CEO's Column
Search
More
Foundation Models

Microsoft May Adopt Anthropic’s MCP to Power a More Connected AI Future

ByRishabh Srihari
2025-05-19.about 2 months ago
Microsoft May Adopt Anthropic’s MCP to Power a More Connected AI Future
Microsoft plans to adopt Anthropic’s Model Context Protocol (MCP) to create a more connected, intelligent AI ecosystem at Build 2025.

Microsoft is reportedly preparing to adopt Anthropic’s Model Context Protocol (MCP) as it pushes for smarter, more collaborative AI systems. This development may be formally announced at Microsoft Build 2025, a major conference for developers working across the company’s platforms.

The MCP protocol, introduced by Anthropic in November 2024, offers a standardized way for AI agents to connect to external data hubs. It addresses one of the industry’s biggest challenges—how AI models retrieve and process third-party information.

Each cloud service and data system has its own methods. This creates inconsistencies in performance, inference time, and output quality. MCP solves this by introducing a uniform protocol. With MCP, AI agents can retrieve and use information from different systems more efficiently.

Also read: Anthropic Explores AI Consciousness and Model Welfare with New Research Program

Creating a Web of Intelligent Agents

Microsoft CTO Kevin Scott emphasized that the company wants to enable multiple AI agents to work together effectively. In comments reported by Reuters, Scott stated that this multi-agent approach is essential to building what he calls an “agentic web.”

He compared MCP to the early days of the internet, where the hypertext protocol (HTTP) helped web pages connect. MCP, in a similar way, could help AI agents cooperate, retrieve information, and perform connected tasks with a consistent context.

New Memory Capabilities Could Redefine Agent Intelligence

Microsoft is also addressing another common limitation: memory. Many AI agents only remember the current task, forgetting what happened before. That limits their usefulness in multi-step workflows.

To solve this, Microsoft is developing a method called structured retrieval augmentation. This system allows agents to recall important parts from earlier user interactions. Over time, this builds a longer-term memory and creates more helpful, context-aware agents.

At Build 2025, Microsoft is expected to reveal how its agents will integrate MCP and use structured retrieval to improve memory. Together, these upgrades could transform AI agents into dynamic collaborators capable of handling complex, multi-step user tasks.

Related Topics

Large Language Models (LLMs)LLMs

Subscribe to NG.ai News for real-time AI insights, personalized updates, and expert analysis—delivered straight to your inbox.