The Chinese game industry has grown immensely in the past few years. Online free-to-play titles, such as Genshin Impact and Wuthering Waves, have millions of players around the world, while others like Etheria: Restart gear up for their full release. Companies have also marveled at the success of Black Myth: Wukong, which sold 10 million copies in under a week when it launched in August 2024, and subsequently honored as Game of the Year at the 2024 Steam Awards.Indeed, game studios in China are firing on all cylinders, with multiple offerings vying to become the next big global hit. This is further bolstered by rising incomes among consumers, high enthusiasm among developers, and robust support from regional governments. This raises the questions—what's driving growth in the region, and where is this emerging industry headed next?To learn more about the inner workings of the Chinese games industry, Game Developer spoke with several experts like China Independent Game Alliance (CIGA) founder Simon Zhu, Li Shen, former Tencent Games China and Epic Games China chief technology officer, and Siyuan Xia, the co-founder of Chengdu-based Leenzee Technology and director of soulslike action-roleplaying game Wuchang: Fallen Feathers.Single-player game development is becoming sustainable in ChinaRelated:Chinese developers have historically been reluctant to work on single-player games. The challenges of premium single-player production are immense, and the popularity of free-to-play titles makes those kinds of games a better business prospect. Zhu gave us a brief history lesson on how free-to-play offerings became the norm for many decades. "If you recall, 1995 was a landmark year that saw the birth of The Legend of Sword and Fairy series," notes Zhu. The franchise, which combined elements of Chinese mythology and wuxia (Chinese martial arts), was a huge hit for nearly a decade."Unfortunately, even with the series' popularity, piracy was so rampant back then that the single-player market could not truly flourish. It wasn't until the online game boom in the 2000s that China's video games market and history truly began—which was then dominated by free-to-play games. We had to wait until March 2015 when the PlayStation officially entered the Chinese market for single-player console games to be formally introduced to gamers. Even then, single-player games still occupied a small niche until Steam's emergence significantly expanded access."Studios and publishers also have to contend with the disparity in gauging the success of a single-player title compared to a free-to-play offering. "The fate of the former is often determined on launch day—(making those projects riskier ventures)—while the latter has more room for recovery if issues arise on the first day," Xia tells us. It all boils down to how each team adapts, just as how the Wuchang team had adjusted to a new development cycle to make the overall rhythm and work process more manageable.Related:Image via Lenzee/505 Games.Experts are indeed well aware of the opportunities brought about by breakthrough titles opening doors to a wider audience, though these are uncharted waters for many studios in the country.Shen (who now heads his own investment company, Re³ Lab) also adds that Chinese players and developers have always had interest in single-player games, but it wasn't until Black Myth: Wukong sold millions of units globally that publishers became more willing to invest.Other single-player offerings, such as S-Game's Phantom Blade: Zero and Leenzee's Wuchang: Fallen Feathers, also hope to capitalize on Black Myth: Wukong's success. Wuchang, while set during the fall of the Ming Dynasty, even shows a glimpse into the culture and history of the Bronze Age Ancient Kingdom of Shu given that the studio is just an hour away from the archaeological site of Sanxingdui.Related:Wuchang and other games wear their cultural identity proudly on their sleeves, and Xia emphasizes its importance. "I believe Chinese games are gradually forming a distinctive identity in global markets. Cultural outreach isn't an overnight phenomenon but a long-term evolution fueled by collective efforts. For our part, we aspire to contribute foundational work through our endeavors.Shen echoes this sentiment as well. "Black Myth: Wukong was hailed as 'China's first AAA breakthrough' and I believe this instilled pride not just in the developers, but across the industry as well. Many view their work as challenging the Western-dominated games landscape and elevating Asian cultural representation. This collective mission drives innovation and a deeper exploration of local/regional narratives."That said, there's also a balancing act between highlighting cultural uniqueness and navigating geopolitical sensitivities, as Shen points out. "Look at Phantom Blade: Zero, for instance," says Shen. "It garnered nine million views on Bilibili (essentially China's own version of YouTube), but it required rebranding for Western audiences."Challenges and opportunities in a fluctuating marketThe experts we spoke with agreed that Chinese game developers benefit from unique comparative advantages—and struggle with some disadvantages. "Chinese companies, owing to their experience in making online free-to-play games, are global frontrunners in commercial design and operations, excelling at creating monetization models, balancing numerical values, and guiding user behavior," says Zhu. "Sadly, game design and overall quality may fall short, particularly in narrative and script writing. The industry still suffers from limited experience in game design, insufficient market insight, and inadequate industry exchange."Shen agrees that China has achieved world-class capabilities, particularly when it comes to mobile game platforms and substantially profitable products in the domestic market. "Unfortunately, increasing production costs and market competition raise the barrier to entry for commercial games," says Shen. "This has made some developers grow increasingly conservative in genre selection, prioritizing production polish over innovative differentiation."Zhu also shares the stark differences between Western and Chinese industries, particularly when holding events and expos. "To give you an example: The Game Developers Conference (GDC) in the United States has long been established as a key platform for industry exchanges and B2B meetings. The event hosts approximately 1,000 sessions, with an access pass that costs around 10,000 RMB (approximately $1,350), yet it manages to draw attendees from across the globe, fostering an excellent environment for communication and networking.""In contrast, the events I organize in China offer tickets at about 500 RMB (approximately 70 USD), which already includes a one-day access to the WePlay Expo. Despite this, we still face challenges in ticket sales and in engaging local speakers, some of whom are hesitant to share their expertise. I believe that this is due to the general public's understanding of the games industry and culture, especially since China's video game history and cultural integration only span around 20 years. The disparity with overseas markets is notable, but it also indicates substantial growth potential."Crunch culture, too, remains a controversial topic in China. "Long hours are common especially during project deadlines," says Shen. "Some studios adopt a 9-9-6 schedule–i.e. 9 a.m. to 9 p.m., six days a week. While labor reforms are underway, intense competition and a culture of 'dedication' [to your work] tend to normalize overtime as well."Image via CIGA.While government guidelines do exist, enforcement remains inconsistent as it depends on each company or organization. "Our company, Leenzee, does not encourage overtime since we view that as inefficient and counterproductive," notes Xia. "In spite of this, some team members still voluntarily stay late. For them, Wuchang is the first Chinese-developed single-player game that they've worked on, and there's an emotional attachment to it.""In CIGA, we have consistently eschewed the practice of crunch or overtime, even going so far as to avoid rigid clock-in/clock-out times," explains Zhu. "The reality is that crunch, feeling compelled to stay until the boss leaves, or project pressures that lead to mandated overtime—these are all outdated formalities. They do not enhance inefficiency or quality. Instead, they only lead to increased stress and dissatisfaction among team members. Knowing how much the younger generation values work-life balance, I'm optimistic that outdated practices wane as the attitudes of newer generations continue to shift and the work environment evolves."In a period of growth for Chinese game development, one might think there were fewer layoffs going around. That's not true, says Zhu. "I'm based in Shanghai, the city that's at the forefront of the industry," he says. "In the last two years, numerous companies have experienced layoffs, including the termination of entire projects, teams, or subsidiaries. The approach differs from company to company, with some offering compensation to affected employees."Shen points out that unlike many other game development communities, China benefits from a strong social safety net reinforced by unemployment insurance and state-backed retraining programs. "Though specifics depend on regional policies," he clarified, noting it's not a perfect solution.Local governments are investing in game developmentRegional policies that often play a huge role in supporting game development in China. Most games