Table of Contents
As Microsoft prepares to unveil its latest advancements at the Build 2025 developer conference, Chief Technologist Kevin Scott revealed a bold vision for the future of artificial intelligence (AI). In a preview of what’s to come, Scott shared plans to make AI agents not just smarter but more collaborative, allowing agents from different companies to seamlessly work together and retain better memory of past interactions.
This vision could fundamentally change the way AI systems operate, transitioning from isolated, task-specific agents to more interconnected and adaptive systems. Scott emphasized that the company’s efforts are focused on laying the groundwork for AI agents to break free from silos and collaborate across firms, creating new possibilities for developers and businesses alike.
Supporting the Model Context Protocol (MCP)
One of the key technologies Microsoft is backing to bring this vision to life is the Model Context Protocol (MCP). This open-source protocol was introduced by Anthropic, a company backed by Google, and has the potential to enable AI agents from different companies to collaborate in a way that mimics the internet’s evolution in the 1990s.
According to Scott, the MCP could pave the way for an “agentic web”, a concept that would allow diverse AI systems to interact and build upon each other’s capabilities. Just as the World Wide Web was propelled by the adoption of hypertext protocols, MCP could be the catalyst for creating a shared foundation for AI agent collaboration.
“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 remarked, suggesting that the shift toward a more open, collaborative AI ecosystem is a powerful opportunity for innovation.
Enhancing AI Memory: A Step Toward More Human-Like Intelligence
While AI collaboration is one major aspect of Microsoft’s vision, improving AI memory is another crucial piece of the puzzle. Currently, most AI systems function in a transactional manner, responding to user inputs without retaining context from previous interactions. However, Scott explained that better memory in AI agents could allow them to act more like human beings, remembering past conversations and offering more personalized responses.
WhatsApp Rolls Out AI-Powered Profile Photos for iOS Users
The challenge, however, is that improving AI memory is computationally expensive and requires significant resources. This is why Microsoft is focusing on a new approach called structured retrieval augmentation. This technique allows an AI agent to extract and store short, key pieces of conversation data, essentially creating a roadmap to the dialogue that helps the agent “remember” the important details without overloading its system.
“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 explained. By implementing a more structured method of memory, Microsoft aims to make AI more efficient and adaptable while still scaling its capabilities.
Implications for the Future of AI
The idea of AI agents that can collaborate across companies and possess longer-term memory opens up exciting possibilities for the future. For businesses, this could mean more efficient problem-solving and better customer interactions, as AI agents would be able to understand context and recall past discussions to improve their performance over time.
Additionally, the agentic web envisioned by Microsoft could lead to an entirely new ecosystem of AI agents, potentially disrupting industries and enabling innovation in ways previously thought impossible. The goal is to create a world where AI collaboration is as seamless as browsing the internet, where agents are interconnected, adaptive, and smarter.
For developers, these advancements could provide new tools and frameworks to build more sophisticated and efficient AI systems, ultimately leading to broader adoption of AI across various sectors, from healthcare to software development.
