pgvector is the PostgreSQL extension that adds a vector data type and approximate-nearest-neighbour search, letting AGNT store embeddings next to relational rows in one ACID transaction.
Last verified: April 2026
What is pgvector?
pgvector is the PostgreSQL extension that adds a vector data type and approximate-nearest-neighbour search, letting AGNT store embeddings next to relational rows in one ACID transaction.
AGNT uses pgvector 0.3 with 1536-dimensional vectors produced by OpenAI's text-embedding-3-small. Semantic search runs as a cosine distance query against IVFFlat or HNSW indexes, returning sub-hundred-millisecond results at the venue graph's current size. Because the extension lives inside the primary database, there is no second-system replication lag and no separate consistency model to reason about.
FAQ
pgvector FAQ.
Common questions about pgvector in the AGNT platform.
pgvector is the PostgreSQL extension that adds a vector data type and approximate-nearest-neighbour search, letting AGNT store embeddings next to relational rows in one ACID transaction.
People also ask.
Related terms
Semantic Recall
Semantic recall is the embedding-based retrieval step that finds memories relevant to the current message, even when the wording has changed since the original fact was stored.
User Memory
User memory is AGNT's per-user fact store: key-value pairs with confidence scores and pgvector embeddings, split into structural keys that always enter the prompt and semantic keys recalled on demand.
Knowledge Pack
A knowledge pack is a venue's documentation bundle — menus, policies, hours, FAQs — ingested and chunked into VenueChunk records with pgvector embeddings for retrieval-augmented generation.
Venue Graph
The Venue Graph is the connected database of every restaurant, bar, gym, and spa on the AGNT network, linking rows, embeddings, and relationships in a single PostgreSQL cluster.
Authority sources
Source: agnt-backend/requirements.txt
Share as social post
What is pgvector? pgvector is the PostgreSQL extension that adds a vector data type and approximate-nearest-neighbour search, letting AGNT store embeddings next to relational rows in one ACID transaction. https://agntdot.com/glossary/pgvector
243 / 280 chars
See it in action.
Now you know what pgvector means. Try the live scan demo or read the developer docs to go deeper.