We kept hitting the same wall: everyone wants AI analytics, but AI without business context hallucinates. The fix is a semantic layer — a model that defines what "revenue" or "churn" actually means. But building one by hand takes weeks.

So we built an AI agent that writes the semantic layer itself, then uses it to answer questions and build dashboards with no hallucinations.

Connect your data → agent builds the model in seconds → ask questions in natural language → get accurate results grounded in your business definitions.

Three months in GA. Running in 200+ companies. 500K+ lines of semantic layer code generated. Teams like Brex and Drata using it in production. OSS core isn't going anywhere.

Agentic analytics is a cloud layer on top. Free tier available. https://cube.dev/

The founders are in the thread to answer your question.