AutoGen: Enabling Next Gen AI Applications via Multi Agent Conversation

In this session, Qingyun Wu provides a deep dive into the key concepts of AutoGen, a programming framework for agentic AI. She demonstrates diverse applications enabled by AutoGen and shares the latest updates and ongoing efforts. AutoGen enables the development of AI agentic applications using multiple agents that can converse with each other to solve tasks. The talk covers agent abstraction and multi-agent orchestration, customizability of agents, and various conversation patterns like two-agent chat, sequential chat, N-chat, and group chat. Advanced topics such as React prompting, Retrieval Augmented Generation (RAG), memory management, and more are also discussed. The presentation highlights use cases in science, engineering, and industry, and outlines the project's future outlook including optimization capabilities and automated agent construction (AutoBuild).