spoken to so far, we’ve caught a glimpse of these predicted futures, seeing how clinicians and patients are actually using AI today and how developers are leveraging the technology in the healthcare products and services they’re creating. In fact, that first fictional account isn’t so fictional after all, as most of the doctors in the real world actually appear to be using AI at least occasionally—and sometimes much more than occasionally—to help in their daily clinical work. And as for the second fictional account, which is more of a science fiction account, it seems we are indeed on the verge of a new way of delivering and receiving healthcare, though the future is still very much open.  As we continue to examine the current state of AI in healthcare and its potential to transform the field, I’m pleased to welcome Bill Gates and Sébastien Bubeck.   Bill may be best known as the co-founder of Microsoft, having created the company with his childhood friend Paul Allen in 1975. He’s now the founder of Breakthrough Energy, which aims to advance clean energy innovation, and TerraPower, a company developing groundbreaking nuclear energy and science technologies. He also chairs the world’s largest philanthropic organization, the Gates Foundation, and focuses on solving a variety of health challenges around the globe and here at home.  Sébastien is a research lead at OpenAI. He was previously a distinguished scientist, vice president of AI, and a colleague of mine here at Microsoft, where his work included spearheading the development of the family of small language models known as Phi. While at Microsoft, he also coauthored the discussion-provoking 2023 paper “Sparks of Artificial General Intelligence,” which presented the results of early experiments with GPT-4 conducted by a small team from Microsoft Research.    [TRANSITION MUSIC]   Here’s my conversation with Bill Gates and Sébastien Bubeck.  LEE: Bill, welcome.  BILL GATES: Thank you.  LEE: Seb …  SÉBASTIEN BUBECK: Yeah. Hi, hi, Peter. Nice to be here.  LEE: You know, one of the things that I’ve been doing just to get the conversation warmed up is to talk about origin stories, and what I mean about origin stories is, you know, what was the first contact that you had with large language models or the concept of generative AI that convinced you or made you think that something really important was happening?  And so, Bill, I think I’ve heard the story about, you know, the time when the OpenAI folks—Sam Altman, Greg Brockman, and others—showed you something, but could we hear from you what those early encounters were like and what was going through your mind?   GATES: Well, I’d been visiting OpenAI soon after it was created to see things like GPT-2 and to see the little arm they had that was trying to match human manipulation and, you know, looking at their games like Dota that they were trying to get as good as human play. And honestly, I didn’t think the language model stuff they were doing, even when they got to GPT-3, would show the ability to learn, you know, in the same sense that a human reads a biology book and is able to take that knowledge and access it not only to pass a test but also to create new medicines.  And so my challenge to them was that if their LLM could get a five on the advanced placement biology test, then I would say, OK, it took biologic knowledge and encoded it in an accessible way and that I didn’t expect them to do that very quickly but it would be profound.   And it was only about six months after I challenged them to do that, that an early version of GPT-4 they brought up to a dinner at my house, and in fact, it answered most of the questions that night very well. The one it got totally wrong, we were … because it was so good, we kept thinking, Oh, we must be wrong. It turned out it was a math weakness [LAUGHTER] that, you know, we later understood that that was an area of, weirdly, of incredible weakness of those early models. But, you know, that was when I realized, OK, the age of cheap intelligence was at its beginning.  LEE: Yeah. So I guess it seems like you had something similar to me in that my first encounters, I actually harbored some skepticism. Is it fair to say you were skeptical before that?  GATES: Well, the idea that we’ve figured out how to encode and access knowledge in this very deep sense without even understanding the nature of the encoding, …  LEE: Right.   GATES: … that is a bit weird.   LEE: Yeah.  GATES: We have an algorithm that creates the computation, but even say, OK, where is the president’s birthday stored in there? Where is this fact stored in there? The fact that even now when we’re playing around, getting a little bit more sense of it, it’s opaque to us what the semantic encoding is, it’s, kind of, amazing to me. I thought the invention of knowledge storage would be an explicit way of encoding knowledge, not an implicit statistical training.  LEE: Yeah, yeah. All right. So, Seb, you know, on this same topic, you know, I got—as we say at Microsoft—I got pulled into the tent. [LAUGHS]  BUBECK: Yes.   