AutoGen is Microsoft's open-source framework for creating multi-agent AI systems where multiple LLM-backed agents collaborate via asynchronous messaging to complete complex tasks. As of 2026, AutoGen is in maintenance mode and no longer receives new features. Microsoft now directs new users to the Microsoft Agent Framework (MAF), which combines AutoGen's agent abstractions with Semantic Kernel's enterprise features including session-based state management, type safety, middleware, telemetry, and graph-based multi-agent orchestration. Key features: - Asynchronous agent messaging supporting event-driven and request/response patterns - AutoGen Studio no-code GUI for building multi-agent workflows - Pluggable components: custom agents, tools, memory, and models - Support for human-in-the-loop interaction patterns - Compatible with OpenAI, Anthropic, Azure, and other LLM providers - Migration path to Microsoft Agent Framework (MAF) successor
Free and open source (MIT license). LLM API costs (OpenAI, Anthropic, etc.) are the primary ongoing expense.