All integrations
API PlatformsLive
OpenAI API
Text embeddings for semantic venue search and agent memory recall.
Official siteOverview
AGNT uses OpenAI's text-embedding-3-small model to generate 1536-dimensional vectors for venue descriptions, reviews, and agent memory entries. These embeddings are stored in PostgreSQL via pgvector alongside the relational data.
The embedding pipeline runs at ingestion time for venues and at query time for user messages, enabling sub-100ms approximate nearest neighbor searches across the entire venue graph and memory store.
How it works
AGNT calls the OpenAI embeddings endpoint at ingestion and query time, storing resulting vectors in pgvector for semantic search.
Integration status
Live— actively used in production