LEE: Because this was a very secret project. And then, um, I had the opportunity to select a small number of researchers in MSR [Microsoft Research] to join and start investigating this thing seriously. And the first person I pulled in was you.  BUBECK: Yeah.  LEE: And so what were your first encounters? Because I actually don’t remember what happened then.  BUBECK: Oh, I remember it very well. [LAUGHS] My first encounter with GPT-4 was in a meeting with the two of you, actually. But my kind of first contact, the first moment where I realized that something was happening with generative AI, was before that. And I agree with Bill that I also wasn’t too impressed by GPT-3.  I though that it was kind of, you know, very naturally mimicking the web, sort of parroting what was written there in a nice way. Still in a way which seemed very impressive. But it wasn’t really intelligent in any way. But shortly after GPT-3, there was a model before GPT-4 that really shocked me, and this was the first image generation model, DALL-E 1.  So that was in 2021. And I will forever remember the press release of OpenAI where they had this prompt of an avocado chair and then you had this image of the avocado chair. [LAUGHTER] And what really shocked me is that clearly the model kind of “understood” what is a chair, what is an avocado, and was able to merge those concepts.  So this was really, to me, the first moment where I saw some understanding in those models.   LEE: So this was, just to get the timing right, that was before I pulled you into the tent.  BUBECK: That was before. That was like a year before.  LEE: Right.   BUBECK: And now I will tell you how, you know, we went from that moment to the meeting with the two of you and GPT-4.  So once I saw this kind of understanding, I thought, OK, fine. It understands concept, but it’s still not able to reason. It cannot—as, you know, Bill was saying—it cannot learn from your document. It cannot reason.   So I set out to try to prove that. You know, this is what I was in the business of at the time, trying to prove things in mathematics. So I was trying to prove that basically autoregressive transformers could never reason. So I was trying to prove this. And after a year of work, I had something reasonable to show. And so I had the meeting with the two of you, and I had this example where I wanted to say, there is no way that an LLM is going to be able to do x.  And then as soon as I … I don’t know if you remember, Bill. But as soon as I said that, you said, oh, but wait a second. I had, you know, the OpenAI crew at my house recently, and they showed me a new model. Why don’t we ask this new model this question?   LEE: Yeah. BUBECK: And we did, and it solved it on the spot. And that really, honestly, just changed my life. Like, you know, I had been working for a year trying to say that this was impossible. And just right there, it was shown to be possible.   LEE: [LAUGHS] One of the very first things I got interested in—because I was really thinking a lot about healthcare—was healthcare and medicine.  And I don’t know if the two of you remember, but I ended up doing a lot of tests. I ran through, you know, step one and step two of the US Medical Licensing Exam. Did a whole bunch of other things. I wrote this big report. It was, you know, I can’t remember … a couple hundred pages.   And I needed to share this with someone. I didn’t … there weren’t too many people I could share it with. So I sent, I think, a copy to you, Bill. Sent a copy to you, Seb.   I hardly slept for about a week putting that report together. And, yeah, and I kept working on it. But I was far from alone. I think everyone who was in the tent, so to speak, in those early days was going through something pretty similar. All right. So I think … of course, a lot of what I put in the report also ended up being examples that made it into the book.  But the main purpose of this conversation isn’t to reminisce about [LAUGHS] or indulge in those reminiscences but to talk about what’s happening in healthcare and medicine. And, you know, as I said, we wrote this book. We did it very, very quickly. Seb, you helped. Bill, you know, you provided a review and some endorsements.  But, you know, honestly, we didn’t know what we were talking about because no one had access to this thing. And so we just made a bunch of guesses. So really, the whole thing I wanted to probe with the two of you is, now with two years of experience out in the world, what, you know, what do we think is happening today?  You know, is AI actually having an impact, positive or negative, on healthcare and medicine? And what do we now think is going to happen in the next two years, five years, or 10 years? And so I realize it’s a little bit too abstract to just ask it that way. So let me just try to narrow the discussion and guide us a little bit.   Um, the kind of administrative and clerical work, paperwork, around healthcare—and we made a lot of guesses about that—that appears to be going well, but, you know, Bill, I know we’ve discussed that sometimes that you think there ought to be a lot more going on. Do you have a viewpoint on how AI is actually finding its way into reducing paperwork?  