AutoGen
AutoGen is Microsoft's open-source framework for building multi-agent AI applications, where agents converse and collaborate to solve tasks. Supports tool use, human-in-the-loop, and a layered architecture (Core, AgentChat, Extensions).
Guidance
Guidance is Microsoft's library for controlling large language models with structured generation. It enables interleaving generation, prompting, and logical control into a single continuous flow, uses context-free grammar (CFG) to guarantee valid output formats like JSON, and achieves 50% faster inference through token healing and prompt caching. Supports all major LLM providers with a simple, intuitive syntax.
ONNX Runtime
ONNX Runtime is Microsoft's cross-platform engine for high-performance inference and training of ONNX models, with hardware acceleration across CPU, GPU, and NPUs and a dedicated generative-AI path for LLMs.
Playwright MCP
Microsoft's Model Context Protocol server that gives AI agents structured, accessibility-tree-based control of a browser via Playwright - navigating pages, clicking, and extracting content without screenshots or pixel guessing.
Semantic Kernel
Semantic Kernel is Microsoft's open-source SDK for integrating LLMs into applications. It supports C#, Python, and Java with features like planners, plugins, and memory management.