semlayer

The open-source semantic layer that infers itself — skip the quarter of hand-writing dbt YAML.

Point it at your warehouse. It profiles every column, discovers the foreign keys nobody declared, decodes the status columns, and recovers the business rules hiding in your aggregate tables — verified against your own data, with confidence, provenance, and a review queue on every claim. Then any AI agent consumes the result over MCP.

On our messy-warehouse benchmark, an agent answers 53% of business questions correctly vs 34% from the raw schema (+54% relative). On clean, well-named schemas it doesn't help — we publish that negative result too. The full eval suite ships in the repo.

pip install -e ".[warehouses]"      # Snowflake, BigQuery, DuckDB (+ Iceberg)
semlayer infer snowflake -o layer.yaml --context ./docs/
semlayer review layer.yaml           # accept/reject what it inferred
semlayer mcp layer.yaml              # serve it to Claude, Cursor, any agent

GitHub → Read the benchmark

Free & complete: Apache-2.0 CLI, runs in your environment with your keys — no caps, no signup. A paid hosted control plane (continuous re-inference, team review, dashboards) comes later; the CLI stays fully functional.
Honest by construction: every claim carries its evidence; statistics alone never auto-accept a foreign key; doc-vs-data contradictions go to review, never silently believed.
Design partners wanted: got a messy warehouse and an AI initiative? hello@semlayer.dev