Decoder. plain-English AI glossary

pgvector

▲ Rising

Also called Postgres Vector Extension

A Postgres extension that lets your regular database do vector search, no separate system needed.

Think of it like

Adding a spice rack to the kitchen you already have instead of building a second kitchen.

Example

A team stores embeddings in a new column and runs similarity queries with SQL, keeping vectors right next to their relational data.

How it actually works

pgvector adds a vector column type and similarity operators to PostgreSQL, with HNSW and IVFFlat indexing. Its big appeal is operational simplicity: one database for both structured data and embeddings, with familiar SQL, transactions, and joins. At very large scale or extreme QPS a dedicated vector DB may pull ahead, but for many apps pgvector is more than enough.

For product teams

Often the pragmatic choice — one fewer system to run when you already use Postgres.

For engineers

Vector type + HNSW/IVFFlat operators in Postgres; unifies relational and similarity queries at the cost of specialized scale.

Related

Read anything AI without the jargon

Look up any term in plain English, or save terms as you read with the free Chrome extension.

Open DecoderAdd to Chrome