Vector database
A database specialized for storing and searching vector embeddings — designed to answer 'find the N vectors closest to this query vector' very fast, at scale.
A vector database stores embeddings and lets you query for nearest-neighbor matches at scale. The operation "find the 10 embeddings closest to this one" is trivial when you have a few thousand embeddings; it becomes non-trivial when you have millions or billions, which is where a vector database earns its keep.
The technical trick most of them use is approximate nearest-neighbor search — HNSW (hierarchical navigable small world) is the most common algorithm — which sacrifices small amounts of precision for orders of magnitude of speed improvement. For most RAG use cases, "the 10th-nearest instead of the actually-10th-nearest" is acceptable.
Popular options in 2026: Pinecone (managed, easy to start), pgvector (Postgres extension, good if you already run Postgres), Weaviate (open source, strong), Turbopuffer (newer, aggressively optimized for cost), Qdrant (open source, fast). Which one is right depends on scale, budget, and whether you want a separate service or an extension of your existing DB.