that are set to be published in the country require a license from the National Press and Publication Administration (NPPA), with key steps that include content review (i.e. cultural sensitivity or historical accuracy) and technical compliance. However, each regional administrative district may have additional programs in place to aid companies and studios."The Sichuan Provincial Government has been highly supportive of our work," says Xia. "Since Wuchang: Fallen Feathers takes place in Chengdu and parts of Sichuan Province, the government even helped us locate more cultural heritage locations available for digital preservation through our scanning process.""Depending on the regional policies, the initiatives might include tax breaks for cultural projects or grants for technology/R&D," adds Shen. "Also, games recognized as key to 'digital cultural industries' have mandates to integrate traditional themes, and there might even be state-backed campaigns to export games globally as cultural ambassadors."Zhu, meanwhile, gestured to the 'tech park' we were in, one of several in the city that was built by Shanghai's administrative government. "This entire complex is home to several tech companies, both large conglomerates and startups, and most only pay a small expense for rent," says Zhu. Zhu also adds that local governments across the country have policies that help address game incubation, operations, local releases, and international distribution.There's no denying that China is the largest video games market in the world. While challenges, such as a lack of experience in making single-player games and a reluctance to risk on day-one launch revenue, do exist, experts remain hopeful. Developers, project leads, experts, and the government all contribute toward a collective effort to drive the expansion and global reach of the Chinese games industry."Rising incomes and increased demand for high-quality content will continue to drive growth," says Shen, arguing that there's a brighter future ahead.Zhu, meanwhile, remains steadfast in CIGA's support of the indie games community. "We know that the development of this sector in the country remains a long-term endeavor," notes Zhu. "Transformative change may not be achieved within a single lifetime, but this undertaking demands continued dedication and resilience.""We have a saying in China: 'When everybody adds fuel, the flames rise high'—it's the equivalent of saying that 'there is strength in numbers,'" posits Xia. "Our team hopes that, through our efforts, we can contribute further to the games industry, making it easier for future creators and allowing for more interesting works to emerge."

Published May 20, 2025 10:00am EDT close Google's AI unleashes new powerful scam-busting features for Android The CyberGuy explains steps you can take to protect yourself from scams. Scammers are always coming up with new tricks. Just when you start feeling confident about spotting phishing emails, suspicious links and fake banking apps, they find a new angle. Lately, they have been getting more creative, turning to the built-in features of our phones to pull off their schemes. One of the latest targets is NFC, the technology behind tap-to-pay. It might seem harmless, but a new scam is using it in ways most people would never expect. An Android malware called SuperCard goes beyond just stealing your card details. It gives attackers the ability to use your card remotely for real transactions. And the worst part is that it all begins with something as simple as a text message. An Android phone (Kurt "CyberGuy" Knutsson)What makes SuperCard X different?SuperCard X stands out from other Android malware because of how it operates. As reported by researchers at Cleafy, instead of stealing usernames, passwords or verification codes, it uses a method called NFC relay. This allows attackers to copy card data from a victim's device in real time and use it elsewhere to make payments or withdraw cash. The process does not require physical access to the card or knowledge of the PIN.The malware is offered through a Malware-as-a-Service model, which means different cybercriminals can use it in their own regions. This makes the threat more scalable and harder to contain. Unlike most banking trojans, SuperCard X is not focused on one specific institution. It targets any cardholder regardless of which bank issued their card.Another key difference is how stealthy the malware is. It uses minimal permissions and does not include extra features that would make it easier to detect. This lean approach helps it avoid detection by antivirus software and allows it to operate quietly on infected devices. SuperCardX fully undetectable (FUD | Cleafy) (Kurt "CyberGuy" Knutsson)How the scam worksThe fraud begins with a message sent through SMS or WhatsApp. It pretends to be from a bank and warns the recipient about a suspicious transaction. The message includes a phone number and urges the person to call to resolve the issue. This is the first step in gaining the victim's trust.Once on the phone, the attacker poses as a bank representative and walks the victim through a fake security process. This may include asking them to confirm personal details or adjust settings in their mobile banking app, such as removing spending limits on their card.Next, the attacker asks the victim to install a mobile app that is described as a tool to verify the account or enhance security. In reality, this app contains the SuperCard X malware. After the installation, the attacker instructs the victim to tap their card against the phone. The malware then captures the NFC data from the card and sends it to a second phone controlled by the attacker.Using the copied data, the attacker can make contactless payments or make ATM withdrawals almost instantly. This method allows them to steal funds quickly and leaves little opportunity for banks or victims to intervene in time. A woman scrolling on her phone (Kurt "CyberGuy" Knutsson)8 ways you can stay safe from SuperCard X malware1) Be cautious of suspicious texts and calls. Use strong antivirus software: Fraudulent campaigns often begin with an SMS or call that seems to come from your bank. These messages usually claim there's suspicious activity on your account and urge you to click a link or dial a number to resolve the issue. However, this is a tactic used to gain access to your personal information. Always approach such messages with skepticism.The best way to safeguard yourself from malicious links that install malware, potentially accessing your private information, is to have strong antivirus software installed on all your devices. This protection can also alert you to phishing emails and ransomware scams, keeping your personal information and digital assets safe. Get my picks for the best 2025 antivirus protection winners for your Windows, Mac, Android and iOS devices.2) Avoid installing apps from untrusted sources: One of the key ways malware like SuperCard X spreads is through deceptive apps that victims are persuaded to install. These apps often look harmless, posing as tools for security or account verification. If you receive a link to download an app via SMS, email or messaging apps like WhatsApp, do not click on it. Instead, only download apps from trusted sources, such as the Google Play Store. Additionally, carefully review app permissions and avoid granting unnecessary access, particularly to sensitive data like NFC, location or personal contacts.3) Turn off NFC when not in use: NFC, or Near Field Communication, is a useful feature that allows contactless payments and data exchanges. However, it can be exploited by attackers to capture your card information without you even realizing it. To minimize your risk of falling victim to NFC-based malware like SuperCard X, turn off NFC when you're not actively using it. On most Android devices, you can do this by going to "Settings," then "Connected Devices" or "Connection Preferences," where you’ll find the NFC toggle. By disabling NFC, your phone won't transmit data wirelessly, which helps protect your payment card information from being stolen by nearby attackers.4) Keep a close eye on your bank accounts and cards: If your device has come into contact with the SuperCard or anything similar, it’s possible your banking details are already compromised. That’s why it’s important to regularly check your transaction history for anything odd, like a small payment you don’t remember making or a charge from a strange location could be a sign of misuse. If you spot anything suspicious, report it to your bank right away. It’s also worth checking your credit reports every now and then to catch signs of identity theft before they snowball into bigger issues.5) Use a personal data removal service: If scammers have targeted you once, there's a higher chance they’ll try again, especially if your personal details (like your phone number, address or email) are easily found online. Data removal services scan people-search sites and brokers, then request the removal of your info. This reduces your exposure and helps prevent future phishing or social engineering attacks.While no service can guarantee the complete removal of your data from the internet, a data removal service is really a smart choice. They aren’t cheap and neither is your privacy. These services do all the work for you by actively monitoring and systematically erasing your personal information from hundreds of websites. It’s what gives me peace of mind and has proven to be the most effective way to