Building Agentic Systems With PydanticAI

This session is presented by our community partner, IDinsight.

This resource was produced by IDInsight, a trusted partner in AI education. While OpenAI Academy shares it here, it does not necessarily represent OpenAI’s own views. IDinsight is a mission-driven global advisory, data analytics, and research organization that helps global development leaders maximize their social impact. For more information, visit their website at IDinsight.org.

Join IDinsight for a hands-on, high-level walkthrough of agentic system design using PydanticAI, a new open-source library that turns LLMs into structured, tool-enabled agents. This talk is designed for developers, ML engineers, and AI enthusiasts who want to move beyond single-shot prompts and explore how role-based agents can collaborate, critique, and evolve outputs over time.

We’ll cover key concepts like schema-based agent design, RCC (Refinement through Critique and Conversation), and the Mind Evolution loop, showing how you can build systems where agents like Creator, Critic, Judge, and Explorer iterate toward better outcomes. The centerpiece demo will showcase a working multi-agent system that evolves product ideas through an island mutation model, grounded in real-time web search and powered by PydanticAI. You'll leave with a clear mental model, working code references, and a roadmap for building your own agentic workflows.