AutoGen
Microsoft's framework for multi-agent systems built around asynchronous, conversational message passing.
AutoGen models agents as participants in a conversation: they exchange messages, call tools, and reach a result through dialogue rather than a fixed pipeline. Its redesigned core is event-driven and asynchronous, which suits agents that run long or wait on each other.
It comes from Microsoft Research and ships with AutoGen Studio, a low-code surface for prototyping agent teams before committing them to code.
Where it fits
A fit for research-flavoured, exploratory multi-agent work where conversation between agents is the point.
Where it does not fit
The conversational model adds nondeterminism you do not want on a business-critical path that must run the same way every time.
Alternatives to AutoGen
The closest options on the map, and where each fits instead.
-
LangGraph Graph orchestration
Reach for LangGraph when an agent needs real branching (loops, retries, human-in-the-loop pauses) and you want that control flow written down explicitly.
Unlike AutoGen Maturity: established
-
CrewAI Multi-agent orchestration
Good when a task decomposes cleanly into distinct roles and you want a fast, readable way to express that collaboration.
Unlike AutoGen Hosted API
-
Semantic Kernel Enterprise agent SDK
The natural choice for .NET and enterprise Microsoft shops bringing agents into existing applications.
Unlike AutoGen Maturity: established
Pairs with
Commonly used alongside it.
