Vector Database — AI Glossary

Reference details

Related terms

• Embedding
• Semantic Search
• HNSW (Hierarchical Navigable Small World)
• RAG (Retrieval-Augmented Generation)

Example: Approximate, and that is the point

Exact nearest-neighbour over 10 million vectors means 10 million comparisons per query. An HNSW index navigates a layered graph and returns very likely neighbours in single-digit milliseconds. You are trading a small amount of recall for orders of magnitude in speed — and the recall you lose is tunable, not fixed.

Commonly confused with

A vector database is not a replacement for your primary store. It holds embeddings and metadata for similarity search; it is not where records of truth live. Most production systems run both and keep the vector index derived from, and reconcilable with, the source of record.

When to use it

Reach for it when:

• Similarity search over a corpus too large to scan per query
• You need metadata filters combined with vector search
• The index changes often enough to need incremental updates

Reach for something else when:

• The corpus is small — a flat in-memory index is simpler and exact
• The query is lexical, where an inverted index outperforms
• You already run Postgres and pgvector would meet the need without new infrastructure

Referenced by

This term is used by the following ModelRefs references:

• vector search tool
• naive rag stack
• learn rag

Continue your research

Use these connected ModelRefs sections to compare alternatives, inspect implementation paths, and review the evidence and governance boundaries relevant to Vector Database — AI Glossary.

• Browse models
• Review providers
• Inspect benchmarks
• Explore workflows
• Learn AI concepts
• Read decision guides
• Review trust and methodology
• Start a decision path