Session abstract
What you’ll learn
This session is about building AI agents that actually hold up in production using Mastra and TypeScript. We’ll go beyond simple prompt chains and look at how to design multi-step workflows, enforce structured outputs with schemas and build memory systems that don’t fall apart at scale. We’ll also cover how to connect agents to real tools using MCP, handle failures and retries and design systems that can reliably take action instead of just generating text. The focus is on practical patterns, real tradeoffs and what it takes to move from demos to systems that actually work.