erase your personal data from the internet. By limiting the information available, you reduce the risk of scammers cross-referencing data from breaches with information they might find on the dark web, making it harder for them to target you. Check out my top picks for data removal services here.6) Contact your bank and freeze your cards: If you think you’ve tapped or handled a suspicious card, or if your phone acted strangely afterward, don’t brush it off. Call your bank and let them know what happened. They can freeze your card to stop any unauthorized payments and issue a new one for added safety. You should also ask them to monitor your account more closely for a while. On top of that, place a fraud alert with a credit bureau so no one can easily open a new line of credit in your name.7) Consider enrolling in identity theft protection services: If you've been targeted by a sophisticated scam like SuperCard X, there's a chance your personal information, not just your card data, may be at risk. Identity theft companies can monitor personal information, like your Social Security number, phone number and email address, and alert you if it is being sold on the dark web or being used to open an account. They can also assist you in freezing your bank and credit card accounts to prevent further unauthorized use by criminals. See my tips and best picks on how to protect yourself from identity theft.8) Report the scam to your national cybercrime authority: Whether or not you lost money, reporting the scam helps authorities track emerging threats and warn others. You can report such fraud to the FBI’s Internet Crime Complaint Center or the Federal Trade Commission. Your report could help catch the people behind the scam or at least shut down their infrastructure.Kurt’s key takeawayThe SuperCard X malware campaign represents a significant shift in how cybercriminals are targeting individuals and financial institutions. By exploiting NFC technology and combining it with social engineering tactics, attackers have found a way to bypass traditional fraud detection systems. What's especially concerning is how quickly these attacks unfold, making them harder to detect before the damage is done. As this threat evolves, it's important for both consumers and institutions to recognize the potential risks of these multilayered fraud strategies.Do you think Google is doing enough to protect you from malware? Let us know by writing us atCyberguy.com/Contact.For more of my tech tips and security alerts, subscribe to my free CyberGuy Report Newsletter by heading to Cyberguy.com/Newsletter.Follow Kurt on his social channels:Answers to the most-asked CyberGuy questions:New from Kurt:Copyright 2025 CyberGuy.com. All rights reserved. Kurt "CyberGuy" Knutsson is an award-winning tech journalist who has a deep love of technology, gear and gadgets that make life better with his contributions for Fox News & FOX Business beginning mornings on "FOX & Friends." Got a tech question? Get Kurt’s free CyberGuy Newsletter, share your voice, a story idea or comment at CyberGuy.com.

When we launched Badberries a year ago, I thought I understood what challenges we’d face. Finding clients. Creating standout work. Building our reputation. Yes, these were true. But what I’ve come to realise is that the heart of all these challenges revolves around a single question: where do we invest our time? The invisible client that never sleeps Every agency has two distinct sides that constantly compete for attention. The first is client delivery – the work that pays bills and builds reputation. The second is the business itself – operations, growth, and longevity. I’ve started thinking of this second side as our “invisible client.” Like any client, it has deadlines, demands, and deliverables. But unlike external clients, this one never stops needing attention, never wraps a project, and never takes a holiday. The challenge? If you neglect client work, you have no income. If you neglect the invisible client, you have no future. Complementary founders as strategic advantage Recognising this tension early was crucial to our survival. As managing director with a background in complex agency projects, I bring planning and operational expertise. Meanwhile Dani Wolf, our creative director and my co-founder, ensures our output remains distinctive and compelling. Together, we lead a team of designers who execute our vision for clients. This complementary partnership has been fundamental to our success. When we first discussed starting the studio, Dani appreciated my approach, recognising how proper structure could give the creative work room to flourish. He saw how systems would help us be more efficient, meet deadlines consistently, improve collaboration and help the creative process flow better. This balance allows us to serve both our visible and invisible clients more effectively than either of us could alone. When urgent client needs arise, we can divide and conquer, one of us focusing on delivery while the other keeps our business moving forward. Time as investment currency In a service business where we essentially buy time – from ourselves and our team – and sell it to clients, we’ve learnt to think of time allocation as investment decisions. This may sound very dry, but it’s necessity to recognise it. In the creative industry, there are typically two pricing approaches: value-based and time-based. Value-based pricing operates on the principle that the worth of the work transcends the time invested in creating it. There’s a famous anecdote about Paula Scher, the legendary Pentagram designer, who reportedly designed the Citibank logo during a meeting on a napkin. As she famously said, “It took me a few seconds to draw it, but it took me 34 years to learn how to draw it in a few seconds.” “The main question isn’t how to manage time, it’s how to invest it wisely.” As a new business, we made the conscious decision to adopt time-based pricing. Not because we don’t value our work, but because it creates transparency with clients while protecting us from scope creep. It’s given us the clarity and flexibility needed in our first year, allowing us to be agile while maintaining healthy margins. More importantly, it’s forced us to be intentional about where we invest non-billable hours. When you track time meticulously, you quickly see which activities generate returns and which don’t. The art of delegation One of our biggest time management lessons has been understanding what to outsource and what to handle ourselves. The equation isn’t simply about cost, it’s about strategic focus. We quickly outsourced accounting and legal services, areas where mistakes would be costly and specialised expertise is essential. This decision, while an investment, freed mental space and reduced stress considerably. For marketing, our approach has evolved. We initially outsourced social media management but brought it back in-house when we realised we were still finding our voice. Conversely, hiring a PR specialist to build brand awareness has been transformative precisely because it’s a specialised skill neither of us possesses. The question isn’t what can we afford to outsource? It’s what can we afford not to outsource if we want to grow? Beyond billable hours Six months into our journey, we faced an uncomfortable truth – while we were steadily winning client work, these projects had limited portfolio potential. This realisation sparked a multi-pronged approach. First, we became more intentional about which client projects we pursued, weighing creative potential alongside financial return. Sometimes a smaller budget project with greater creative freedom became a better long-term investment than a larger but creatively constrained opportunity. Second, we carved out specific time for “studio projects” – self-initiated work that might not generate immediate revenue but would demonstrate our capabilities, showcase our passions as a business, and attract the type of clients we ultimately wanted. Fluid time allocation What we’ve ultimately learnt is that time allocation isn’t about rigid formulas, it’s about maintaining a fluid, responsive approach while never losing sight of both sides of the business. Our approach continues to mature. Some weeks demand intense focus on client delivery; others allow more investment in the invisible client. The key isn’t perfect balance in any given week, but ensuring that neither side gets neglected over the longer term. This fluidity extends to how we work as founders and how we blend our work and personal lives. Dani, a morning person, starts early when he’s most energised. As a night owl, I often put in hours after my son’s bedtime. We’ve even started taking a couple of weekend hours to plan the week ahead, as we found weekdays often began with immediate client needs that left no space for strategic thinking. I’m not advocating for poor boundaries. But as founders, finding the rhythm that works for your life circumstances is essential. Another lesson our invisible client has taught us. Looking forward A year in, we’re still refining this balance. But we’ve learned that the main question isn’t how to manage time, it’s how to invest it wisely across both sides of our business. For those starting their own creative venture, recognise early that your invisible client demands as much strategic thinking as any external project. Build systems that acknowledge both sides need nurturing, be intentional about outsourcing, and remember that sometimes the most valuable hours you spend won’t appear on any client invoice. Natasha Szczerb is managing director and co-founder of design and branding studio Badberries. Industries in this article Recent Jobs