GATES: Well, I’m stunned … I don’t think there should be a patient-doctor meeting where the AI is not sitting in and both transcribing, offering to help with the paperwork, and even making suggestions, although the doctor will be the one, you know, who makes the final decision about the diagnosis and whatever prescription gets done.   It’s so helpful. You know, when that patient goes home and their, you know, son who wants to understand what happened has some questions, that AI should be available to continue that conversation. And the way you can improve that experience and streamline things and, you know, involve the people who advise you. I don’t understand why that’s not more adopted, because there you still have the human in the loop making that final decision.  But even for, like, follow-up calls to make sure the patient did things, to understand if they have concerns and knowing when to escalate back to the doctor, the benefit is incredible. And, you know, that thing is ready for prime time. That paradigm is ready for prime time, in my view.  LEE: Yeah, there are some good products, but it seems like the number one use right now—and we kind of got this from some of the previous guests in previous episodes—is the use of AI just to respond to emails from patients. [LAUGHTER] Does that make sense to you?  BUBECK: Yeah. So maybe I want to second what Bill was saying but maybe take a step back first. You know, two years ago, like, the concept of clinical scribes, which is one of the things that we’re talking about right now, it would have sounded, in fact, it sounded two years ago, borderline dangerous. Because everybody was worried about hallucinations. What happened if you have this AI listening in and then it transcribes, you know, something wrong?  Now, two years later, I think it’s mostly working. And in fact, it is not yet, you know, fully adopted. You’re right. But it is in production. It is used, you know, in many, many places. So this rate of progress is astounding because it wasn’t obvious that we would be able to overcome those obstacles of hallucination. It’s not to say that hallucinations are fully solved. In the case of the closed system, they are.   Now, I think more generally what’s going on in the background is that there is something that we, that certainly I, underestimated, which is this management overhead. So I think the reason why this is not adopted everywhere is really a training and teaching aspect. People need to be taught, like, those systems, how to interact with them.  And one example that I really like, a study that recently appeared where they tried to use ChatGPT for diagnosis and they were comparing doctors without and with ChatGPT (opens in new tab). And the amazing thing … so this was a set of cases where the accuracy of the doctors alone was around 75%. ChatGPT alone was 90%. So that’s already kind of mind blowing. But then the kicker is that doctors with ChatGPT was 80%.   Intelligence alone is not enough. It’s also how it’s presented, how you interact with it. And ChatGPT, it’s an amazing tool. Obviously, I absolutely love it. But it’s not … you don’t want a doctor to have to type in, you know, prompts and use it that way.  It should be, as Bill was saying, kind of running continuously in the background, sending you notifications. And you have to be really careful of the rate at which those notifications are being sent. Because if they are too frequent, then the doctor will learn to ignore them. So you have to … all of those things matter, in fact, at least as much as the level of intelligence of the machine.  LEE: One of the things I think about, Bill, in that scenario that you described, doctors do some thinking about the patient when they write the note. So, you know, I’m always a little uncertain whether it’s actually … you know, you wouldn’t necessarily want to fully automate this, I don’t think. Or at least there needs to be some prompt to the doctor to make sure that the doctor puts some thought into what happened in the encounter with the patient. Does that make sense to you at all?  GATES: At this stage, you know, I’d still put the onus on the doctor to write the conclusions and the summary and not delegate that.  The tradeoffs you make a little bit are somewhat dependent on the situation you’re in. If you’re in Africa, So, yes, the doctor’s still going to have to do a lot of work, but just the quality of letting the patient and the people around them interact and ask questions and have things explained, that alone is such a quality improvement. It’s mind blowing.   LEE: So since you mentioned, you know, Africa—and, of course, this touches on the mission and some of the priorities of the Gates Foundation and this idea of democratization of access to expert medical care—what’s the most interesting stuff going on right now? Are there people and organizations or technologies that are impressing you or that you’re tracking?  GATES: Yeah. So the Gates Foundation has given out a lot of grants to people in Africa doing education, agriculture but more healthcare examples than anything. And the way these things start off, they often start out either being patient-centric in a narrow situation, like, OK, I’m a pregnant woman; talk to me. Or, I have infectious disease symptoms; talk to me. Or they’re connected to a health worker where they’re helping that worker get their job done. And we have lots of pilots out, you know, in both of those cases.   