What's going on with AutoGen and AG2?

AutoGen has split into two paths: AG2 and Microsoft's version. Learn what this means for your AI projects and which path might be right for you.

If you've been working with AutoGen or are planning to, there's a major shake-up you need to know about. The framework just split into two paths, and this might affect your AI projects going forward.

Let me break this down for you.

đź’ˇ
Follow this URL to listen to my take on the latest AutoGen / AG2 updates on my YouTube channel.

What's Happening?

The earliest contributors and creators of AutoGen have moved away from the official Microsoft repo and rebranded their version as AG2. This isn't a new framework - it's basically AutoGen 0.2.34 continuing under a new name, now at version 0.3.2. Their goal? Keep it community-driven and maintain the architecture you're familiar with.

Meanwhile, Microsoft is taking AutoGen in a different direction. They're maintaining version 0.2 while working on a complete rewrite with version 0.4. This new version looks promising, with features like:

  • TypeScript support
  • More built-in agents and workflows
  • Distributed agent deployment
  • AutoGen Studio improvements
  • Advanced RAG techniques
  • Better integration with other frameworks

The Two Paths Forward

Here's what this means for AutoGen developers:

AG2 Path

  • Continues the current AutoGen architecture you know
  • Available through pyautogen, autogen, or ag2 packages
  • Full backward compatibility
  • No breaking changes planned
  • Community-driven development

Microsoft's Path

  • Maintaining AutoGen 0.2 for now
  • Complete rewrite coming with version 0.4
  • Deeper integration with Microsoft's AI ecosystem
  • Potential merger with Semantic Kernel in 2025
  • Significant architectural changes ahead

What Should You Do?

If you're running AutoGen in production:

  • Sticking with AG2? Your code is safe, no immediate action needed.
  • Sticking with Microsoft 0.2? Plan for potential migration work when 0.4 lands.

Starting a new project? Consider this:

  • Need stability and familiar architecture? Go with AG2.
  • Want cutting-edge features and Microsoft ecosystem integration? Wait for 0.4.

My Take

Both paths make sense for different use cases. AG2 offers stability and community-driven development, while Microsoft's version promises exciting new features but with some migration work ahead.

The choice really depends on your needs.

Want to explore more? Check out both repositories and see which path aligns better with your project goals.

I'd love to hear your thoughts on this split and which path you're choosing.

-

Further readings

More from Getting Started with AI

AutoGen - Getting Started with Artificial Intelligence
AutoGen is an open-source framework that is based on research from teams at Microsoft Research, Penn State University, and the University of Washington. The framework enables building AI agents that communicate with each other to accomplish a given task.
Let’s compare AutoGen, crewAI, LangGraph and OpenAI Swarm
The more options we have, the harder the decision to take. Luckily, we’re going to explore the most popular AI multi-agent frameworks in this post so you can make the right choice!
How to choose the best AI models, frameworks, and tools?
Let’s face it, AI is trending. The space is overcrowded and choosing from the many tools and frameworks IS confusing. Read this to simplify your decision.
AutoGen Conversation Patterns - Overview for Beginners
AI Agents are cool. They can solve a lot of problems in a few minutes or seconds. That’s assuming you set up the right conversation patterns in your application so that your agents work together not against each other.
AutoGen vs crewAI - Which framework should you choose
So, you’re wondering which framework to choose for your agentic app and you’ve come across AutoGen and CrewAI. Both are popular, but which one is more suitable for you?
AutoGen Agents - Building agentic apps for beginners
Building agentic apps is not a complicated task! Let’s see how this framework from Microsoft approaches multi-agent conversations and allows you to get started in no time!
Installing and exploring AutoGen Studio 2.0
Let’s explore AutoGen Studio 2.0, the web application that lets us easily manage AI Agents, their skills, and how they collaborate to complete a task.

More from the Web