AutoGen — Multi-Agent Conversation Framework

Core Concepts

Conversable Agents

Every agent can send and receive messages, forming conversations:

assistant = AssistantAgent("coder", model_client=model)
reviewer = AssistantAgent("reviewer", model_client=model,
    system_message="Review code for bugs and suggest improvements.")

Multi-Agent Teams

Agents work together in teams with defined conversation patterns:

from autogen_agentchat.teams import RoundRobinGroupChat
team = RoundRobinGroupChat([coder, reviewer, tester])
result = await team.run(task="Build a REST API for todos")

Code Execution

Agents can write and execute code in sandboxed environments.

Human-in-the-Loop

Insert human approval at any point in the agent conversation.

AutoGen Studio

Visual UI for building and testing multi-agent workflows without code.

FAQ

Q: What is AutoGen? A: Microsoft's open-source multi-agent conversation framework. Agents collaborate through chat to solve complex tasks with tool use, code execution, and human feedback. 56K+ GitHub stars.

Q: How is AutoGen different from CrewAI? A: AutoGen focuses on conversational patterns between agents (chat-based collaboration), while CrewAI uses role-based task delegation. AutoGen also includes code execution and AutoGen Studio for visual building.