Product
A diagnostic that has to earn the right to change your orders.
Most tools forecast and then optimise against the forecast. We start from what you already do, and only move away from it when your own history shows it would have paid.
Your history becomes a policy
From purchase orders, goods receipts, sales and stock counts we infer how you order each SKU today: when you reorder, how much, how often, and how you react to a late supplier. That policy is the incumbent. Every candidate must beat it.
The state you are operating in
Per supplier and per lane, the diagnostic asks whether the recent past looks like the normal past. Late deliveries, short shipments and demand shifts move a probability, not a switch. Public signals for your lanes, chokepoint transits, port calls, freight and fuel indices, add an outlook with its quality stated.
Candidates on your own past
Each candidate optimizer is replayed through rolling windows of your history under many demand and lead-time scenarios. We keep the mean, the conservative lower bound and the worst fold, per candidate.
Adopt only when proven
A candidate is adopted only if its lower bound beats your policy on held-out folds. Otherwise the report keeps your plan and says exactly why. There is no forecast of savings, only what happened on your data.
A purchase plan you can act on
When a candidate passes, you get order quantities and timing per SKU with the reason for every change from your current plan, a stability cost so the plan does not churn week to week, and the lane risk that shaped it.
Try it on a year of your files.
A private workspace opens in seconds. If the answer is “keep doing what you do”, you will hear it plainly.