Data Engineering Is AI Engineering

​"Context engineering" is everywhere right now. But for data agents, most teams are solving it wrong. They're reaching for RAG pipelines, multi-agent frameworks, and semantic layers when the highest-leverage fix is something data teams already know how to do.

Jacob Matson spent months running controlled experiments on 460 real business questions. The results were surprising: well-modeled schemas with clear naming outperformed expensive documentation, retrieval systems, and complex agent architectures at a fraction of the cost. Medium reasoning beat high reasoning. Simpler setups beat elaborate ones. And when schema and documentation disagreed, accuracy got worse, not better.

Join Jacob for a frank look at what actually makes AI agents work on real data and why the most important AI skill might be one data engineers have had all along.