Database extension · Established · Intermediate
pgvector
A PostgreSQL extension that adds vector storage and similarity search to a standard relational database.
What pgvector is
pgvector lets teams keep embeddings alongside their existing relational data instead of running a separate vector service.
How it works
A vector column type stores embeddings, with index types for approximate search and normal SQL for metadata filtering and joins.
Why it matters
For most applications below very large scale, it removes an entire piece of infrastructure while keeping transactional consistency.
Common uses
- →RAG on existing Postgres data
- →Semantic search in web apps
- →Hybrid keyword and vector queries
Strengths
- ✓One database instead of two
- ✓Full SQL filtering and joins
Watch for
- ✓Tuning needed at large scale
- ✓Fewer specialised retrieval features
Continue exploring
More in this collection
Browse all AI Technology