The right material.
Exactly where it belongs.
A feature strip can belong inside one appliance and on the door of another. Explore a Steve pilot that checks point-of-sale materials against your model-specific placement guide: product identity, artwork version, and installation surface. Unclear evidence goes to review.
Early access. Ask us about joining.
Steve{ "product_model": "COOL-400", "material_id": "FRESH-DRAWER-v3", "model_match": true, "artwork_match": true, "expected_surface": "interior_drawer_front", "observed_surface": "exterior_door", "placement_match": false, "verdict": "flagged", "reason": "Correct artwork, wrong installation surface"}The evidence and your rules.
- Product catalogue with model identifiers, reference views, and approved material variants
- Placement guide for each model: required pieces, interior or exterior surface, position, and campaign version
- Store photos showing the complete product, its open interior where needed, and readable material and model details
- Store and visit identifiers; retailer price labels defined separately from the brand materials being checked
Fields, a verdict, and the proof.
- Structured fields with a confidence score each, delivered by API or webhook
- A verdict against your rules: approved, flagged for review, or blocked
- The evidence and the reason, side by side, in a review queue your team can act on
Model, material, and placement. Checked separately.
Identify the exact product
Match a readable model label or catalogue identifier with the photographed unit. Similar-looking models need corroborating evidence; appearance alone must not silently confirm identity.
Match the material, not just the logo
Compare the feature strip, sticker, or panel with the approved artwork for that model and campaign. Similar colours and a shared logo do not make two variants interchangeable.
Check interior versus exterior
Compare the observed surface and placement zone with the brief. A drawer strip on an outer door is a mismatch even when the artwork is correct. Hidden surfaces remain not assessable.
Account for every required piece
Check each material separately for presence, orientation, and visibility. Distinguish a visible missing piece from a view that never showed the required area.
Separate retailer labels from brand materials
Treat retailer price cards and other allowed labels as their own objects. They must not count as installed POSM, while any obstruction can be checked against the brand's rules.
Keep evidence tied to the visit
Screen for reused photos and keep results linked to the supplied store, visit, model, and brief version. Metadata supports the check; a timestamp alone does not prove where or when a photo was taken.
Start with a pilot on real evidence.
This workflow is in early access. A placement guide defines the standard; representative store photos are needed to establish which checks can be automated reliably. The example output illustrates the proposed workflow.
- Agree the model-to-material mapping, placement zones, and permitted exceptions with the brand.
- Label real examples of correct and incorrect installations, lookalike models, glare, occlusion, and missing views.
- Evaluate unseen submissions for false approvals, missed errors, and review volume before enabling automatic decisions.
From capture to decision.
Define the reference
Turn your approved placement guide into model-specific checks and version the artwork.
Capture the evidence
Collect the overview, interior, material detail, and identifier views needed for that model.
Compare each requirement
Check identity, artwork, and placement separately. