My tokens, your tokens, her tokens and no tokens at all

This is a session for data architects and engineers who are being asked to make architectural decisions in real time as AI tooling proliferates. It takes a position: the current conversation about AI in data engineering overweights the intelligence layer and underweights the foundation it depends on. Better foundations mean better AI, lower inference cost, and stronger governance, and that foundation is a governed context store, models, lineage, and transformation logic held in one place, not scattered across scripts and someone's memory.

Every practitioner already knows context quality determines output quality — this session gives you the vocabulary and trade-offs to argue it with the AI-tooling decisions landing on your desk right now.