All sectors

Natural language questions on the data warehouse

Teams ask their questions in French and an agent queries the data warehouse itself through SQL tools.

Talk to an engineer

Illustrative mock-up: the names, values and documents shown are fictitious.

Context

A data warehouse of more than 600 tables spread across 70 schemas.

The problem

Every business question requires a query written by a specialist.

What we built

A chatbot API and web interface: an MCP server exposes SQL tools on the warehouse, and the model calls them autonomously, on a documented subset of tables.

Steps

01Selection of the relevant tables
02MCP server and SQL tools
03Agent and interface
04Evaluation (not carried out)

Hosting and models

Azure Container Apps; Azure SQL warehouse; Gemini 3 Flash via Vertex AI and LiteLLM.

Services involved

AI in your applications · Agentic platforms

We name a client only with their written agreement. The budgets and detailed results of our engagements remain confidential.