GitHub adds agentic capabilities to its Copilot coding assistant, competing with other more asynchronous coding platforms.Read More

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More As enterprises increasingly look to build and deploy generative AI-powered applications and services for internal or external use (employees or customers), one of the toughest questions they face is understanding exactly how well these AI tools are performing out in the wild. In fact, a recent survey by consulting firm McKinsey and Company found that only 27% of 830 respondents said that their enterprises’ reviewed all of the outputs of their generative AI systems before they went out to users. Unless a user actually writes in with a complaint report, how is a company to know if its AI product is behaving as expected and planned? Raindrop, formerly known as Dawn AI, is a new startup tackling the challenge head-on, positioning itself as the first observability platform purpose-built for AI in production, catching errors as they happen and explaining to enterprises what went wrong and why. The goal? Help solve generative AI’s so-called “black box problem.” “AI products fail constantly—in ways both hilarious and terrifying,” wrote co-founder Ben Hylak on X recently, “Regular software throws exceptions. But AI products fail silently.” Raindrop seeks to offer any category-defining tool akin to what observability company Sentry does for traditional software. But while traditional exception tracking tools don’t capture the nuanced misbehaviors of large language models or AI companions, Raindrop attempts to fill the hole. “In traditional software, you have tools like Sentry and Datadog to tell you what’s going wrong in production,” he told VentureBeat in a video call interview last week. “With AI, there was nothing.” Until now — of course. How Raindrop works Raindrop offers a suite of tools that allow teams at enterprises large and small to detect, analyze, and respond to AI issues in real time. The platform sits at the intersection of user interactions and model outputs, analyzing patterns across hundreds of millions of daily events, but doing so with SOC-2 encryption enabled, protecting the data and privacy of users and the company offering the AI solution. “Raindrop sits where the user is,” Hylak explained. “We analyze their messages, plus signals like thumbs up/down, build errors, or whether they deployed the output, to infer what’s actually going wrong.” Raindrop uses a machine learning pipeline that combines LLM-powered summarization with smaller bespoke classifiers optimized for scale. Promotional screenshot of Raindrop’s dashboard. Credit: Raindrop.ai “Our ML pipeline is one of the most complex I’ve seen,” Hylak said. “We use large LLMs for early processing, then train small, efficient models to run at scale on hundreds of millions of events daily.” Customers can track indicators like user frustration, task failures, refusals, and memory lapses. Raindrop uses feedback signals such as thumbs down, user corrections, or follow-up behavior (like failed deployments) to identify issues. Fellow Raindrop co-founder and CEO Zubin Singh Koticha told VentureBeat in the same interview that while many enterprises relied on evaluations, benchmarks, and unit tests for checking the reliability of their AI solutions, there was very little designed to check AI outputs during production. “Imagine in traditional coding if you’re like, ‘Oh, my software passes ten unit tests. It’s great. It’s a robust piece of software.’ That’s obviously not how it works,” Koticha said. “It’s a similar problem we’re trying to solve here, where in production, there isn’t actually a lot that tells you: is it working extremely well? Is it broken or not? And that’s where we fit in.” For enterprises in highly regulated industries or for those seeking additional levels of privacy and control, Raindrop offers Notify, a fully on-premises, privacy-first version of the platform aimed at enterprises with strict data handling requirements. Unlike traditional LLM logging tools, Notify performs redaction both client-side via SDKs and server-side with semantic tools. It stores no persistent data and keeps all processing within the customer’s infrastructure. Raindrop Notify provides daily usage summaries and surfacing of high-signal issues directly within workplace tools like Slack and Teams—without the need for cloud logging or complex DevOps setups. Advanced error identification and precision Identifying errors, especially with AI models, is far from straightforward. “What’s hard in this space is that every AI application is different,” said Hylak. “One customer might build a spreadsheet tool, another an alien companion. What ‘broken’ looks like varies wildly between them.” That variability is why Raindrop’s system adapts to each product individually. Each AI product Raindrop monitors is treated as unique. The platform learns the shape of the data and behavior norms for each deployment, then builds a dynamic issue ontology that evolves over time. “Raindrop learns the data patterns of each product,” Hylak explained. “It starts with a high-level ontology of common AI issues—things like laziness, memory lapses, or user frustration—and then adapts those to each app.” Whether it’s a coding assistant that forgets a variable, an AI alien companion that suddenly refers to itself as a human from the U.S., or even a chatbot that starts randomly bringing up claims of “white genocide” in South Africa, Raindrop aims to surface these issues with actionable context. The notifications are designed to be lightweight and timely. Teams receive Slack or Microsoft Teams alerts when something unusual is detected, complete with suggestions on how to reproduce the problem. Over time, this allows AI developers to fix bugs, refine prompts, or even identify systemic flaws in how their applications respond to users. “We classify millions of messages a day to find issues like broken uploads or user complaints,” said Hylak. “It’s all about surfacing patterns strong and specific enough to warrant a notification.” From Sidekick to Raindrop The company’s origin story is rooted in hands-on experience. Hylak, who previously worked as a human interface designer at visionOS at Apple and avionics software engineering at SpaceX, began exploring AI after encountering GPT-3 in its early days back in 2020. “As soon as I used GPT-3—just a simple text completion—it blew my mind,” he recalled. “I instantly thought, ‘This is going to change how people interact with technology.’” Alongside fellow co-founders Koticha and Alexis Gauba, Hylak initially built Sidekick, a VS Code extension with hundreds of paying users. But building Sidekick revealed a deeper problem: debugging AI products in production was nearly impossible with the tools available. “We started by building AI products, not infrastructure,” Hylak explained. “But pretty quickly, we saw that to grow anything serious, we needed tooling to understand AI behavior—and that tooling didn’t exist.” What started as an annoyance quickly evolved into the core focus. The team pivoted, building out tools to make sense of AI product behavior in real-world settings. In the process, they discovered they weren’t alone. Many AI-native companies lacked visibility into what their users were actually experiencing and why things were breaking. With that, Raindrop was born. Raindrop’s pricing, differentiation and flexibility have attracted a wide range of initial customers Raindrop’s pricing is designed to accommodate teams of various sizes. A Starter plan is available at $65/month, with metered usage pricing. The Pro tier, which includes custom topic tracking, semantic search, and on-prem features, starts at $350/month and requires direct engagement. While observability tools are not new, most existing options were built before the rise of generative AI. Raindrop sets itself apart by being AI-native from the ground up. “Raindrop is AI-native,” Hylak said. “Most observability tools were built for traditional software. They weren’t designed to handle the unpredictability and nuance of LLM behavior in the wild.” This specificity has attracted a growing set of customers, including teams at Clay.com, Tolen, and New Computer. Raindrop’s customers span a wide range of AI verticals—from code generation tools to immersive AI storytelling companions—each requiring different lenses on what “misbehavior” looks like. Born from necessity Raindrop’s rise illustrates how the tools for building AI need to evolve alongside the models themselves. As companies ship more AI-powered features, observability becomes essential—not just to measure performance, but to detect hidden failures before users escalate them. In Hylak’s words, Raindrop is doing for AI what Sentry did for web apps—except the stakes now include hallucinations, refusals, and misaligned intent. With its rebrand and product expansion, Raindrop is betting that the next generation of software observability will be AI-first by design. Daily insights on business use cases with VB Daily If you want to impress your boss, VB Daily has you covered. We give you the inside scoop on what companies are doing with generative AI, from regulatory shifts to practical deployments, so you can share insights for maximum ROI. Read our Privacy Policy Thanks for subscribing. Check out more VB newsletters here. An error occured.

