Case Study

SafariMart Africa

After 13 years of customer data, SafariMart had zero useful queries. Meet "Jack": a custom-trained AI with product intelligence that handles even compound questions.

AI Chatbot Retail Intelligence
The Challenge

13 Years of Data, Zero Answers

A decade of sales history, product knowledge, and customer behaviour sat hidden in a database that no one could ask questions of.

Untapped History

Years of transactional data held every answer about what sells, when, and to whom, but it was locked in raw tables.

Manual Customer Service

Every product question was answered by staff from memory. Best sellers, stock, recommendations, all in one brain.

Simple Bots, Wrong Answers

Off-the-shelf chatbots failed on compound questions like "which waterproof camera under 40k is in stock in Westlands?"

The Solution

What Was Built

A custom-trained AI assistant with live access to products, prices, and stock, built to answer the way a top salesperson would.

Custom-Trained on Real Data

Trained on SafariMart's 13 years of product and sales history, not generic internet knowledge.

Compound Query Handling

Understands multi-part questions that combine category, budget, brand, and location in one sentence.

Product Intelligence

Knows the catalogue, live stock, and pricing, and recommends from what is actually available now.

Everywhere Customers Chat

Deployed on the storefront and messaging channels, so history and recommendations are available 24/7.

Key Results

The Impact

13yrs
Of data, now queryable
24/7
Instant product answers
Multi
Condition queries handled
Live
Stock-aware recommendations
Tech Stack

Technologies Used

Frontend

React, Next.js, Tailwind CSS

Backend

Node.js, PostgreSQL, Custom AI Model + RAG

Integrations

Storefront API, WhatsApp Business API, SMS

Want Similar Results?

Let's turn years of your data into an assistant that answers like your best salesperson, any hour.

Book Strategy Call Start the Conversation