Inventory optimization without replacing your planning system
Turn your forecasts into decisions: how much to hold, where to position it, and when to reorder. MetaLearner works on top of the planning tools you already run.
Too much stock, or too many stockouts
Every inventory decision is the same trade-off: hold too little and you stock out, hold too much and you tie up cash in the wrong SKUs. Most teams manage it with static safety-stock rules and a gut feel for which products are risky.
The result is predictable. Working capital sits in slow movers while your best sellers run short, and nobody can point to the number that says how much either one is costing you.
The fix is setting cover from what the forecast actually tells you, SKU by SKU, so inventory stays in the band that protects service without trapping cash.
Cover set by the forecast, not a flat rule
Reorder and allocation recommendations
Every SKU and location gets its own recommendation, sized by its demand risk rather than a single company-wide rule.
Cover from uncertainty, not guesswork
We size cover from the forecast's uncertainty range for each SKU, so risky items get protection and steady items stop hoarding cash.
Solved with robust optimization
Recommendations are optimized against a range of demand outcomes, not a single forecast, so the plan holds up when reality deviates from the expected line.
Service level balanced against carrying cost
Every recommendation weighs stockout risk against the cost of holding, so you can see the trade-off instead of defaulting to more stock.
Exception alerts before it hurts
MetaLearner flags SKUs drifting toward a stockout or piling up as excess, in plain language, while there is still time to act.
On top of your planning stack
We work with the exports your team already produces from SAP, Oracle, NetSuite, Kinaxis, and similar tools. No replacement project.
Built on robust optimization
Most planning tools optimize against a single forecast. The plan looks optimal in the model and breaks the moment reality deviates: a demand spike, a supplier delay, a tariff change.
MetaLearner solves the decision against a range of demand outcomes at once. The reorder you commit to stays feasible across the futures you actually face, not just the expected one.
In a published 52-week backtest across 17 warehouses and 63 items, this approach held a 94.9% service level with 40% less inventory than the strongest forecast-plus-buffer benchmark. The method draws on decades of robust-optimization research, with Dr. Melvyn Sim of NUS advising. Read the full write-up.
What better inventory decisions are worth
Start free, expand when it pays off
Diagnostic
Free, 1-2 weeksWe quantify the working capital exposed to forecast error in your categories, on your data. No IT project, no procurement.
Pilot
Paid pilotInventory recommendations on a focused subset of SKUs, with light integration, so you can measure the working capital released before committing.
Production
Scoped to your deploymentFull rollout across categories and locations, with continuous recommendations, exception alerts, and quarterly reviews.
Built for the people who own cover
Inventory optimization fits mid-market manufacturers, distributors, and high-SKU retailers where cash and service level are both on the line. If your team still sets cover with static rules and spreadsheets, this is built for you.
Questions buyers ask
Do you replace our planning system?
No. We sit alongside whatever your team plans in today, whether that is SAP, Oracle, NetSuite, Kinaxis, Anaplan, RELEX, spreadsheets, or something else, and we use that output as an input. You keep your current stack and add MetaLearner as the decision layer.
How is this different from our ERP's inventory module?
ERP modules apply fixed rules. We set cover per SKU from the forecast's uncertainty range and balance it against carrying cost, so the recommendation reflects each item's actual risk.
What is robust optimization?
A method for making decisions that stay sound across many possible futures instead of betting on a single forecast being right. It is what turns MetaLearner's forecasts and uncertainty ranges into reorder recommendations you can trust.
What about safety stock we already set?
The diagnostic shows where your current safety stock is over- or under-protecting, in dollars, before you change anything.
What data do you need to start?
Sales history and inventory records, plus any forecast files you keep. Spreadsheets and exports are fine, with no data warehouse requirement at the diagnostic stage.
How long until we see a number?
The free diagnostic produces a defensible working-capital exposure figure by category in 1-2 weeks.
What does it cost?
The diagnostic is free. The pilot is a small, fixed-scope paid engagement, and production pricing is scoped to your deployment.
Go deeper on the method
See the cash trapped in your inventory
Schedule the free Forecast-to-Inventory Diagnostic. In 1-2 weeks you get a defensible working-capital exposure figure by category, with no IT or procurement involved.
