Cube’s semantic layer as the shared surface for BI, embed, and AI agents

Cube targets internal business intelligence and embedded analytics on one product. Data teams maintain semantic models in code; analysts and business users explore via workbooks, dashboards, and natural-language Analytics Chat grounded in that model. AI agents connect through an MCP server or Chat API to the same governed model. Cube Core, the open-source semantic layer, centralizes metric definitions, joins, access rules, and caching upstream of every consumer; models port between Cube Core and the commercial product. Without a semantic layer, agents writing SQL against the warehouse get inconsistent metrics and ungoverned access. With one, they reason over a stable surface. Semantic SQL is a Postgres-compatible interface with a MEASURE function; queries are validated and access policies applied in the semantic layer runtime.

Analytics and AI teams need one metric vocabulary for dashboards and agents. Embedding the same APIs in customer products keeps definitions aligned with internal BI.

Governed shared meaning sits in the semantic model: metrics and access rules every human and agent tool must pass through before hitting the warehouse.