AutoGen
Microsoft's framework for building multi-agent conversations where agents play roles and negotiate solutions.
Think of it like
A Slack channel where bots argue and agree their way to a solution—structured debate automation.
Example
Create a 'user proxy' agent, a 'coder' agent, and a 'planner' agent. They chat with each other over token limits until they converge on a plan, code, and tests.
How it actually works
AutoGen treats multi-agent as a conversational protocol. Agents are Python objects with state; they send messages to each other, have conversation history, and can interrupt or agree. It's more stateful than CrewAI. Useful for back-and-forth reasoning but tokenization can explode if agents are verbose.
For product teams
Good for workflows that need negotiation, debate, or internal validation before action.
For engineers
Conversation-first design; built-in state management and tool registry; Python native.
Read anything AI without the jargon
Look up any term in plain English, or save terms as you read with the free Chrome extension.
Open DecoderAdd to Chrome