Building a production AI agent system with Pydantic

​Pydantic AI now has native Temporal support, bringing durable execution to type-safe AI agents. Join Douwe Maan & David Montague from Pydantic and Shy Ruparel from Temporal for a live coding session where we'll build a real AI agent system from scratch—showing you how to create agents that survive API failures, handle long-running workflows, and coordinate multiple agents without manual checkpointing or fragile retry logic.

​We'll demonstrate the new integration through practical examples: handling human-in-the-loop approvals, and managing complex workflows that preserve state across failures and restarts. You'll see how combining Pydantic AI's type safety and built-in observability with Temporal's replay-based fault tolerance gives you production-grade AI applications with minimal complexity.

​Whether you're building customer support agents, research assistants, or approval workflows, this session will show you patterns you can use immediately. Bring your questions about scaling AI systems, debugging workflows, and building reliable agents.