Home / Glossary / Retrieval
What is Retrieval?
Retrieval is the step in RAG where relevant documents or chunks are pulled from a knowledge store based on the user’s query. osFoundry combines vector embedding search with BM25 keyword search and optional reranking.
Detail
Retrieval is typically implemented as: take the user query, embed it into a vector, search a vector index for nearest neighbours, optionally combine with BM25 for keyword matches, return top-k. Quality depends on the embedding model, the chunking strategy, and the reranker.
Hybrid retrieval (vector + keyword) tends to beat pure vector — keywords catch acronyms and specific terms that embeddings sometimes miss.
How osFoundry approaches Retrieval
osFoundry’s default pipeline uses Voyage embeddings + tantivy BM25 + Voyage reranker. Every stage is configurable in osStudio.
Related terms
Related features