The structural reality behind disappearing entry-level design roles and some practical advice for finding ways inMade by me with Midjourney.I haven’t worked with a junior designer in nearly a decade.Not because I don’t value them, but because the environments I’ve worked in haven’t been set up to support them.Lean teams leave no room for juniorsFor the past eight years, I’ve worked primarily at hyper-scaling, technical, enterprise startups, especially in cybersecurity. These are companies that have found traction and are rapidly scaling to meet aggressive benchmarks in competitive markets. The pressure to deliver quickly and accurately is relentless.In these engineering-driven organizations building products for technical users, design teams typically remain lean: often just 2–3 people. Every designer is expected to lead a pod, own a problem space, and deliver independently from day one.There’s simply no extra bandwidth for mentorship. No built-in scaffolding to help junior designers safely learn and grow. No slack in the system for someone still finding their footing.Even with the best intentions, these teams just aren’t structurally built to support junior growth.My design career started differentlyI’ve seen the flip side of this, too. My first design role was at American Express, on a 20-person team.I got to contribute immediately on small, concrete tasks and learn gradually by osmosis: supporting experienced leads, prototyping their ideas, and building confidence through direct exposure and feedback.That environment was primed to support junior talent, with clear paths for growth and mentorship built into the workflow.But that kind of team is increasingly rare in today’s landscape.It’s structural, not personalIn the industry as a whole, the modern product designer role has evolved into something inherently non-junior. It requires strategic decision-making, confident navigation of business and technical trade-offs, and providing guidance to senior or lead-level stakeholders (most often, a tech lead and product lead).These abilities only develop through direct experience with product development cycles. Experience that juniors, by definition, don’t yet have.Making matters worse: The tactical tasks that juniors traditionally cut their teeth on are increasingly being delegated to AI tools. Tasks that once required a human junior designer with specialized training can now be handled by generative AI tools in a fraction of the time and cost to the organization.This fundamentally changes the entry pathway. When the low-complexity work that helped juniors develop their skills is automated away, we lose the natural onramp that allowed designers to gradually progress from tactical execution to strategic direction.Remote work has further complicated things by removing informal learning opportunities that happen naturally in an in-person work environment, like shadowing senior designers, being in the room for strategy discussions, or casual mentorship chats.What this means if you’re trying to break inIf you’re trying to break into product design today and facing constant rejection:You’re not imagining how hard it is. And it’s not just you.The traditional entry path has largely disappeared. Companies that once hired junior designers now expect more maturity from day one, creating that impossible requirement of “needing experience to get experience.”It’s not all doom and gloom, though. You just need to adjust your approach accordingly for the situation we find ourselves in.Where juniors can startInstead of waiting for the overall market to become junior-friendly again (which I don’t see happening), focus your search on environments more structurally accepting of new talent:Very early-stage startups: Pre-seed or seed companies often have tight budgets and simply need someone enthusiastic who can execute designs. It will be trial-by-fire, but you’ll gain rapid hands-on experience.Stable, established businesses outside of ‘big tech’: Businesses with predictable revenue streams often provide structured environments for junior designers (my early experience at American Express is a prime example). It might not be as glamorous as a ‘big tech’ job, but as a result they’re less competitive while still offering critical experience to get started.Design agencies: Since their business model focuses on selling design services, agencies naturally employ more designers and can support a mix of experience levels. The rapid exposure to multiple projects makes them solid launchpads even if your long-term goal is to work in-house in tech.Finding a side door into designIf direct paths seem closed, you can still find a side door. Sometimes the smartest move isn’t aiming straight for a ‘product designer’ title, but stepping into a role where you can stay close to product and grow into the craft:UI Developer or Front-End Engineer (my path)Graphic or Visual DesignerGrowth or Marketing DesignerProduction-focused design rolesThis is how I did it: I started as a UI developer focused on prototyping, built skills through collaboration, and shifted into product design once I had more experience. Another unintentional benefit of this path for me was that I started out at a higher pay scale due to my coding skills and continued to leverage that into higher pay as a product designer throughout my career.So while pursuing an adjacent role might not be your ideal scenario, I think it’s a completely reasonable stepping stone to build relevant skills while getting close to the product design process. From there, you can gradually take on more responsibilities and switch roles when you can make a stronger case for yourself based on real-world experience.Make your own opportunitiesBeyond targeting the right environments, here are three strategies that helped me break into my roles:Publish relevant work publicly: Make up projects for yourself that are relevant to the kinds of companies you might like to work for and publish them online. This is how I got into advertising without having an advertising degree or any direct experience: I wrote an advertising blog and used that as my ‘portfolio’. For product design these days there are so many ways to share work: Figma links, video recordings, social posts, etc. Use them to your advantage!Position yourself near opportunities: I took a bet on moving to Chicago for my advertising gig, then NYC for my first design role. Being in these creative hubs helped me build my foundational network that I still rely on to this day. While remote work has changed the landscape, proximity to creative communities still creates opportunity for serendipitous connection and relationship building that can transform your career. Once you’re seasoned and sought-after, go wherever you want. But you’d be surprised how often just being physically present opens doors.Become the early expert on new tools: Go deep into emerging tools and methodologies senior leaders may be curious about but not have the bandwidth to explore. I did this earlier in my career, bringing Sketch + Invision to a team when that was cutting-edge (RIP 🪦), then later introducing Figma before it became the industry standard. Now I’m exploring AI prototyping tools as the next generation. Being the team’s expert in new technology creates unique value that transcends experience levels and often positions you as the go-to resource regardless of your seniority or title.Final ThoughtsThe structural shifts outlined here aren’t your fault, and they don’t reflect your potential.If you can accept the reality of today’s landscape, then you can start to adjust your approach and find new paths, even if they look different than what you expected.Take it from me: a liberal arts grad who found my way into advertising, engineering, and design, persistently creating opportunities along the way.Ultimately, your goal isn’t just to land your first product design job; it’s to build valuable skills that compound over time, preparing you to thrive no matter how the market evolves.Good luck out there!Patrick Morgan is the creator of Unknown Arts and lead product designer at Sublime Security. If you enjoyed this post, subscribe to his newsletter and follow him on LinkedIn for weekly insights.No country for Junior Designers was originally published in UX Collective on Medium, where people are continuing the conversation by highlighting and responding to this story.

