Word that a group of former Google executives plans to build an operating system for AI agents underscores the fact that there is a distinction between what application platforms are doing with agents today and what an AI agent-first OS would entail, said industry analyst Brian Jackson.
Jackson, principal research director at Info-Tech Research Group, said Wednesday that the OSes we use today “were built for a software architecture that is file-based and designed to be used with a computer and a mouse. An OS that is built around AI could look at defining computing as data-based, with intelligent models that continuously learn and adapt based on new data.”
He made the comments following the announcement from venture capitalist CapitalG, Alphabet’s independent growth fund, that it is co-leading a $56 million seed round in /dev/agents, an organization co-founded by former Google vice president of engineering David Singleton, along with former colleagues Hugo Barra, Ficus Kirkpatrick, and Nicholas Jitkoff, all of whom had held senior positions in the company.
On his LinkedIn page, Singleton, who will serve as CEO for the new firm, stated, “modern AI will fundamentally change how people use software in their daily lives. Agentic applications could, for the first time, enable computers to work with people in much the same way people work with people.”
He added that modern AI will not “happen without removing a ton of blockers. We need new UI [user interface] patterns, a reimagined privacy model, and a developer platform that makes it radically simpler to build useful agents. That’s the challenge we’re taking on.”
While there are few details being released on how the new OS will be developed — the /dev/agents web site is definitely a work in progress — according to a posting on the CapitalG web site the company, which will be based in San Francisco, is building a team to “work closely with users and developers across the full stack (from tuning models up through to the UI layer) to solve a novel class of system and UX [user experience] problems created by AI.”
Every tech vendor, said Jackson, “will always want to win over developers to use their platforms to solve different enterprise problems. The more developers creating functionality and extending your platform, the more value for customers and the more money for the platform.”
It is, he said, “a win-win-win (In theory). AI agents are just the next wave of where these tech vendors want to compete to get developers invested. But by focusing on building a new OS from the ground up, /dev/agents has the chance to fundamentally change our relationship with computers.”
In a new operating system, he added, the UI, “could be based on less abstracted modes of interaction like natural language, gestures, or even eye movements. And AI agents would be like the next generation of applications built on top of the OS, where developers could use tooling to combine those AI elements in new ways to solve problems.”
There is no questioning AI agents’ importance, as evidenced by vendors ranging from Salesforce and Microsoft to ServiceNow and Snowflake now offering development platforms they claim make it simpler to build useful agents.
In a blog post released last month, Gartner analyst Tom Coshow wrote, “today’s AI models perform tasks such as generating text, but these are ‘prompted’ — the AI isn’t acting by itself. That is about to change with agentic AI (AI with agency). By 2028, 33% of enterprise software applications will include agentic AI, up from less than 1% in 2024, enabling 15% of day-to-day work decisions to be made autonomously.”
While agentic AI is still in early stages, it is, he said, “not too soon to gain an understanding of the technology, determine how to manage risk, and prepare your tech stack.”
Intelligent agents in AI, wrote Coshow, “will change decision making and improve situational awareness in organizations through quicker data analysis and prediction intelligence. While you’re sleeping, agentic AI could look at five of your company’s systems, analyze far more data than you ever could, and decide the necessary actions.”
Other analysts are in agreement. Deloitte, for example, is predicting that a quarter of companies that use generative AI (genAI) will launch agentic AI pilots or proofs of concept in 2025, with that number growing to half of genAI users by 2027. “Some agentic AI applications, in some industries, and for some use cases, could see actual adoption into existing workflows in 2025, especially by the back half of the year,” said Deloitte staff in a recent report.