Pointing an LLM at raw data gets you interesting demos. Pointing one at a well-governed semantic layer gets you answers that stakeholders trust.

Learn how Gousto moved from a centralised monorepo to a domain-owned data mesh with the dbt platform and cut analyst onboarding from five hours to 15 minutes. We’ll also cover how a well-structured dbt semantic layer and Databricks Genie became the foundation for our early AI use cases, and we’ll share the framework for measuring whether AI self-serve is actually sticking.

You'll also learn:

• Why context beats raw data for grounding LLMs
• An approach to piloting AI self-serve without overpromising
• The signals that tell you a use case is ready to scale