At the annual Build Developer Conference, Microsoft made a strong push into the artificial intelligence field by launching Windows AI Foundry, a new platform that replaces the previous Copilot Runtime.
"At Microsoft, we believe the future of artificial intelligence will be built in the cloud, at the edge, and on Windows. Windows will continue as an open platform, offering developers the best capabilities and ultimate flexibility," the company announced in its official blog. "Windows AI Foundry is the evolution of Windows Copilot Runtime, providing a unified and reliable platform that supports the AI developer lifecycle from model selection, optimization, fine-tuning to client and cloud deployment." It added that Windows AI Foundry includes several features.
Windows AI Foundry can run various AI models directly on users' devices, including open-source options and Microsoft's own built-in models. The platform supports third-party contributions from companies like Nvidia, particularly including NVIDIA NIM (Nvidia Inference Microservices). These models are fully optimized to run on different hardware configurations, including CPUs, GPUs, and dedicated NPUs (Neural Processing Units) in Microsoft Copilot+ PCs.
Windows AI Foundry offers a set of APIs for instant access to commonly used AI functions such as text intelligence, object detection, image description, and text recognition. These features are supported by Microsoft's integrated models on Copilot+ devices, designed to support general and domain-specific applications. Another notable feature is support for LoRA (Low-Rank Adaptation), allowing developers to fine-tune Microsoft's built-in Phi Silica models with custom datasets. This customization ability is particularly important for organizations looking to adapt AI for niche or specific use cases without large-scale retraining.
To enable richer user interactions, Microsoft has also introduced APIs for semantic search and knowledge retrieval. These features support the implementation of Retrieval-Augmented Generation (RAG) workflows, enabling applications to perform context-sensitive queries using structured and unstructured data sources.
Microsoft has simplified local deployment through Foundry Local, a module that can be installed with simple terminal commands. Developers can leverage command-line tools (CLI) and software development kits (SDKs) to test, debug, and deploy AI models from their local environments. These tools will be fully integrated into Windows 11 and the Windows App SDK, promoting native support for AI-driven applications within the Windows ecosystem. The integration of Windows ML further enhances deployment efficiency by eliminating the need for developers to manually include ML runtimes or hardware drivers in their application packages.
The success of Windows AI Foundry partly depends on Microsoft's collaboration with major hardware vendors. The company stated that it is actively working with AMD, Intel, Nvidia, and Qualcomm to ensure seamless model performance across various chipsets and devices. These partnerships are expected to help standardize AI deployment on Windows, making the platform attractive not only to individual developers but also to enterprise software teams.
