Building Multi-Agent Generative-AI Applications with AutoGen
summary
Shows AutoGen open-source Python framework wiring a chat manager, assistant and user-proxy so agents coordinate to decompose planning tasks, call tools, persist memory (Chromadb by default) and self-refine by writing and running code. Demos build a travel-booking flow and a stock-plot pipeline that largely works but trips on a stale Gandam Factory booking. Integrations include LlamaIndex LangChain and multimodal models (GPT‑4V); cost/latency and a Gaia benchmark note that success drops on level‑3 reasoning tasks.
AI-written summary · about
chapters (9)
- 0:00AutoGen and multi-agent systems
- 3:16Complex tasks need multi-agent AI
- 9:21Domain-driven conversational agents
- 14:50Limitations of single LLM
- 16:57Multi-agent productivity gains
- 20:33Why multi-agent AI now
- 23:03Building with AutoGen
- 26:58Live demo: travel agent
- 31:12AutoGen ecosystem and challenges
related talks
33mAutoGen: A Multi-Agent AI Framework
45mBuild Agentic AI Apps with the Autogen framework
10mKeynote: Next-Gen AI Applications via Multi-Agent Conversation- 10mKeynote: Next-Gen AI Applications via Multi-Agent Conversation
25mFrontiers of Agentic AI
20mGenAI agents - Evolution of automation
16mBuilding Applications with AI Agents
36mGenerative AI + Python: Unlocking Efficiency, Personalization, and Insight