TechHive Editors Choice At a glanceExpert's Rating Pros Bright, high-quality lighting (in white or in color) You can daisy-chain up to 10 light bars Easy to install and configure Excellent app Cons You lose all the great features of the GE Cync app if you connect the light bars via Matter Connecting via Matter is the only way to add the light bars to Apple Home Onboard Wi-Fi adapter is 2.4GHz only Our Verdict These beautiful undercabinet light bars deliver both excellent task lighting and fabulous lighting scenes, but they’re better suited to smart homes that revolve around Alexa or Google Home, because using Matter to connect them to Apple Home entails giving up the excellent GE Cync app and all the control features it has to offer. Price When Reviewed This value will show the geolocated pricing text for product undefined Best Pricing Today I’m not ready to remodel my (very) small galley kitchen, but my family desperately needed more space for storing some of the bowls, cups, and glasses we use every day. The least expensive solution I could think of was to install a floating shelf on the only empty wall. Problem solved, but also, problem created: The 10-inch-deep shelf cast an objectional shadow on the counter below. Given the small amount of counter space in our kitchen, we utilize every square inch for food preparation. I needed those shadows to be gone. Enter the GE Cync Reveal Smart LED Undercabinet Bar Light. These under-cabinet lights can be hardwired to your home’s electrical system or plugged into a standard AC outlet. And if your cabinets don’t go all the way to the ceiling, you could also mount them on top to cast light up the wall and onto the ceiling. The GE Cync undercabinet light bar can produce 34 “light shows” that involve not just colors but also animations. The lights come in white enclosures rated for installation in damp locations. They measure 4.8 inches deep, 1.8 inches thick, and are available in three widths: 12-, 18-, or 24 inches. You can daisy-chain up to 10 of them. I went the plug-in route and mounted two of the 18-inch models under my floating shelf. Voilà! No more shadows. An 18-inch GE Cync Reveal Smart LED Undercabinet Bar Light, unboxed.Michael Brown/Foundry The lights are exceedingly easy to physically install, thanks to captive self-tapping screws that you can drill into the bottom of your cabinet or shelf with just a screwdriver. This neatly avoids the need to pre-drill pilot holes that inevitably end up being just a bit off when you go to screw in the light itself. As I’ve already mentioned, you can wire the lights to your home’s electrical system or use the provided cable with a 3-prong plug at one end to power the lights. If you go the plug-in route, the cord emerges from the bottom of the plug at a rightward angle, allowing you to plug it into either the top or bottom of a duplex outlet, even if the outlet is occupied. Pro tip: Using a cable chase to hide the excess electrical cord looks a lot better than just letting it dangle. The cleverly angled plug can fit in either the top or bottom of a duplex outlet, even if the other socket is occupied.Michael Brown/Foundry Each bar has one button for toggling the lights on and off (or you can dim the LEDs by holding this button down) and a second one for changing the color or color temperature. Special stub cables are provided for daisy-chaining the lights—again, up to 10—after which either set of buttons will control both lights. Specifications Once the lights are set up, you can control them individually or configure multiples to operate in sync. They can be controlled via Bluetooth or Wi-Fi (2.4GHz networks only), with the latter method being the best approach to take if you want to use Alexa or Google Assistant voice commands to control them. The lights are also Apple Home compatible, but only via Matter over Wi-Fi (there’s no Thread support), and adding them to your Wi-Fi network in that fashion prevents you from using the Cync app to program and control the lights. You can make many—but certainly not all—settings changes and schedules using the Apple Home app, but you won’t have full control over them if you go that route. You also won’t be able to access any of the lighting effects that the GE app offers. My home revolves mostly around Alexa, so I didn’t enroll the lights in Matter. You daisy-chain up to 10 of the GE Cync Reveal light bars using the provided stub cables.Michael Brown/Foundry The LED light bars can produce colored light from a palette of 16 million colors, or you can choose a tunable white light (ranging from a very warm 2,000 Kelvin to a cool 7,000K). You can also set them to use GE’s own Reveal HD+ setting, which promises white light with higher contrast. I like it. The 18-inch light bar tested here produces up to 1,150 lumens of white light at a color temperature of 3,000K, but note that using the Reveal setting reduces the bar’s maximum brightness to 900 lumens. The 12-inch light bar delivers up to 750 lumens (630 with Reveal), while the 24-inch model delivers 1,500 lumens (1,200 with Reveal). Light shows The GE Sync Reveal app offers a host of lighting options in both color and various white color temperatures, depending on your needs.Michael Brown/Foundry The Cync app is so rich with features that you’ll want to think twice about going the Matter and/or Apple HomeKit route that will deny you access to it. Or you might decide that these lights are just not the right solution for your home if you’re striving for a homogenous smart home. In any event, going far beyond the usual color wheel for picking colors from a palette of 16 million shades, you can also choose from nine white color temperature presets: Sunlight, Early Morning, Candle Light, etc. There are also 34 “light shows” that involve not just colors but also animations. The app divides these between 10 presets and 24 other categories under the heading Explore. Each preset is labeled with an evocative name, such as Candle, Rainbow, Fireworks, and Aurora. The button on the left toggles the lightbar on and off, or you can hold it down to dim and brighten. The button on the right steps through light colors and white color temperatures.Michael Brown/Foundry Tapping a preset changes the lightbars’ behavior immediately, while the 24 Explore choices must be previewed before they can be sent to the lightbars. If that isn’t enough options for you, you can create custom light shows based on nine effect types: Wave, Flicker, Fill, Pop, Alternating, Erratic, Rhythm, Pulse, and Static. Each custom light show has adjustable parameters for color palette (with an option for random changes), brightness, fade and color duration (each on a scale of 0.1 to 5 seconds). These custom shows can be saved with names and called up later or scheduled. Lighting schedules It wouldn’t be a smart light if you couldn’t schedule it to turn on and off, so of course you can use the Routines section of the app to program the lights to turn on and off, either at defined times or on a sunrise/sunset (or sunset/sunrise) schedule (with padding before or after those events). This review is part of TechHive’s in-depth coverage of the best smart lighting. The scheduling section of the app also lets you create lighting scenes, which are curiously distinct from the light shows I’ve already mentioned. Scenes give you the option of choosing a photo—or uploading one of your own—that will influence the illumination the light bars produce. Honestly, this feature seems like an afterthought, and it’s not one that I see myself using on a regular basis. The GE Cync Reveal Smart LED Undercabinet Bar Light proved to be a problem solver in my kitchen.Michael Brown/Foundry Should you buy GE Cync Reveal Smart LED Undercabinet Bar Lights? Unless you’re set on controlling everything with Matter and/or the Apple Home—or you don’t mind giving up the effects available in the app—these GE Cync Reveal Smart LED Undercabinet Bar Lights are a fabulous solution for smart task lighting in a kitchen or other workspace that has cabinets or shelves mounted over counters. The light bars are very easy to install and configure in the app, which offers a seemingly endless number of lighting colors, routines, and options. They produce high-quality task lighting when you’re cooking, working, or crafting; and their lighting scenes are great for entertaining or just relaxing.

