WrenAI

Why we included this project

WrenAI is for teams that want to query their existing data warehouse in plain language without exposing raw schemas to an LLM. It sits between your databases and whatever agent or application is asking questions, backed by a semantic layer that captures business definitions, approved join paths, and column meanings in versionable MDL files. That context is what lets a plain question become governed SQL and charts you can share as dashboards, across sources like BigQuery, Snowflake, PostgreSQL, and ClickHouse. Data and analytics engineers will find it useful for giving non-technical colleagues or AI agents a trustworthy way to ask business questions while keeping control over what the model can see and compute. Since the context layer is open and Git-friendly, it can also serve as a standalone foundation for other agentic analytics surfaces beyond the bundled BI app.