Skip to navigation
Use cases

Retrieval for RAG

Talqora Vector is the retrieval layer for a RAG application. Your ingestion service splits source documents, embeds each chunk, and writes chunks together with the source URL, title, and access metadata.

At answer time:

  1. Embed the user question.
  2. Query an index with dense or hybrid search.
  3. Apply access and tenant metadata filters.
  4. Pass the returned chunks and citations to your model.

The vector API does not generate answers. It gives the application a predictable retrieval contract and observable usage data.