Microsoft wants AI 'agents' to work together and remember things - CTO tells conference

Microsoft envisions a future where any company's artificial intelligence agents can work together with agents from other firms and have better memories of their interactions, its chief technologist said on Sunday, 18 May ahead of the company's annual software developer conference.

Microsoft - which is due to visit Cyprus this month for talks to open offices, following President Nikos Christodoulides' recent visit to Silicon Valley - is holding its Build conference in Seattle on 19 May, where analysts expect the company to unveil its latest tools for developers building AI systems.

• Read more: President announces May visits of Tenstorrent, Microsoft and Google

Speaking at Microsoft's headquarters in Redmond, Washington, ahead of the conference, Chief Technology Officer Kevin Scott told reporters and analysts the company is focused on helping spur the adoption of standards across the technology industry that will let agents from different makers collaborate. Agents are AI systems that can accomplish specific tasks, such as fixing a software bug, on their own.

Scott said that Microsoft is backing a technology called Model Context Protocol (MCP), an open-source protocol introduced by Google-backed Anthropic. Scott said MCP has the potential to create an "agentic web" similar to the way hypertext protocols that helped spread the internet in the 1990s.

"It means that your imagination gets to drive what the agentic web becomes, not just a handful of companies that happen to see some of these problems first," Scott said.

Scott also said that Microsoft is trying to help AI agents have better memories of things that users have asked them to do, noting that, so far, "most of what we're building feels very transactional."

But making an AI agent's memory better costs a lot of money because it requires more computing power. Microsoft is focusing on a new approach called structured retrieval augmentation, where an agent extracts short bits of each turn in a conversation with a user, creating a roadmap to what was discussed.

"This is a core part of how you train a biological brain - you don't brute force everything in your head every time you need to solve a particular problem," Scott said.

(Source: Reuters)