Build Multi-Agent Architectures

Pablo Salvador Lopez, Principal AI Architect at Microsoft, dives into building and productionizing multi-agent architectures using Semantic Kernel and AutoGen. The talk demystifies agents, explains their core components like perception, planning, and action, and explores how LLMs can be leveraged for more complex reasoning. It covers single-agent patterns, the evolution from RAG to reasoning and tool utilization, and delves into multi-agent paradigms. The presentation highlights the practical applications of these architectures, particularly in areas like content generation and healthcare claims processing, and introduces key frameworks like AutoGen and Semantic Kernel for enterprise-ready AI solutions.