Platform module

Embeddings

Opt-in text-to-vector generation and text-in semantic search.

What you can build

Key capabilities for embeddings — explore the docs for setup and examples.

Point source string fields at a Vector target. The first upsert can provision the index; keep the config disabled until it is queryable.

Queue bounded onlyMissing runs. Cancel and resume from a stored cursor — never scan in the request thread.

Client POST /embeddings/search takes schemaName and text. Raw vectors stay on Database Admin vector-search.

Modules teams often enable alongside embeddings.

Flexible data models on MongoDB or PostgreSQL — CRUD, relationships, and powerful queries in one interface.

Explore the full module catalog or jump into the docs.