Vector database
A store optimized for similarity search over embedding vectors: the retrieval layer behind many RAG and semantic-search systems.
A vector database indexes high-dimensional embeddings produced by an LLM or dedicated encoder at inference time. Given a query embedding, it returns nearest neighbors (chunks, documents, images) by cosine distance or related metrics.
It is the retrieval backbone of many RAG setups: ingest text, embed offline or on write, search at query time, inject hits into the prompt. Quality depends on chunking, the embedding model, and freshness; it complements but does not replace training or fine-tuning the generator.