MindsDB Query Engine
Self-hosted query engine that joins live data sources and searches documents through SQL, with Docker deployment and MySQL or PostgreSQL clients.
MindsDB Query Engine is a self-hosted server for developers building AI agents that need access to business data and documents. It queries connected sources through one SQL dialect and adds semantic search for unstructured content. You can run it locally or on your own server with Docker. Its core uses the Elastic License 2.0, with separate licenses in some directories.
Live queries leave data in the connected source rather than requiring a separate import pipeline. The engine can join records across PostgreSQL, MongoDB and other systems, including warehouses such as Snowflake and BigQuery. Connections also cover Slack, S3, files and web content.
Document search uses knowledge bases that index content with an embedding model and a vector store such as pgvector. You can search by meaning, match keywords, or combine both, then narrow results with metadata filters. Hybrid retrieval combines vector similarity with BM25 keyword matching, useful when a question includes an exact error code or acronym. Knowledge bases can also use a reranking model. OpenAI's text-embedding-3-large is a supported embedding choice; that setup uses a cloud API even though the query engine runs on your hardware.
A built-in browser SQL editor and compatibility with MySQL and PostgreSQL clients let teams use DBeaver, SQLAlchemy and BI tools. It can also supply data to MindsHub agents. Saved views retain queries across sources, while scheduled SQL jobs can refresh knowledge base indexes.
