It is more efficient to bring logic to the data than to ship data to the logic. This is especially true for "real time" RAG scenarios where OLTP data needs to be searched and processed in real time. Azure Postgres now has the ability to generate vector embeddings in real-time within the database. Check it out! #postgres #azure #ai #copilot #rag #llm #postgresql
Today at Build, we announced the preview of in-database embedding generation for Azure Database for PostgreSQL which enables you to build blazing fast AI apps. This feature leverages open source embedding models running locally in the Postgres server and accessible through a Postgres extension. In-database embedding creation time is typically under 10ms to provide true OLTP performance for your AI apps. You can learn more in the blog from Josh below: #postgres #postgresql #oss #opensource #ai #MSBuild