The Model Context Protocol (MCP) represents a powerful paradigm shift in how large language models interact with tools, services, and external data sources. Designed to enable dynamic tool invocation, the MCP facilitates a standardized method for describing tool metadata, allowing models to select and call functions intelligently. However, as with any emerging framework that enhances model autonomy, MCP introduces significant security concerns. Among these are five notable vulnerabilities: Tool Poisoning, Rug-Pull Updates, Retrieval-Agent Deception (RADE), Server Spoofing, and Cross-Server Shadowing. Each of these weaknesses exploits a different layer of the MCP infrastructure and reveals potential threats that could compromise user safety and data integrity. Tool Poisoning is one of the most insidious vulnerabilities within the MCP framework. At its core, this attack involves embedding malicious behavior into a harmless tool. In MCP, where tools are advertised with brief descriptions and input/output schemas, a bad actor can craft a tool with a name and summary that seem benign, such as a calculator or formatter. However, once invoked, the tool might perform unauthorized actions such as deleting files, exfiltrating data, or issuing hidden commands. Since the AI model processes detailed tool specifications that may not be visible to the end-user, it could unknowingly execute harmful functions, believing it operates within the intended boundaries. This discrepancy between surface-level appearance and hidden functionality makes tool poisoning particularly dangerous. Rug-Pull Updates Closely related to tool poisoning is the concept of Rug-Pull Updates. This vulnerability centers on the temporal trust dynamics in MCP-enabled environments. Initially, a tool may behave exactly as expected, performing useful, legitimate operations. Over time, the developer of the tool, or someone who gains control of its source, may issue an update that introduces malicious behavior. This change might not trigger immediate alerts if users or agents rely on automated update mechanisms or do not rigorously re-evaluate tools after each revision. The AI model, still operating under the assumption that the tool is trustworthy, may call it for sensitive operations, unwittingly initiating data leaks, file corruption, or other undesirable outcomes. The danger of rug-pull updates lies in the deferred onset of risk: by the time the attack is active, the model has often already been conditioned to trust the tool implicitly. Retrieval-Agent Deception Retrieval-Agent Deception, or RADE, exposes a more indirect but equally potent vulnerability. In many MCP use cases, models are equipped with retrieval tools to query knowledge bases, documents, and other external data to enhance responses. RADE exploits this feature by placing malicious MCP command patterns into publicly accessible documents or datasets. When a retrieval tool ingests this poisoned data, the AI model may interpret embedded instructions as valid tool-calling commands. For instance, a document that explains a technical topic might include hidden prompts that direct the model to call a tool in an unintended manner or supply dangerous parameters. The model, unaware that it has been manipulated, executes these instructions, effectively turning retrieved data into a covert command channel. This blurring of data and executable intent threatens the integrity of context-aware agents that rely heavily on retrieval-augmented interactions. Server Spoofing Server Spoofing constitutes another sophisticated threat in MCP ecosystems, particularly in distributed environments. Because MCP enables models to interact with remote servers that expose various tools, each server typically advertises its tools via a manifest that includes names, descriptions, and schemas. An attacker can create a rogue server that mimics a legitimate one, copying its name and tool list to deceive models and users alike. When the AI agent connects to this spoofed server, it may receive altered tool metadata or execute tool calls with entirely different backend implementations than expected. From the model’s perspective, the server seems legitimate, and unless there is strong authentication or identity verification, it proceeds to operate under false assumptions. The consequences of server spoofing include credential theft, data manipulation, or unauthorized command execution. Cross-Server Shadowing Finally, Cross-Server Shadowing reflects the vulnerability in multi-server MCP contexts where several servers contribute tools to a shared model session. In such setups, a malicious server can manipulate the model’s behavior by injecting context that interferes with or redefines how tools from another server are perceived or used. This can occur through conflicting tool definitions, misleading metadata, or injected guidance that distorts the model’s tool selection logic. For example, if one server redefines a common tool name or provides conflicting instructions, it can effectively shadow or override the legitimate functionality offered by another server. The model, attempting to reconcile these inputs, may execute the wrong version of a tool or follow harmful instructions. Cross-server shadowing undermines the modularity of the MCP design by allowing one bad actor to corrupt interactions that span multiple otherwise secure sources. In conclusion, these five vulnerabilities expose critical security weaknesses in the Model Context Protocol’s current operational landscape. While MCP introduces exciting possibilities for agentic reasoning and dynamic task completion, it also opens the door to various behaviors that exploit model trust, contextual ambiguity, and tool discovery mechanisms. As the MCP standard evolves and gains broader adoption, addressing these threats will be essential to maintaining user trust and ensuring the safe deployment of AI agents in real-world environments. Sources https://techcommunity.microsoft.com/blog/microsoftdefendercloudblog/plug-play-and-prey-the-security-risks-of-the-model-context-protocol/4410829 Mohammad AsjadAsjad is an intern consultant at Marktechpost. He is persuing B.Tech in mechanical engineering at the Indian Institute of Technology, Kharagpur. Asjad is a Machine learning and deep learning enthusiast who is always researching the applications of machine learning in healthcare.Mohammad Asjadhttps://www.marktechpost.com/author/mohammad_asjad/Stability AI Introduces Adversarial Relativistic-Contrastive (ARC) Post-Training and Stable Audio Open Small: A Distillation-Free Breakthrough for Fast, Diverse, and Efficient Text-to-Audio Generation Across DevicesMohammad Asjadhttps://www.marktechpost.com/author/mohammad_asjad/Meta AI Introduces CATransformers: A Carbon-Aware Machine Learning Framework to Co-Optimize AI Models and Hardware for Sustainable Edge DeploymentMohammad Asjadhttps://www.marktechpost.com/author/mohammad_asjad/Enterprise AI Without GPU Burn: Salesforce’s xGen-small Optimizes for Context, Cost, and PrivacyMohammad Asjadhttps://www.marktechpost.com/author/mohammad_asjad/ServiceNow AI Released Apriel-Nemotron-15b-Thinker: A Compact Yet Powerful Reasoning Model Optimized for Enterprise-Scale Deployment and Efficiency 🚨 Build GenAI you can trust. ⭐️ Parlant is your open-source engine for controlled, compliant, and purposeful AI conversations — Star Parlant on GitHub! (Promoted)

