New Delhi:

Microsoft is positioning MCP as the next step in how AI integrates into operating systems, not just passively generating responses, but actively engaging with apps, files, and services on behalf of users.

At Build 2025, Microsoft laid out its next big bet for Windows: native infrastructure for AI agents. The company introduced the Model Context Protocol (MCP) as a new system-level capability designed to let AI agents interact with Windows apps in a standardized, secure way. The goal is to make Windows 11 a foundational layer for what Microsoft calls the “agentic future”, one where intelligent agents can perform meaningful tasks across tools and platforms, directly from your PC.

What MCP brings to Windows

At its core, MCP is an open, lightweight protocol, essentially JSON-RPC over HTTP, that allows apps and AI agents to discover and invoke each other’s capabilities. The idea is to enable seamless orchestration between local tools and intelligent systems.

Windows 11 will support three roles within this setup:

MCP Clients (agents that initiate requests)

MCP Servers (apps or tools exposing functionality)

MCP Hosts (platforms like VS Code that enable these connections)

These interactions will be handled through a secure proxy built into Windows, which Microsoft says will enforce consistent policy checks, authentication, and logging.

Read mode: Wondering what is MCP? We explain it for you

Security-first design for agentic workflows

As agent-based systems grow in complexity, Microsoft has made security a top priority. According to David Weston, VP of Enterprise and OS Security at Microsoft, MCP was designed with threat models like cross-prompt injection, tool poisoning, and credential leakage in mind.

To mitigate risks, Windows will implement:

Runtime isolation of MCP servers

A centralized registry of trusted MCP servers

Mandatory code signing

Fine-grained user controls

“All sensitive actions done on behalf of the user must be auditable and transparent,” Weston said. Only apps that meet Microsoft’s baseline security requirements will be listed in the official MCP registry.

Industry partners backing the ecosystem

Microsoft will release MCP in a private developer preview in the coming months, with broader access expected later. While MCP is still early in its rollout, the company’s long-term vision is clear, a tightly integrated AI-native Windows environment, where apps and agents operate together securely and efficiently.

Major AI developers including OpenAI, Anthropic, Perplexity, and Figma are already working with Microsoft to bring MCP integration into their apps. “We’re excited to see Windows embracing AI agent experiences through its adoption of the Model Context Protocol,” said Kevin Weil, Chief Product Officer at OpenAI. “This paves the way for ChatGPT to seamlessly connect to Windows tools and services.”

Read More:

Google brings NotebookLM to iPhones and Android