Domain Modeling in the Age of AI: Why the Semantic Layer Is the New Bottleneck

The hardest problem in scaling a data platform is not choosing between Kafka and Kinesis or picking a lakehouse format. It's getting six teams to agree on what a "customer" is — and whether that's the same thing as a user, a consumer, or an account.

This talk draws on domain modeling work at Zillow, where multiple acquired brands produce overlapping data: same concepts with different names, same names with different meanings. Dashboards undercounted because teams named the same action differently. Schemas broke silently across team boundaries. Data dictionaries grew stale within weeks.

Tooling helps — catalogs, registries, and governance dashboards all have their place. But none of them create agreement. The real needle-mover was the unglamorous discipline of getting producers and consumers in a room and deciding what words mean before anything gets built. Tooling augments that agreement; it doesn't replace it.

This talk covers what that looks like in practice: how to identify the entities worth aligning on, how to build a shared vocabulary that teams actually adopt, and how AI changes the stakes — amplifying whatever foundation you have, whether it's solid or chaotic.

What you'll take away:

• A practical framework for identifying and resolving naming conflicts across teams
• Which standardization decisions are worth investing in — and which ones to
  let go
• How to make the case for domain modeling when leadership wants features, not foundations
• A realistic picture of where AI helps with this work and where it gives a false sense of progress