Skip to content

Retrieval Augmented Generation (RAG)

Versa uses RAG as a key component to improve the quality of LLM responses. It does this by:

  • Storing business information in a database using structured schemas with Pydantic
  • Indexing and embedding data for efficient retrieval
  • Using hybrid search that combine FTS and vector search
  • Reranking responses using a Cohere reranker