The semantic layer for embedded AI analytics

What teams ask us most about governance, AI, and the tools they already use

• Does Luzmo have a semantic layer?+

  Yes. Luzmo covers the five capabilities the market means by the term: governed metric definitions through formulas, derived columns and SQL views; AI grounded in the model rather than freeform SQL; multi-tenant security enforced on every surface including AI; sub-second performance through Rust push-down connectors and Warp, our data acceleration layer; and full openness through a documented API, MCP server and Agent API.

• Do I still need dbt or Cube if I use Luzmo?+

  Keep them. They stay where they are, and Luzmo does not read their definitions: there is no dbt or Cube connector, so the metrics you want Luzmo to serve are defined in Luzmo, as formulas, derived columns or SQL views. Luzmo connects to the data source underneath, so the tables dbt builds are the tables Luzmo queries.

• Can I bring my existing metric definitions?+

  Not as a live connection, and this is the part worth being blunt about. You re-express them in Luzmo, as formulas, derived columns or SQL views. There is no sync from an external semantic layer, so a definition that changes in dbt does not change in Luzmo. What keeps that from being manual work is the API: definitions can be created and maintained programmatically, so they can be scripted from whatever already holds them.

• Is the AI writing raw SQL against my database?+

  No. Every question is resolved by the Luzmo query engine, which applies your data model, table relationships, column metadata and the requesting user's permissions. The model selects from governed structures rather than authoring queries against your database.

• How does Luzmo compare with other semantic layer tools?+

  Tools like dbt, Cube, AtScale and Dremio are data infrastructure. They govern definitions for your own analysts. Luzmo governs them inside the product you ship to your customers, which is why multi-tenant scope applies to dashboards, the API and AI answers alike.

• How does this apply per customer in a multi-tenant product?+

  Tenant scope is enforced by the embed token on every surface, including AI answers, so each of your customers sees only their own rows through exactly the same governed definitions. The definitions themselves stay shared, which is what keeps the numbers consistent from one customer to the next.