AutoGen is a Python-based, open-source multi-agent framework from Microsoft Research that enables the orchestration of LLM-powered conversational agents. Agents are defined as ConversableAgent subclasses and can be configured with system prompts, tool/function-calling capabilities, code execution environments, and custom reply functions. It supports synchronous and asynchronous agent conversations, group chat topologies, and integrates with OpenAI, Azure OpenAI, and other LLM providers via a unified config interface. AutoGen v0.4+ introduces an actor-model runtime for scalable, distributed agentic applications.