Grocery & FMCG

Short-life categories like produce, deli, bakery, and prepared foods are forecasted at store-day and intraday granularity — giving grocers the precision needed to match supply to actual demand patterns before product spoils, not after.

Lost sales are estimated by reading sales data against store-level inventory — identifying when an out-of-stock suppressed demand and quantifying the units customers would have bought if the shelf had been full. That recovered demand is fed back into the forecast, so the baseline reflects true customer demand rather than censored sales history — breaking the self-reinforcing cycle where a stockout teaches the model to under-order the very item that ran out, causing it to stock out again.

For fresh and loose items, the recorded store inventory rarely matches what’s actually on the shelf — produce sold by weight, spoilage, shrink, and in-store sampling all erode the record between counts, leaving replenishment to run on a number that’s already wrong. Solvoyo predicts true sellable inventory at the store-item level by combining deliveries, sales, shelf-life, and historical shrink patterns into a continuously updated estimate of what’s genuinely available to sell. With an accurate on-hand position, ordering and waste decisions reflect shelf reality instead of a phantom count — eliminating the double penalty where inflated records suppress orders into stockouts while unrecorded spoilage quietly drives waste.

Safety stock is set dynamically to hit the service-level target you choose — recalculated as demand and supply conditions shift, instead of sitting in fixed min/max levels that go stale the moment they’re entered. Each buffer is sized to the actual volatility it has to absorb: the more variable a SKU’s demand and the less reliable a supplier’s OTIF, the more cover it gets — and the more predictable the flow, the leaner it runs. The result is availability held to target without the blanket overstock that fixed thresholds force across every item, so working capital sits where variability genuinely demands it rather than spread evenly out of caution.

Store and DC orders are generated automatically using live demand signals, upcoming promotions, and shelf-life constraints — with service-level focused inventory optimization that simulates the trade-off between waste and lost sales before a single order is placed. The result: no spreadsheets, no manual approvals, and no guesswork about whether you’re cutting waste at the cost of availability.

Perishable waste is predicted at the batch level across the full planning horizon — projecting, for each production or receipt batch, how much will go unsold before its expiry date given forecasted demand, current inventory age, and remaining shelf life. Acting on that forecast, batch-level dispatch recommendations are generated using FEFO logic — sequencing which batches ship from the DC to which stores first so the stock closest to expiry moves to the locations that can sell it through in time. Markdowns are then timed store by store to recover value on the units still at risk — replacing reactive end-of-day write-offs with a continuously optimized plan that prevents waste before it happens, not after.

AI-powered promotion planning and list price management are executed against live demand signals — with models that account for halo effects, cannibalization, and pantry-loading behaviors to provide full visibility into the net impact of each promotion across sales volume, revenue, and profitability. Every decision is governed by a complete audit trail across every banner, so promotional strategy is grounded in true incrementality, not just lift on the promoted item.

Replenishment and transportation are planned together in a single optimization — so the order that’s right for the shelf is also the order that fills the truck, rather than a replenishment plan being thrown over the wall to logistics to move at whatever cost it takes. Store and DC orders are shaped with load building, truck capacity, delivery frequency, and routing solved concurrently, balancing on-shelf availability against freight cost in the same decision rather than in sequence. The result is fuller trucks, fewer under-utilized runs, and lower cost-to-serve — availability protected and waste controlled without freight quietly eating the margin that better ordering just recovered.

Solvi, Solvoyo’s AI planning agent, continuously monitors demand signals, stock cover, waste levels, and supplier OTIF performance across every store and category — detecting perishable stockout risks, excess inventory build-ups, and promotional allocation imbalances before they translate into shrink, lost sales, or margin write-offs. Solvi diagnoses the root cause behind each exception, quantifies the waste and lost sales cost of inaction at the store and SKU level, and generates prioritized action recommendations — from replenishment adjustments and markdown timing to promotional reallocation and supplier escalations — that grocery planners can review and execute directly within the platform.