The dream would be eventually to have the thing the patient consults be so broad that it’s like having a doctor available who understands the local things.   LEE: Right.   GATES: We’re not there yet. But over the next two or three years, you know, particularly given the worsening financial constraints against African health systems, where the withdrawal of money has been dramatic, you know, figuring out how to take this—what I sometimes call “free intelligence”—and build a quality health system around that, we will have to be more radical in low-income countries than any rich country is ever going to be.   LEE: Also, there’s maybe a different regulatory environment, so some of those things maybe are easier? Because right now, I think the world hasn’t figured out how to and whether to regulate, let’s say, an AI that might give a medical diagnosis or write a prescription for a medication.  BUBECK: Yeah. I think one issue with this, and it’s also slowing down the deployment of AI in healthcare more generally, is a lack of proper benchmark. Because, you know, you were mentioning the USMLE [United States Medical Licensing Examination], for example. That’s a great test to test human beings and their knowledge of healthcare and medicine. But it’s not a great test to give to an AI.  It’s not asking the right questions. So finding what are the right questions to test whether an AI system is ready to give diagnosis in a constrained setting, that’s a very, very important direction, which to my surprise, is not yet accelerating at the rate that I was hoping for.  LEE: OK, so that gives me an excuse to get more now into the core AI tech because something I’ve discussed with both of you is this issue of what are the right tests. And you both know the very first test I give to any new spin of an LLM is I present a patient, the results—a mythical patient—the results of my physical exam, my mythical physical exam. Maybe some results of some initial labs. And then I present or propose a differential diagnosis. And if you’re not in medicine, a differential diagnosis you can just think of as a prioritized list of the possible diagnoses that fit with all that data. And in that proposed differential, I always intentionally make two mistakes.  I make a textbook technical error in one of the possible elements of the differential diagnosis, and I have an error of omission. And, you know, I just want to know, does the LLM understand what I’m talking about? And all the good ones out there do now. But then I want to know, can it spot the errors? And then most importantly, is it willing to tell me I’m wrong, that I’ve made a mistake?   That last piece seems really hard for AI today. And so let me ask you first, Seb, because at the time of this taping, of course, there was a new spin of GPT-4o last week that became overly sycophantic. In other words, it was actually prone in that test of mine not only to not tell me I’m wrong, but it actually praised me for the creativity of my differential. [LAUGHTER] What’s up with that?  BUBECK: Yeah, I guess it’s a testament to the fact that training those models is still more of an art than a science. So it’s a difficult job. Just to be clear with the audience, we have rolled back that [LAUGHS] version of GPT-4o, so now we don’t have the sycophant version out there.  Yeah, no, it’s a really difficult question. It has to do … as you said, it’s very technical. It has to do with the post-training and how, like, where do you nudge the model? So, you know, there is this very classical by now technique called RLHF [reinforcement learning from human feedback], where you push the model in the direction of a certain reward model. So the reward model is just telling the model, you know, what behavior is good, what behavior is bad.  But this reward model is itself an LLM, and, you know, Bill was saying at the very beginning of the conversation that we don’t really understand how those LLMs deal with concepts like, you know, where is the capital of France located? Things like that. It is the same thing for this reward model. We don’t know why it says that it prefers one output to another, and whether this is correlated with some sycophancy is, you know, something that we discovered basically just now. That if you push too hard in optimization on this reward model, you will get a sycophant model.  So it’s kind of … what I’m trying to say is we became too good at what we were doing, and we ended up, in fact, in a trap of the reward model.  LEE: I mean, you do want … it’s a difficult balance because you do want models to follow your desires and …  BUBECK: It’s a very difficult, very difficult balance.  LEE: So this brings up then the following question for me, which is the extent to which we think we’ll need to have specially trained models for things. So let me start with you, Bill. Do you have a point of view on whether we will need to, you know, quote-unquote take AI models to med school? Have them specially trained? Like, if you were going to deploy something to give medical care in underserved parts of the world, do we need to do something special to create those models?  