Advancing Windows for AI Development
At Build 2025, Microsoft introduced Windows AI Foundry and a new suite of developer tools aimed at strengthening Windows as a leading platform for AI development. This update brings powerful APIs, secure AI model deployment, open-source support for Windows Subsystem for Linux (WSL), and new hardware partnerships. With these enhancements, Windows targets flexibility, performance, and security for AI innovation both on-device and in the cloud.
Windows AI Foundry: A Comprehensive AI Platform
Windows AI Foundry builds on the Windows Copilot Runtime to cover the AI development lifecycle—from model selection and optimization to deployment on client devices and cloud environments. Key features include:
- Windows ML: A native AI inferencing runtime optimized for AMD, Intel, NVIDIA, and Qualcomm hardware.
- Foundry Local: Integrations with model catalogs like Ollama and NVIDIA NIMs offering open-source models tuned for Windows silicon.
- Inbox AI APIs: On-device models for language and vision tasks such as text summarization and object removal, tailored for Copilot+ PCs.
- LoRA support: Enables fine-tuning of the Phi Silica small language model with custom data.
- Semantic Search and Knowledge Retrieval APIs: Designed for building retrieval-augmented generation (RAG) scenarios and natural language search over local data.
Windows ML: The Backbone for AI Inference
Windows ML simplifies AI model deployment by removing the need for separate runtimes or drivers. It adapts to new AI hardware generations and offers a suite of optimization tools through the AI Toolkit for VS Code. Developers at companies like Adobe and McAfee are already leveraging Windows ML for efficient AI applications across diverse hardware setups.
Built-In AI APIs for Easy Integration
With the Windows App SDK 1.7.2, Microsoft offers ready-to-use AI APIs for language and vision tasks running locally. These APIs prioritize privacy and performance on neural processing units (NPUs) and include:
- Text summarization and rewriting
- Image description and recognition
- Super-resolution and image segmentation
Applications from developers such as Dot Vista and Filmora are already incorporating these features to accelerate AI integration.
Semantic Search and Retrieval-Augmented Generation (RAG)
New semantic search APIs enable context-aware and privacy-optimized searches on app data, supporting both semantic and lexical queries. These APIs also facilitate RAG, enhancing large language model outputs with precise, app-specific data.
Model Context Protocol (MCP) for AI Agents
Windows 11 now includes native support for the Model Context Protocol, which allows AI agents to communicate directly with apps and expose specific functionalities securely. MCP features:
- MCP Registry: A secure directory of MCP servers accessible to AI agents.
- MCP Servers: System-level services and developer-defined APIs that expose app features to agents.
Leading AI companies like Anthropic, OpenAI, and Figma are already integrating MCP support.
App Actions: Enhancing Discoverability
App Actions let developers define discrete, contextually surfaced actions within their apps, improving user engagement. Early adopters include Zoom, Todoist, and Goodnotes.
LoRA Fine-Tuning for Phi Silica Models
LoRA support simplifies fine-tuning of the Phi Silica language model by adjusting fewer parameters, making customization more accessible. Currently in public preview for Snapdragon X NPUs, it will soon support Intel and AMD platforms. Developers can perform fine-tuning in VS Code and deploy adapters using the Phi Silica API.
AI-Capable Developer Workstations
Microsoft is collaborating with OEMs to offer AI-focused hardware tailored for local inferencing and compute-intensive workloads. Notable devices include:
- Dell Pro Max Tower for desktop development
- HP Z2 Mini G1a for compact workstation needs
- Copilot+ laptops such as Dell Pro Max 16 Premium and HP Zbook Ultra G1a
Security Enhancements: VBS Enclave SDK and Post-Quantum Cryptography
The VBS Enclave SDK is now in public preview, providing a Trusted Execution Environment for secure data processing, including cryptographic operations. It supports interface definition, parameter validation, enclave lifecycle management, encryption, and telemetry.
Microsoft is also integrating post-quantum cryptography (PQC) into Windows Insider builds and Linux through SymCrypt-OpenSSL. This prepares developers for upcoming cryptographic standards.
Open Sourcing Windows Subsystem for Linux (WSL)
WSL is now open source, allowing developers to contribute to and customize the Linux virtualization layer on Windows. This move aims to boost flexibility and performance, responding to long-standing community requests.
Refinements to Developer Tools
- WinGet Configuration: Now supports DSC V3 for capturing and replicating application states.
- Advanced Windows Settings: Centralizes options such as GitHub integration in File Explorer.
- PowerToys Command Palette: Provides quick access to commands and apps, reducing context switches.
- Edit: A new command-line text editor for fast file edits directly in the terminal.
Microsoft Store: Expanding Developer Support
The Microsoft Store reaches over 250 million monthly users and offers developers features like free account registration, Win32 app FastTrack submission, Beta App Campaigns, and enhanced analytics. The new AI Hub highlights AI-enabled apps with badges and featured experiences, underlining Windows’s commitment to consumer-facing AI tools.
Looking Ahead
Build 2025 highlights Microsoft’s focus on an open, secure, and intelligent Windows platform for AI development. With new APIs, security improvements, hardware partnerships, and developer-centric tools, Windows is set to support the next wave of AI applications and experiences.
For developers seeking to deepen their AI skills on Windows and beyond, exploring targeted AI courses can provide practical guidance on integrating these new capabilities.
Your membership also unlocks:
