Vector Database
A specialized data system purpose-built to efficiently store, index, and query massive collections of high-dimensional embeddings for semantic search.
Think of It Like This
Like an impossibly massive library organized not by alphabet, but by grouping books whose plots, themes, and feelings are geometrically close together.
Unlike relational databases, vector DBs like Pinecone or Milvus execute similarity searches using metrics like Cosine Similarity or Euclidean Distance. Because comparing a query to billions of dense vectors is computationally intractable, they rely on Approximate Nearest Neighbor (ANN) indexes like HNSW for blazing-fast retrieval at scale.