What It Takes to Make RAG Work in the Real World

Cake CTO Skyler Thomas shares an unfiltered look at what it takes to move RAG (retrieval-augmented generation) from prototype to production, especially in complex, high-stakes environments. This isn’t about small-scale demos; it’s about what teams face when GenAI becomes business-critical. From why most projects stall out, to how to structure evaluation, orchestration, and observability for scale, Sky lays out practical insights, hard-earned lessons, and the tooling patterns that actually make a difference. Whether you're a builder or decision-maker, this talk demystifies the work behind successful, enterprise-grade AI systems.