GATES: We certainly need to teach them the African languages and the unique dialects so that the multimedia interactions are very high quality. We certainly need to teach them the disease prevalence and unique disease patterns like, you know, neglected tropical diseases and malaria. So we need to gather a set of facts that somebody trying to go for a US customer base, you know, wouldn’t necessarily have that in there.  Those two things are actually very straightforward because the additional training time is small. I’d say for the next few years, we’ll also need to do reinforcement learning about the context of being a doctor and how important certain behaviors are. Humans learn over the course of their life to some degree that, I’m in a different context and the way I behave in terms of being willing to criticize or be nice, you know, how important is it? Who’s here? What’s my relationship to them?   Right now, these machines don’t have that broad social experience. And so if you know it’s going to be used for health things, a lot of reinforcement learning of the very best humans in that context would still be valuable. Eventually, the models will, having read all the literature of the world about good doctors, bad doctors, it’ll understand as soon as you say, “I want you to be a doctor diagnosing somebody.” All of the implicit reinforcement that fits that situation, you know, will be there. LEE: Yeah. GATES: And so I hope three years from now, we don’t have to do that reinforcement learning. But today, for any medical context, you would want a lot of data to reinforce tone, willingness to say things when, you know, there might be something significant at stake.  LEE: Yeah. So, you know, something Bill said, kind of, reminds me of another thing that I think we missed, which is, the context also … and the specialization also pertains to different, I guess, what we still call “modes,” although I don’t know if the idea of multimodal is the same as it was two years ago. But, you know, what do you make of all of the hubbub around—in fact, within Microsoft Research, this is a big deal, but I think we’re far from alone—you know, medical images and vision, video, proteins and molecules, cell, you know, cellular data and so on.  BUBECK: Yeah. OK. So there is a lot to say to everything … to the last, you know, couple of minutes. Maybe on the specialization aspect, you know, I think there is, hiding behind this, a really fundamental scientific question of whether eventually we have a singular AGI [artificial general intelligence] that kind of knows everything and you can just put, you know, explain your own context and it will just get it and understand everything.  That’s one vision. I have to say, I don’t particularly believe in this vision. In fact, we humans are not like that at all. I think, hopefully, we are general intelligences, yet we have to specialize a lot. And, you know, I did myself a lot of RL, reinforcement learning, on mathematics. Like, that’s what I did, you know, spent a lot of time doing that. And I didn’t improve on other aspects. You know, in fact, I probably degraded in other aspects. [LAUGHTER] So it’s … I think it’s an important example to have in mind.  LEE: I think I might disagree with you on that, though, because, like, doesn’t a model have to see both good science and bad science in order to be able to gain the ability to discern between the two?  BUBECK: Yeah, no, that absolutely. I think there is value in seeing the generality, in having a very broad base. But then you, kind of, specialize on verticals. And this is where also, you know, open-weights model, which we haven’t talked about yet, are really important because they allow you to provide this broad base to everyone. And then you can specialize on top of it.  LEE: So we have about three hours of stuff to talk about, but our time is actually running low. BUBECK: Yes, yes, yes.   LEE: So I think I want … there’s a more provocative question. It’s almost a silly question, but I need to ask it of the two of you, which is, is there a future, you know, where AI replaces doctors or replaces, you know, medical specialties that we have today? So what does the world look like, say, five years from now?  GATES: Well, it’s important to distinguish healthcare discovery activity from healthcare delivery activity. We focused mostly on delivery. I think it’s very much within the realm of possibility that the AI is not only accelerating healthcare discovery but substituting for a lot of the roles of, you know, I’m an organic chemist, or I run various types of assays. I can see those, which are, you know, testable-output-type jobs but with still very high value, I can see, you know, some replacement in those areas before the doctor.   The doctor, still understanding the human condition and long-term dialogues, you know, they’ve had a lifetime of reinforcement of that, particularly when you get into areas like mental health. So I wouldn’t say in five years, either people will choose to adopt it, but it will be profound that there’ll be this nearly free intelligence that can do follow-up, that can help you, you know, make sure you went through different possibilities.  