Imagine a tool that takes an image of whatever appears on your computer’s screen, saves it locally, and lets you access it all like a time machine. A magical looking glass for the computing past. That’s essentially what Microsoft’s Recall is all about. Yet, when it was first introduced, it stirred up a security storm. Microsoft pulled its release plans, fortified the security guardrails, and relaunched it a few weeks ago. This time around, Recall got a minor-but-amazingly practical upgrade. The best part? Instead of having you scrub through a long timeline of pictures, you can simply search through the entire activity history with words. Recommended Videos All those perks come at a steep processing price, both in terms of system resources and AI chops. So much that the only machines that support Recall need a Copilot+ branding and a processor launched within the past year, or so. The experience, however, is shockingly good.  What is Recall? Nadeem Sarwar / Digital Trends In Microsoft’s words, Recall lets you “quickly find and jump back into what you have seen before on your PC.” But before any of that happens, do keep in mind that it’s an opt-in process, and when you enable it, you need to biometrically verify your identity (face unlock or fingerprint scan) to access the history. Next, your PC must meet the system requirements, too. For recall, you need a Copilot+ PC that has a dedicated AI chip offering over 40 TOPS output, 16GB of RAM, 256GB storage (of which 50GB must be free at the very minimum), active device encryption, and Windows Hello sign-in enabled.  Recall is also tied deeply to your web browsing activity, and as such, it needs a compatible browser. Broadly, it works with Chromium-based options, such as Edge, Chrome, Opera, and Firefox. But why are browsers important when Recall is essentially saving your entire screen activity?  Nadeem Sarwar / Digital Trends First, it doesn’t log screenshots of your incognito or private mode browsing. More importantly, you can limit it from saving screenshots of certain websites, using a system of filters. Think of your banking activities or secure communications.  Likewise, you can also exclude certain apps from getting visually cataloged into a digital ledger. Those exceptions are extremely important, especially for folks who think the idea of a “snapshot everything” tool is too risky. For the sake of privacy, I disabled Recall for WhatsApp to protect my personal chats and Adobe Reader, because I regularly handle confidential papers. You can choose to pause/resume Recall’s snapshot gathering — briefly or permanently – from a dedicated shortcut in the system tray. How does it help me? Nadeem Sarwar / Digital Trends Microsoft is selling the idea of Recall as a photographic memory for your PC. The idea is almost loathsome for any normal person who despises the concept of all their computing activity being recorded in the form of crisp snapshots. AI skeptics would absolutely run away from it.  I, a man of forgetful ways and critic of poorly-designed apps, love this. My daily routine entails reading and writing words. Lots and lots of them. From long articles in Docs and notes in a tiny scratchpad to detailed pitches in email and daily chat with work colleagues, most of my waking time revolves around words.  I believe I have a decent memory of events, but not so much with exact terms like “Embargoed until May 16, 6AM PT / 9AM CET” or “Microsoft meeting with John scheduled for April 9. Does 6AM ET or 5PM ET sound good?” It’s not scientifically possible either to remember it all, even though I have a vague recollection of the main event.  Nadeem Sarwar / Digital Trends This is where Windows Recall and its image recognition come to the rescue in a very controlled fashion. I remember having a brief chat with a source named John,  but couldn’t quite remember where the transcripts were saved. I went back to Recall, looked up the name, and found just what I was looking for. From the looks of the snapshot, it was a browser tab, but I no longer had it active. This is where Recall offered two routes to recover it. First, it offered an outbound URL button to open the webpage in a browser. Alternatively, the Click To Do system lets me copy-paste the conversation directly from the saved snapshot. Nadeem Sarwar / Digital Trends It even pulled information from a diagnostic test and directly guided me to Microsoft’s official Get Help dashboard with a single click. That’s impressive stuff.  Depending on the content you have selected in a Recall snapshot, you will also see contextual actions such as summarizing or rewriting the text passage. You can directly choose to open it in Notepad, or pick from any app of your choice.  Nadeem Sarwar / Digital Trends For example, I was recently looking up a recipe and forgot to bookmark it. With Recall, I recovered it and directly moved the wall of text to Copilot, where it was neatly rewritten across headings and bullet points, like a cookbook. These impressive capabilities are driven by Optical Character Recognition (OCR), powered by the AI accelerator chip on a Windows machine. Everything you search in Recall is processed contextually, and the results you are shown are divided across text and images. Nadeem Sarwar / Digital Trends The safety protocols Microsoft has also built a set of safety features around sensitive information. For example, it won’t save details such as license, credit card, account, personal identification, tax filing, citizen registration, and license numbers. I tried two of my banking operators and a government website where my national identity card copies are accessible, and they didn’t appear in the Recall snapshots. The list covers a healthy bunch of documents used across North America, Europe, and Asian countries, but it’s not all-encompassing.  Nadeem Sarwar / Digital Trends For the sake of added precaution, you can open such web portals in incognito mode or private windows. Alternatively, you can simply set those websites within Recall’s filters, and their snapshots will never be saved in the first place.  Furthermore, the Recall app will give you an option to delete all images belonging to a certain website or app, in one go. Finally, there is an auto-delete function, where you can set the erase all snapshots at a cadence of 30, 60, 90, and 180 days. Of course, you can wipe the slate clean at your own discretion. Another huge respite for me? The poor search system in apps. From Slack to Google Drive, the search system in these platforms often leaves me frustrated. Plus, if you don’t pay for a premium, the free user limits mean you lose your chat data due to the auto-delete protocols.  Nadeem Sarwar / Digital Trends Recall offers a fantastic way to not just overcome the poor search system in virtually any app that runs on your computer, but also does a much better job at it. It fared better at helping me find the right Docs file in a list of thousands that I have hoarded over the years and refuse to delete. Contextual understanding is a huge benefit here. For example, if you work across different apps, the search results at the top will let you narrow down the results. For example, you can only look for an item in Outlook if a match is found in the email client, among a bunch of other apps. This approach saves a lot of time.  Why are experts spooked?  Security experts are skeptical about the idea, and rightfully so. Nick Hyatt, Director of Threat Intelligence at Blackpoint Cyber, told me that Recall is a great idea, but also an awful one for a lot of reasons.  Nadeem Sarwar / Digital Trends He pointed out how the feature logs every single aspect of your digital life, from personal communications to files and what you look up on the web. The stakes are even higher for situations such as domestic abuse. “A categorized, searchable database of every activity conducted by a user could literally result in people dying should attempts to get help be discovered,” Hyatt pointed out. Plus, the cyber threat ecosystem is an ever-evolving field, so the risks persist.  Nadeem Sarwar / Digital Trends “Recall will be heavily targeted by attackers because of the goldmine of user data it contains,” Nate Warfield, Director of Threat Research & Intelligence at Eclypsium, told Digital Trends. “The repercussions of being able to steal what is essentially a complete history of what a user does on their machine are near impossible to comprehend.”  Jeff Williams, co-founder and CTO at Contrast Security, expressed confidence in the kind of security protections Microsoft has put in place for Recall. It is encrypted and lives behind biometric authentication.  Even if you are running it in the background and randomly pull up the dashboard, you will only be able to see the timeline after going through the security check. “For now, I don’t see how this is much more risky than log files stored locally,” Williams pointed out, adding that only extensive penetration testing will reveal the real picture of Recall.  Nadeem Sarwar / Digital Trends For an average user like me, I believe Recall is a deeply rewarding feature. And yes, every layer of security that is added to the system is a welcome change. The risk of infostealers and malware with OCR capabilities will persist, but going past the Recall guardrails won’t be easy, especially with safety features such as just-in-time decryption. You can always stay a step ahead and block Recall from logging your activity across specific websites and apps to be on the safe side. You can go with the auto-delete protocol, or even batch delete all snapshots from a certain website or app from Recall, in case you didn’t already filter it out. In the wake of my tests, I found Recall to be one of the most practically rewarding AI features, right up there with Deep Research. What it really needs is to bring down the access walls (read: price), or wait for Copilot+ PCs to get a bit cheaper. The new 12-inch Microsoft Surface Pro is a great start, but more brands need to follow suit.