Let’s talk
Home / Case studies
Retail / Applied AI

Retail assistant chatbot on an MCP server

A chatbot that answers product, stock and order questions by calling tools over the Model Context Protocol against a live retail database.

Architecture

The problem

A retailer wanted a conversational assistant that uses real inventory and order data instead of a static FAQ, without giving the model raw database access.

Approach

  • Exposed retail operations as typed MCP tools, so the model can only run approved queries.
  • Built the FastAPI backend and React chat front end, deployed on AWS EC2.
  • Automated build and deploy with a GitHub Actions CI/CD pipeline.

Outcome

  • The model never sees credentials or writes free-form SQL — every data access goes through a reviewed tool.
  • Push-to-deploy pipeline for fast iteration during the proof of concept.
More in this area
Applied AI

AI Data Explorer: ask your plant data questions in plain English

A natural-language-to-SQL assistant that lets operations staff query production data without writing SQL.

Oracle APEXOracle Select AIOCI Autonomous DBOpenAI GPT-4o-miniFastAPI
Read case study
Database migration

78 GB PostgreSQL migration with logical replication

Moving a live production database from AWS EC2 to a new server with minimal downtime.

PostgreSQLLogical replicationAWS EC2Linux
Read case study

Have a platform that needs to scale, stay up, or move clouds?

Tell me what you’re building. You’ll get a reply within one working day — with questions, not a sales pitch.

Start a conversation