And so I’d say, yes, we’ll have doctors, but I’d say healthcare will be massively transformed in its quality and in efficiency by AI in that time period.  LEE: Is there a comparison, useful comparison, say, between doctors and, say, programmers, computer programmers, or doctors and, I don’t know, lawyers?  GATES: Programming is another one that has, kind of, a mathematical correctness to it, you know, and so the objective function that you’re trying to reinforce to, as soon as you can understand the state machines, you can have something that’s “checkable”; that’s correct. So I think programming, you know, which is weird to say, that the machine will beat us at most programming tasks before we let it take over roles that have deep empathy, you know, physical presence and social understanding in them.  LEE: Yeah. By the way, you know, I fully expect in five years that AI will produce mathematical proofs that are checkable for validity, easily checkable, because they’ll be written in a proof-checking language like Lean or something but will be so complex that no human mathematician can understand them. I expect that to happen.   I can imagine in some fields, like cellular biology, we could have the same situation in the future because the molecular pathways, the chemistry, biochemistry of human cells or living cells is as complex as any mathematics, and so it seems possible that we may be in a state where in wet lab, we see, Oh yeah, this actually works, but no one can understand why.  BUBECK: Yeah, absolutely. I mean, I think I really agree with Bill’s distinction of the discovery and the delivery, and indeed, the discovery’s when you can check things, and at the end, there is an artifact that you can verify. You know, you can run the protocol in the wet lab and see [if you have] produced what you wanted. So I absolutely agree with that.   And in fact, you know, we don’t have to talk five years from now. I don’t know if you know, but just recently, there was a paper that was published on a scientific discovery using o3- mini (opens in new tab). So this is really amazing. And, you know, just very quickly, just so people know, it was about this statistical physics model, the frustrated Potts model, which has to do with coloring, and basically, the case of three colors, like, more than two colors was open for a long time, and o3 was able to reduce the case of three colors to two colors.   LEE: Yeah.  BUBECK: Which is just, like, astounding. And this is not … this is now. This is happening right now. So this is something that I personally didn’t expect it would happen so quickly, and it’s due to those reasoning models.   Now, on the delivery side, I would add something more to it for the reason why doctors and, in fact, lawyers and coders will remain for a long time, and it’s because we still don’t understand how those models generalize. Like, at the end of the day, we are not able to tell you when they are confronted with a really new, novel situation, whether they will work or not.  Nobody is able to give you that guarantee. And I think until we understand this generalization better, we’re not going to be willing to just let the system in the wild without human supervision.  LEE: But don’t human doctors, human specialists … so, for example, a cardiologist sees a patient in a certain way that a nephrologist …  BUBECK: Yeah. LEE: … or an endocrinologist might not. BUBECK: That’s right. But another cardiologist will understand and, kind of, expect a certain level of generalization from their peer. And this, we just don’t have it with AI models. Now, of course, you’re exactly right. That generalization is also hard for humans. Like, if you have a human trained for one task and you put them into another task, then you don’t … you often don’t know. LEE: OK. You know, the podcast is focused on what’s happened over the last two years. But now, I’d like one provocative prediction about what you think the world of AI and medicine is going to be at some point in the future. You pick your timeframe. I don’t care if it’s two years or 20 years from now, but, you know, what do you think will be different about AI in medicine in that future than today?  BUBECK: Yeah, I think the deployment is going to accelerate soon. Like, we’re really not missing very much. There is this enormous capability overhang. Like, even if progress completely stopped, with current systems, we can do a lot more than what we’re doing right now. So I think this will … this has to be realized, you know, sooner rather than later.  And I think it’s probably dependent on these benchmarks and proper evaluation and tying this with regulation. So these are things that take time in human society and for good reason. But now we already are at two years; you know, give it another two years and it should be really …   LEE: Will AI prescribe your medicines? Write your prescriptions?  BUBECK: I think yes. I think yes.  LEE: OK. Bill?  