Microsoft is stepping up its efforts to help developers create AI enhanced applications for Windows. During its Build 2025 conference, the company introduced Windows AI Foundry, building on and rebranding the Windows Copilot Runtime platform launched the previous year.
This platform is meant to serve as a unified space where developers can fine tune, optimize, and deploy AI models for Windows based software. Microsoft sees both local and cloud driven AI technology as critical to its future growth, with AI technologies already generating billions in annual revenue.
New Tools for On Device AI
A key feature of Windows AI Foundry is its ability to identify the hardware on a Windows computer and automatically supply the right software components needed for specific AI models. The company also promises automatic updates to these components to support new devices as they appear, making it easier for developers to keep their AI applications current.
Another major announcement is Foundry Local, which extends AI capabilities directly to client devices, including support for both macOS and Windows. Foundry Local includes specialized modules designed to run AI models natively and tools that use ONXX Runtime, an open source project focused on boosting cross platform AI performance.
Developers also benefit from command line interface options with Foundry Local, allowing them to interactively browse, test, and run AI models directly on their local machines. Commands such as “Foundry model list” and “Foundry model run” streamline this hands on experience.
With these tools, detection of hardware like CPUs, GPUs, and NPUs becomes automatic, so developers can quickly find models compatible with their systems. Integration is further simplified through the Foundry Local SDK, enabling fast embedding of these features within developer applications.
