
Microsoft Research's open-source multi-agent framework, now archived: Microsoft no longer fixes, secures or supports it and points users to Microsoft Agent Framework.
AutoGen models AI work as a conversation between agents — a coder, a reviewer, a planner — that message each other until a task is done. It is research-grade and fast-moving, with AgentChat for high-level patterns and a lower-level Core API for custom orchestration. For enterprise teams it is a credible way to prototype multi-agent designs, but the usual warning applies hard here: conversational multi-agent systems are difficult to evaluate, can loop, and burn tokens unpredictably. Treat AutoGen builds as experiments until you have evals and cost controls around them. AutoGen is now in maintenance mode per its README: it will not receive new features, and Microsoft points new projects to Microsoft Agent Framework, its successor.
Get a new AI workflow each week — many feature Microsoft AutoGen and other tools in this category.