Microsoft Unveils Windows AI Foundry at Build 2025

Microsoft has introduced Windows AI Foundry at Build 2025, a new platform designed to enhance AI development on Windows. The platform, an evolution of Windows Copilot Runtime, provides a unified environment for model selection, optimization, fine-tuning, and deployment across client and cloud environments.

Windows AI Foundry integrates Windows ML, the built-in AI inferencing runtime, which allows developers to deploy models efficiently across CPUs, GPUs, and NPUs from partners like AMD, Intel, NVIDIA, and Qualcomm. The platform also includes Foundry Local and other model catalogs, offering developers quick access to ready-to-use open-source models.

Additionally, Microsoft announced new APIs for semantic search and knowledge retrieval, enabling developers to build natural language search and retrieval-augmented generation scenarios in their applications. The introduction of LoRA (low-rank-adaption) for fine-tuning models with custom data further enhances the platform's capabilities.

The company also revealed native support for the Model Context Protocol (MCP) in Windows 11, which will allow AI agents to connect with native Windows apps, facilitating seamless agentic interactions. This feature will be available in a private developer preview with select partners in the coming months.