MongoDB RAG for AI Agents

MongoDB RAGfor AI Agents.

Semantic search with automated embeddings. Ask questions, get answers.

Automated embedding generation
Configurable score thresholds
Integrate via MCP or API

Create Free Account Try Live Demo

Free forever plan available · No credit card required · Setup in minutes

Already answering 35,000+ questions every day

Works with

+ any MCP-ready agent

Quick Start

Setup in a coffee break

Connect your MongoDB database, activate collections, and start semantic search in minutes.

Connect your database

Add your MongoDB Atlas or Community Edition connection string.

Activate collections

Florentine detects vector search indexes automatically. Activate the collections you want to query semantically.

Integrate via MCP or API

Use the MCP server or REST API to connect with Claude, ChatGPT, or any custom agent.

Features & Concepts

Built for production use cases

Every feature is designed around real-world requirements: accuracy, privacy and seamless integration.

Response Control

Choose between raw query, query result, or natural language answer. Execute read-only queries automatically or review them first — you decide the level of control.

Schema Exploration

Florentine automatically explores and analyzes the schema of your collections and tables, enabling accurate, structure-aware query generation without sending your data to the LLM.

Secure Data Separation

Key/value based isolation enforces mandatory filters in every query — no relying on LLM compliance. Perfect for multi-tenant SaaS applications.

MCP Server & API

Integrate with Claude, Cursor, or any MCP-compatible agent via the MCP Server, or use the REST API to embed Florentine directly into your application.

Vector Search & RAG for MongoDB

Supports semantic searches with automated embedding creation. Set custom score thresholds and use your own embeddings.

Advanced Lookup Support

Automatically creates $lookup stages across MongoDB collections or JOINs across MySQL tables. Manual mapping assistance available for complex schemas.

Custom Prompt Instructions

Control query behavior at the database or collection/table level with custom prompt instructions. Fine-tune exactly what the LLM should and should not do.