Concurrent with the rise of artificial intelligence in recent years has been an uptick in fear of it. With every new use of artificial intelligence comes the fear that it will cost jobs. Now that fear has expanded to not only costing humans their jobs, but killing off software as a service (SaaS) applications as well. Generative AI was the first boogieman, and now concern is growing around the impact of AI agents, otherwise known as agentic AI. Agentic AI is an artificial intelligence system designed to operate autonomously or semi-autonomously on behalf of a user or organization to perform specific tasks. AI agents take initiative and make decisions based on data input from other applications or sensors. They can adjust their behavior based on context, user preferences, new data, or environmental changes. It is growing at a decent rate. By 2028, Gartner estimates that at least 15% of day-to-day work decisions will be made by autonomous AI agents, up from 0% in 2024. This has led to the fear that AI agents will replace SaaS apps by rendering the software irrelevant. A simple search under the phrase “Will AI agents replace SaaS?” yields dozens of articles and postings asking that same question, fueled in part by remarks Microsoft CEO Satya Nadella made to podcasters Bill Gurley and Brad Gerstner on their BG2 podcast in December. “I think the notion that business applications exist — that’s probably where they’ll all collapse in the agent era. Because if you think about it, they are essentially CRUD [create, read, update, delete] databases with a bunch of business logic,” Nadella said. “The business logic is all going to these agents, and these agents are going to be multi-repo CRUD. So they’re not going to discriminate between what the back end is. They’re going to update multiple databases, and all the logic will be in the AI tier, so to speak. Once the AI tier becomes the place where all the logic is, then people will start replacing the back ends,” he said. But Nadella went on to discuss how AI agents could be used in cooperation with business apps, where apps like Excel and Word act as specialized canvases for the agents. “So yes, I think there will be disruption,” he said, but AI agents are more likely to complement apps than to eliminate them. “I think that we’re a long way away from AI agents replacing SaaS,” said Tom Coshow, senior director and analyst at Gartner. “It’s important to remember that inside SaaS platforms, there are automated workflows that people have built that they still want to use. The SaaS platform still serves as a system of record for them.” AI agents may make it so that people can do their work without going into a SaaS app’s user interface, he added, but the app will still be in use. “[Agentic AI] is going to enhance it [SaaS], it’s not going to replace it,” he said. The relationship between agents and SaaS The reason agentic AI is perceived as a threat to SaaS and not traditional apps is that traditional apps have all but disappeared, replaced in favor of on-demand versions of former client software. But it goes beyond that. AI is considered a potential threat to SaaS for several reasons, mostly because of how it changes who is in control and how software is used. Agentic AI changes how work gets done because agents act on behalf of users, performing tasks across software platforms. If users no longer need to open and use SaaS apps directly because the agents are doing it for them, those apps lose their engagement and perceived usefulness. That ultimately translates into lost revenue, since SaaS apps typically charge either per user or by usage. An advanced AI agent can automate the workflows of an entire department, which may be covered by multiple SaaS products. So instead of all those subscriptions, you just use an agent to do it all. That can lead to significant savings in software costs. (See also: How will AI agents be priced? CIOs need to pay attention) On top of the cost savings are time savings. Jeremiah Stone, CTO with enterprise integration platform vendor SnapLogic, said agents have resulted in a 90% reduction in time for data entry and reporting into the company’s Salesforce system. “That doesn’t mean that we’re going to see an impact to the workforce, per se, but rather, we’re going to remove a lot of the drudgery, so that people can do their jobs more effectively,” he said. Mike Wertz, program engineering lead at Optio.ai, developer of a customer data platform designed to help organizations integrate and analyze customer data, said that because agentic AI has the ability to learn, it can adapt and modify the rules that a SaaS application operates by. So, for example, suppose a form has multiple data entry windows for variables like name, address, phone number, etc., and the form layout changes. Inputs have been moved around the screen and are in different places. Perhaps new data entry elements have been added. A SaaS app would require a change to the rules for input and data entry, but the AI agent would simply learn the new structure and adapt. Agents don’t use the UI; they use the back-end interface and have the ability to learn changes to data input. “In using AI, it’s adapting to changes within that document,” said Wertz. “The more you use it, the more you train models. They learn. That’s the benefit of agentic AI: it gains flexibility.” The state of agentic AI The current maturity level of agentic AI is low, Gartner’s Coshow said, and as such, there are potential drawbacks, risks, and other potential issues to note. The main pitfall of AI agents is that most people are building them using large language models (LLMs), and LLMs are probabilistic. That means that you might be working with data that is incomplete or biased, so you might get an answer that is not as precise as you want — or worse, biased in some way. “What’s happening in an AI agent is you’re giving it data so that it can make a decision, and if the data itself is biased or wrong, then you’re going to ultimately get a wrong action,” said Coshow. He advises gradual use of data-driven AI agents until a company or entity has its data “in shape,” as it were. (See also: AI agents can (and will) be scammed) Another question to ask is: is that data properly secured? To function, AI agents need access to data — sometimes highly restricted data. “If you build an AI agent using your permissions, and I get my hands on the agent, can I now see data that I shouldn’t be able to see?” he asked. “I would say, make sure your data is secured before you start building AI agents, and that the user security will stay in force from the data level up to the agent level.” While agentic AI is still a maturing technology, SaaS providers like Salesforce and ServiceNow are rapidly closing the gap by building no-code AI agent platforms that promise to make it easy for enterprises to automate parts of a workflow. “So as we go through this year, I think we’ll see more and more workflows being automated and a lot of time saved,” Coshow said. The agentic AI market is a mix of startups, large established players, and homegrown, privately developed applications, said Coshow. “All the big players are going to be building in the ability to create AI agents inside their platform, and for obvious reasons. If I’m using a software platform and I can’t build automation, I can go someplace else to build that automation,” he said. “I would say that AI agents inside SaaS platforms is a relationship that is going to evolve… The relationship between AI agents, the SaaS platform, and the human beings that work in that environment is going to be an evolution that will be really interesting to watch, but it will definitely be gradual,” he added. المصدر: https://www.computerworld.com/article/3981415/does-agentic-ai-spell-doom-for-saas.html #Does #agentic #spell #doom #for #SaaS