Building AI Applications the Pydantic Way (Sponsor: Pydantic)

AI is becoming an increasingly common part of software development, but best practices for building and maintaining AI-powered applications are still evolving. Many developers are experimenting with LLMs, but figuring out how to structure AI applications effectively—let alone debug and monitor them—can be a challenge. The good news? AI development is just software development—with a few unique twists. You don't need to reinvent engineering principles to work with LLMs, but you do need the right design patterns and development tools to build reliable AI applications efficiently. In this talk, we'll share an opinionated blueprint for AI development in Python, covering both PydanticAI, an open-source AI agent framework, and Pydantic Logfire, an observability platform that helps you understand and debug your AI applications. You'll learn practical techniques for structuring LLM-powered systems, tracking and evaluating model behavior, and integrating AI features into larger software projects—using familiar tools and proven patterns. Whether you're already building AI applications or just curious about how best to approach them, this session will give you a clear, pragmatic foundation for writing AI-powered software in Python.