GATES: Well, I think the next two years, we’ll have massive pilots, and so the amount of use of the AI, still in a copilot-type mode, you know, we should get millions of patient visits, you know, both in general medicine and in the mental health side, as well. And I think that’s going to build up both the data and the confidence to give the AI some additional autonomy. You know, are you going to let it talk to you at night when you’re panicked about your mental health with some ability to escalate? And, you know, I’ve gone so far as to tell politicians with national health systems that if they deploy AI appropriately, that the quality of care, the overload of the doctors, the improvement in the economics will be enough that their voters will be stunned because they just don’t expect this, and, you know, they could be reelected [LAUGHTER] just on this one thing of fixing what is a very overloaded and economically challenged health system in these rich countries.  You know, my personal role is going to be to make sure that in the poorer countries, there isn’t some lag; in fact, in many cases, that we’ll be more aggressive because, you know, we’re comparing to having no access to doctors at all. And, you know, so I think whether it’s India or Africa, there’ll be lessons that are globally valuable because we need medical intelligence. And, you know, thank god AI is going to provide a lot of that.  LEE: Well, on that optimistic note, I think that’s a good way to end. Bill, Seb, really appreciate all of this.   I think the most fundamental prediction we made in the book is that AI would actually find its way into the practice of medicine, and I think that that at least has come true, maybe in different ways than we expected, but it’s come true, and I think it’ll only accelerate from here. So thanks again, both of you.  [TRANSITION MUSIC]  GATES: Yeah. Thanks, you guys.  BUBECK: Thank you, Peter. Thanks, Bill.  LEE: I just always feel such a sense of privilege to have a chance to interact and actually work with people like Bill and Sébastien.    With Bill, I’m always amazed at how practically minded he is. He’s really thinking about the nuts and bolts of what AI might be able to do for people, and his thoughts about underserved parts of the world, the idea that we might actually be able to empower people with access to expert medical knowledge, I think is both inspiring and amazing.   And then, Seb, Sébastien Bubeck, he’s just absolutely a brilliant mind. He has a really firm grip on the deep mathematics of artificial intelligence and brings that to bear in his research and development work. And where that mathematics takes him isn’t just into the nuts and bolts of algorithms but into philosophical questions about the nature of intelligence.   One of the things that Sébastien brought up was the state of evaluation of AI systems. And indeed, he was fairly critical in our conversation. But of course, the world of AI research and development is just moving so fast, and indeed, since we recorded our conversation, OpenAI, in fact, released a new evaluation metric that is directly relevant to medical applications, and that is something called HealthBench. And Microsoft Research also released a new evaluation approach or process called ADeLe.   HealthBench and ADeLe are examples of new approaches to evaluating AI models that are less about testing their knowledge and ability to pass multiple-choice exams and instead are evaluation approaches designed to assess how well AI models are able to complete tasks that actually arise every day in typical healthcare or biomedical research settings. These are examples of really important good work that speak to how well AI models work in the real world of healthcare and biomedical research and how well they can collaborate with human beings in those settings.  You know, I asked Bill and Seb to make some predictions about the future. You know, my own answer, I expect that we’re going to be able to use AI to change how we diagnose patients, change how we decide treatment options.   If you’re a doctor or a nurse and you encounter a patient, you’ll ask questions, do a physical exam, you know, call out for labs just like you do today, but then you’ll be able to engage with AI based on all of that data and just ask, you know, based on all the other people who have gone through the same experience, who have similar data, how were they diagnosed? How were they treated? What were their outcomes? And what does that mean for the patient I have right now? Some people call it the “patients like me” paradigm. And I think that’s going to become real because of AI within our lifetimes. That idea of really grounding the delivery in healthcare and medical practice through data and intelligence, I actually now don’t see any barriers to that future becoming real.  [THEME MUSIC]  I’d like to extend another big thank you to Bill and Sébastien for their time. And to our listeners, as always, it’s a pleasure to have you along for the ride. I hope you’ll join us for our remaining conversations, as well as a second coauthor roundtable with Carey and Zak.   Until next time.   [MUSIC FADES]
    0 Commentarios 0 Acciones
Resultados de la búsqueda