Redacted.
Redacted.
AI orchestration completely embedded across the operations of a UK food & beverage manufacturer. Built to save millions. Approaching operational pilot.
One operator. Multiple connected surfaces — supply chain, demand planning, sales, market intelligence, commodity & FX exposure, financial reporting. Designed end-to-end. Built solo. 78 specifications define the system. 250 SKUs live in the operational pilot.
The supply side runs a Prophet tournament — multiple forecasting models compete per SKU, and the best-performing model wins each product’s forecast. An XGBoost demand-sensing layer sits on top, catching what the statistical models miss: seasonality, cultural demand, the lift and hangover of a promotion. The system watches itself for drift and retrains on a schedule, keeping the new model only when it beats the one it replaces.
A 39-tool AI copilot handles the operational surface — from natural-language queries across any dataset to automated report generation for the C-suite. Every tool is governance-aligned: decision gating ensures no AI action executes without appropriate human approval. From forecast to safety stock to reorder, the chain holds itself together. Operations first. Finance is next.
What follows is the application, with customer name, product names, and counterparty names blacked out. The numbers stay, because the numbers are the point — they show the scale of what one platform is doing across one organisation.
