What does it really take to move from a RAG chatbot to an agentic AI system? In this talk, I’ll share the technical and organizational lessons from our migration: the importance of framework choice, common design patterns for building AI agents, the hidden blockers in data foundations, and how we approach evaluation using Langfuse. Through live demos, I’ll illustrate key challenges and design decisions that shaped our system. Beyond the technical aspects, we’ll explore the evolving skill sets for AI teams, the impact of AI-assisted coding, and practical ways to measure real business impact.