AutoGen

What is AutoGen?

AutoGen is an open-source framework by Microsoft Research for building next-generation LLM applications using multi-agent conversations. Agents can act as assistants, critics, code executors, or domain experts — and they talk to each other to solve tasks.

Core Concept

Instead of one LLM doing everything, you define a team of agents with different roles and instructions. They converse in a structured way until the task is complete.

Example team:

• UserProxy — represents the human, can execute code
• AssistantAgent — writes Python code to solve a problem
• CriticAgent — reviews the code for bugs or improvements

Key Features

• ConversableAgent — base class for all agents; configure name, system prompt, LLM backend.
• GroupChat — orchestrate conversations between 3+ agents.
• Code execution — agents can write and run code in a sandbox.
• Human-in-the-loop — pause and ask the human for input at any point.
• Works with any LLM — OpenAI, Azure, Anthropic, local models.

Getting Started

pip install pyautogen

AutoGen Studio also provides a no-code GUI for building and testing multi-agent workflows without writing Python.