Tesco : AI-Powered Supply Chain Revolution

Retail & E-commerce

Tesco – AI for Smart Supply Chain Optimization

Our AI supply chain solution enabled a leading retail chain to refine its stock management and replenishment workflows. By analyzing product demand trends, customer buying behavior, and seasonal shifts, the AI continuously forecasts stock requirements with precision. We built machine learning pipelines that integrate with their existing ERP, offering real-time alerts for low-stock items and automating restocking decisions. This dynamic system significantly reduced inventory waste, improved product availability on shelves, and enhanced operational efficiency across stores. Tesco now relies on intelligent automation to keep its shelves stocked and customers satisfied, while also saving millions annually on logistics and warehousing costs.

Project Overview

Client: Tesco, the UK’s largest retailer, makes £65 billion.
Challenge: 32% perishable waste in fresh food categories + £220M annual overstock costs
Goal: Create a real-time AI inventory system for:
- Cut waste by ≥25%
- Improve shelf availability to >98%
- Integrate with SAP S/4HANA ERP
My Role: Lead AI Architect (GenX Software Team)
Team: 8 (4 Data Scientists, 2 ERP Specialists, 2 Cloud Engineers)
Timeline: 9 months (Pilot to Full Scale)

“GenX’s artificial intelligence now makes 83% of our replenishment decisions autonomously.” It’s like having 200 veteran category managers working 24/7.”

Anup Shetty

Tesco Chief Supply Chain Officer

The Challenge

Critical Pain Points:

£1.2M/day in spoiled fresh produce
11% out-of-stock rates during peak demand
Outdated system failed to analyze:
- 2.5M+ daily POS transactions
- 200+ demand drivers: weather, events, promotions

Technical Hurdles:

Seamless SAP system integration
Explainable AI for compliance audits
500ms latency for store-level decisions

Tech Stack

Component	Technologies
Data Engineering	Azure Databricks, Delta Lake
ML Models	Prophet, XGBoost, PyTorch
MLOps	MLflow, Kubeflow
Cloud	Azure (AKS, Functions)
ERP Integration	SAP BTP, OData APIs
Monitoring	Prometheus, Grafana

Key Innovations

Tesco’s “Just-in-Time Perishables” system dynamically routed short-shelf-life items to high-traffic stores. AI detected viral trends and adapted inventory in real-time to prevent stockouts. Self-learning order policies helped navigate strikes and supply delays, making restocking smarter and faster.

Just-in-Time Perishables” System

Dynamic routing of short-life commodities to high-traffic retailers.

Result : 28% reduction in dairy waste

Promotion Shock Absorber

AI detected and adjusted for unplanned viral trends (e.g., #TikTokPasta)

Result : Prevented £4.7M stockouts during 2023 Christmas

Self-Learning Order Policies

Continuously adapted to :

Truck driver Strikes
Supplier delays

Result : 19% fewer emergency deliveries

Our AI/ML Architecture

Core Models

Demand Forecasting
- Hybrid Prophet+XGBoost model (MAPE 3.2% vs legacy 11.7%)
- Incorporated
  - Micro-local trends (e.g., school holidays)
  - Promotional lift curves
Perishability Optimizer
- Computer vision-powered ripeness monitoring
- Reinforcement learning for dynamic markdowns
Autonomous Replenishment:
- Adaptive decision-making via multi-armed bandits
- Real-time SAP PO generation