wtf is autogen?

AutoGen is a framework for building applications where multiple AI agents work together to accomplish tasks. Instead of having a single AI assistant answer a question, you can set up a team of specialized agents, for example, one that's an expert at math, another that knows chemistry, and a general coordinator that routes questions to the right specialist. This approach lets you tackle more complex problems than a single AI model could handle on its own. At a high level, you create individual agents, each with a specific role and set of instructions. You then connect them so they can hand work back and forth or call on each other when needed. The framework handles the plumbing, message passing between agents, connecting to AI models like OpenAI's GPT, and integrating external tools such as web browsers. A no-code visual interface called AutoGen Studio lets you prototype these multi-agent workflows without writing code, which is useful for quickly sketching out ideas before committing to a full implementation. This would appeal to builders creating sophisticated AI assistants, say, a customer support system where one agent triages incoming questions, another looks up account details, and a third drafts responses. Or a research tool where agents browse the web, execute code, and summarize findings. The project highlights Magentic-One, a multi-agent team that can browse the web, run code, and manage files, as an example of what's possible. One important caveat: AutoGen is now in maintenance mode. It will not receive new features, and the project's creators recommend that anyone starting a new project use Microsoft Agent Framework instead, which they describe as the enterprise-ready successor. Existing users can continue with AutoGen, and migration guidance is available for those who want to switch. The framework is built in layers, from low-level message-passing components up to higher-level APIs for rapid prototyping, which lets developers work at whatever level of abstraction suits them.