Three steps.Same visit.
No new hardware, no new route plan, no second system for the field team to resent. The only change is that the shelf gets photographed instead of counted.
Scan
The rep opens the app and photographs the fixture. A wide aisle is captured in one pan, with an on-screen guide showing what has been covered so far.
It takes seconds, and it replaces the least valuable task in the visit — counting facings by hand onto a form.
Works with no signal
Captures are held on the phone and upload themselves when the rep gets signal. Nothing is lost in a basement aisle or a rural store.
Any shelf width
Long fixtures are stitched into a single image on the phone, so a six-meter bay is one capture rather than eight disconnected photos.
Quality checked on the spot
Blurred or badly-angled shots are caught before the rep walks away, instead of being rejected hours later when it's too late.
Privacy handled
Shoppers and staff can be masked on the phone before the image is ever uploaded.
Analyze
The image is read against a model trained on your catalog. Every facing is matched to a SKU and every gap is marked as a gap — then turned into the KPIs your commercial team already reports on.
Part of this runs on the phone itself, which is why a result comes back in the aisle rather than after the sync.
Trained on your products
Not a generic object detector. The model learns your packs, your pack sizes and your seasonal variants.
Handles packaging refreshes
When a design changes, the model is retrained on the new pack rather than quietly mis-counting it for a quarter.
Your team keeps the final say
Anything the model gets wrong is corrected in one click in the review screen — and that correction trains the next version.
Consistent by construction
The same shelf photographed twice gives the same answer. Two different reps give the same answer as each other.
Act
Results land in the back office as the visits happen. Head office sees the estate; the rep sees their own stores; the store gets told what is missing while it still matters.
Availability alerts
The stores out of your priority SKUs, ranked, rather than a national average that hides them.
Commercial dashboards
Share of shelf, compliance and promotion execution by chain, region and rep.
Field coaching
Coverage and visit quality per rep — so managers can help the ones who need it.
Feeds your existing stack
Results export into the reporting and ERP systems you already run, rather than becoming another portal nobody opens.
What the first eight weeks look like.
A pilot is designed to answer one question honestly: does this work well enough on your catalog, in your stores, to be worth rolling out?
Weeks 1–2
Catalog and scope
We take your product list and agree the stores, categories and KPIs the pilot will cover.
Weeks 2–4
First captures and training
Your reps start photographing shelves. Those first images train the model on your actual packs in your actual lighting.
Weeks 4–6
Live results and correction
Results appear in the back office. Your team reviews and corrects, which sharpens the model on the SKUs you care about most.
Weeks 6–8
Measured outcome
We report accuracy on your catalog and what the pilot found — including the availability gaps nobody knew about.
Start with a pilot.
Your catalog, your stores, your numbers — before you commit.