Enterprise Platform
Bring AI to Your Data
MariaDB grounds your AI applications in your own documents and databases through a managed retrieval layer, so you ship accurate, agentic apps faster without building your own retrieval stack.
Grounded in your own data
MariaDB grounds your AI applications in your business data. Retrieve from unstructured documents through a managed, in-region pipeline or query your live database in plain English using simple retrieval APIs. Native vector and full-text search run underneath, with no separate stack to build.
Benefits
Why build AI on MariaDB
Accurate, grounded answers
RAG grounds language models in your own data, so answers come from your records instead of guesses, cutting the hallucinations that make AI a liability.
Ship faster, no stack to build
A unified retrieval layer grounds your AI in documents and data through simple APIs, so your team ships applications instead of building a retrieval stack.
Ground AI in docs and data
Retrieve from unstructured documents and from your live operational database, so your AI draws on everything your business knows, not just one source.
Your data stays in region
Meet data residency requirements with the entire document-processing pipeline and retrieval process running in your chosen region.
Capabilities
Building blocks for AI
Native vector & hybrid search
Power RAG with semantic, full-text, and hybrid searches. Store and query embeddings with HNSW indexing alongside relational data using standard SQL.
Unstructured RAG (documents)
A fully managed pipeline, including layout-aware parsing, chunking, embedding, vector and full-text search, and cross-encoder reranking, behind a simple retrieval API.
Structured RAG (text-to-SQL)
Turn a plain-English question into validated, read-only SQL against your live data, with multi-turn refinement. No copying into a separate warehouse or store.
Enterprise MCP Server
Connect AI agents and LLMs straight to MariaDB. They retrieve, reason over and act on live data through standard database and vector operations.
Also included:
Native vector data type
HNSW vector indexing
Isolated knowledge bases over single vector store
Flexible document chunking strategies
Multi-turn query refinement
Text-to-SQL with automatic semantic mapping
Leading embedding and generation models
Built-in cross-encoder reranker
Built-in layout-aware document parser
Looking for Enterprise Platform pricing?
Recommended resources
Get started
Start building with MariaDB
One real-time database platform you can trust, for everything you run today and everything you build next.
Frequently asked questions
RAG grounds a language model in your own data. Before it answers, the system retrieves relevant passages or records and passes them as context, so responses reflect your information instead of the model's guesses.
Yes. Structured retrieval turns a plain-English question into validated, read-only SQL that runs against your live data, with multi-turn refinement. Nothing is copied into a separate warehouse or vector store.
Document processing runs in-region. Parsing, chunking, vector storage and reranking all happen inside your selected region, keeping document data there to help you meet data-residency requirements.
MariaDB Cloud AI Retrieval gives you access to leading models for embedding and generation through integration with Amazon Bedrock and Gemini. On AWS with Bedrock models, data stays in your selected region. Gemini requests are processed outside it.
Retrieval runs against your own data with no copying into a separate store, so there is no second security perimeter to manage. Document processing stays in-region, and database retrieval is read-only.
We provide integrations with Amazon Bedrock and Gemini.