Semantic modeling is the highest-leverage work in analytics — and historically some of the slowest. The new Semantic Model automation task collapses it from weeks to minutes.
How it works
- Point the task at your table schemas.
- Tell it, in plain language, who will use the data and what business questions they need answered.
- The AI drafts a complete semantic model: entities, relationships, joins, and business definitions.
- Review it on a visual entity canvas — drag, edit, refine, approve.
Why it matters
- It runs inside the automation framework, so models can be scheduled, versioned, and re-run like any other job
- Table aliasing and data-warehouse semantic support landed in the same release window
- The output feeds everything downstream: it's the gold layer Databasin One reasons over and the foundation for curated views across the platform
The questions you ask shape the model you get. That's the inversion that matters: instead of modeling the data and hoping it answers the business, you start from the business and model backwards.