04Full stack · Internal product
Analysts used to work out every inter-store transfer by hand. The engine proposes what moves where, and they run it themselves.
Purchasing analysts worked out every inter-store stock transfer by hand: which store had too much of a product, which had too little, and how much to move. A single plan took around four hours to elaborate.
The engine now proposes which products should move from stores with surplus to stores running short, checked against real inventory and sales. The analysts review the plan and run it themselves in a web app, in two to five minutes.
The tools, the architecture and the engineering numbers, one line each.
01MLOps · Time series
Over 100,000 SKU+Store combinations, forecast weekly by a routing policy that picks a model per cluster.
View full case study: ML demand forecasting in production02Data platform · Generative AI
Weekly public data, parsed into a 52-week forecast the buying team takes into supplier negotiations.
View full case study: Market price intelligence platform03Applied AI · Guardrails
A Telegram assistant for questions about sales, inventory and purchasing. Every figure it gives comes from a query it actually ran.
View full case study: AI assistant for business data05Multi-agent · Local-first
The memory is markdown in Git, so it outlives the runtime that reads it.
View full case study: Personal AI operating system