Microsoft is accelerating its AI-first vision for Windows by introducing native support for the Model Context Protocol (MCP) and unveiling the Windows AI Foundry, a platform aimed at simplifying AI model deployment on Windows devices.
MCP, originally introduced by Anthropic, is an open-source standard dubbed the “USB-C of AI apps.” It allows AI agents to seamlessly interact with various applications, web services, and now Windows system components. With Windows embracing MCP, developers can build smarter, context-aware AI agents capable of performing advanced tasks like file retrieval, app control, and real-time system interaction.
“We want Windows to evolve into a platform where agents become a core part of how users interact with their devices,” said Pavan Davuluri, Microsoft’s Windows chief.
To support this AI agent ecosystem, Microsoft is implementing an MCP registry that securely lists accessible MCP servers, such as those for the Windows File System or Subsystem for Linux. In a live demo, Microsoft showcased Perplexity AI using MCP to naturally query files without manual folder selection—just a voice command like “find vacation files in my documents” does the job.
However, the move comes with serious security considerations. Microsoft is implementing strong safeguards to prevent threats such as prompt injection and token theft. “We’re treating large language models as untrusted by default,” said David Weston, Microsoft VP of enterprise and OS security.
Alongside MCP, Microsoft also launched the Windows AI Foundry, which enables developers to use models from Foundry Local, Nvidia NIMs, and others directly on Copilot Plus PCs. The initiative works with AMD, Intel, Nvidia, and Qualcomm to optimize AI model performance on Windows hardware.
If this article helped you, please consider supporting our work. Every small contribution keeps Abijita.com independent and running.
