AI Documentation
From 30,000 undocumented columns to a populated catalog — using the LLM you already trust, with nothing sent to a vendor control plane.
- Reads schema, sample values, lineage neighbours and query context through the DataHub API
- Bring your own LLM: any OpenAI-compatible endpoint — Azure OpenAI, AWS Bedrock, Google Vertex, or vLLM/Ollama you host. Model choice, contract and cost stay yours
- Writes descriptions into native DataHub fields, so they render in stock DataHub
- Review queue: descriptions land as proposals, and governance signs off before publication
- Batch backfill plus event-triggered runs for newly ingested assets
Built as: An external service on the GraphQL/OpenAPI surface and the Actions framework.