Cube Semantic Layer Platform

Cube is a semantic layer platform that serves as a centralized data model between data warehouses and consuming applications including BI tools, AI agents, and embedded analytics interfaces. Data engineers define metrics, dimensions, and joins in JavaScript or YAML data model files that describe how raw database tables and columns map to business concepts like revenue, active users, churn rate, and customer segments. The semantic model enforces consistent definitions across every downstream consumer, so the same metric yields the same number whether queried through a dashboard, a natural-language AI assistant, or a REST API call. The caching layer stores pre-computed query results in Redis with configurable refresh policies, reducing query latency for expensive aggregations against data warehouses like Snowflake, BigQuery, Databricks, and ClickHouse from minutes to milliseconds. The REST and GraphQL APIs expose the semantic model to frontend applications, enabling developers to build custom analytics interfaces without writing SQL or managing database connections. For embedded analytics scenarios, the platform provides iframe-based dashboards and a chat API that renders AI-powered analytics interfaces branded with the embedding application's identity, with multi-tenant permission scoping that controls which data each customer organization can access. Analytics Chat translates natural language questions into validated queries against the semantic model, ensuring that AI-generated answers are grounded in governed metric definitions rather than hallucinated SQL. An MCP server integration lets coding agents and AI development tools query the semantic layer during development, enabling agentic analytics workflows where AI agents build and ship analytics artifacts. The platform has over twenty thousand GitHub stars and serves customers including Tesla, SpaceX, Walmart, Intuit, Webflow, and Brex.