pg_vectorize turns any Postgres instance into a vector database by automating embedding generation and orchestration via SentenceTransformers, with hooks into popular LLMs. It offers two deployment modes: a standalone HTTP server with REST API and a native Postgres extension with SQL functions like vectorize.table() and vectorize.search(). The system handles maintenance and background workers through pgmq, enabling rapid development of retrieval-augmented generation pipelines and search engines on Postgres.