gel

Why we included this project

Gel is a graph-relational database that runs on top of Postgres, and it is one of the few data stores where AI support feels like part of the product rather than an add-on. Its schema is built around object types and links instead of foreign keys, queries use a language called EdgeQL, and a built-in migration system takes much of the pain out of changing your data model. The optional AI extension is the real draw: you mark fields in your schema, Gel handles the embedding pipeline, and it exposes a retrieval-augmented generation endpoint that works with OpenAI, Anthropic, or Mistral. A small team can get a vector-backed RAG setup running without wiring together a separate embedding service, vector index, and orchestration layer. If you want one platform for both your relational application data and AI retrieval, with the option to export back to plain Postgres later, Gel is worth a close look.