AutoGen Framework

Why This Project Was Stopped

This multi-agent framework attempted to solve orchestration when the real problem is tool quality.

• Architecture became the product - Spent more time on agent coordination than making agents useful
• Complexity doesn't compound value - Multiple agents communicating adds failure points, not capabilities
• Claude Code proved the opposite works better - One capable agent + sub agent for specific tasks + native tools > many agents + orchestration layer
• Infrastructure is overhead - Kafka, Redis, Kubernetes, service mesh don't make AI smarter

The lesson: Better tools in one agent's hands beats complex agent choreography. Build capabilities, not coordination.

Prerequisites

• Python 3.10 or later

Virtual Environment Setup (Recommended)

This document records the setup steps for AutoGen based on the official installation guide.

Option 1: Using venv

# Create virtual environment
python3 -m venv .venv

# Activate environment
source .venv/bin/activate    # Linux/Mac
.venv\Scripts\activate.bat   # Windows

# Deactivate when done
deactivate

Installation

Core Package

pip install -U "autogen-agentchat"

OpenAI Model Client Extensions

pip install "autogen-ext[openai]"

Azure OpenAI Support (Optional)

pip install "autogen-ext[azure]"