Services How we work Work About Blog Book a discovery call →
Case study · Retail

From AI pilot to the operating system of a retail group.

Demand forecasting, replenishment agents and the planner console, live in 14 weeks across 200+ stores.

From pilot to production One conveyor carries work through three gated stages. Pilot, at week 0, is greyed out and paused beside a prototype dashboard. Shadow mode, at week 11, sits inside a dashed sandbox where a dashed copy of the agent runs beside the existing process, which an eye watches. Production, at week 14, ends at the Awesomist core building with live services, data stores and status pins. PILOTWeek 0 SHADOW MODEWeek 11 PRODUCTIONWeek 14
) Results
Fewer stock-outs
38%

across 200+ stores, year over year

Pilot → production
14wks

from kickoff to live in their cloud

Forecast accuracy
2.4×

improvement across every category

Platform uptime
99.98%

in production since Q1 2025

) The story
Context

A regional retail leader with 200+ stores, running replenishment on spreadsheets and gut feel.

Challenge

Three previous AI pilots had stalled in the lab. Nothing had survived contact with real operations.

Build

Demand forecasting, replenishment agents and the web console planners actually use, deployed into their cloud.

Result

Live in 14 weeks. Forecast accuracy up across every category, stock-outs down group-wide.

) What we built
One system, three parts

Models, agents and the software people use.

The forecast only matters if a planner acts on it. So the console was built alongside the models, not after them.

Machine learning

Demand forecasting

Store- and category-level forecasts that learn from every week of sales.

Custom ML →
Agentic AI

Replenishment agents

Agents that turn forecasts into proposed orders, with planners approving the exceptions.

Agentic AI →
Engineering

Planner console

The web console planners open every day to review, adjust and release orders.

Web & enterprise apps →
) Start
One conversation is enough to start

Have a pilot that never left the lab? Let's get it live.

Tell us what stalled. A senior member of the team replies within one business day.

Book a discovery call →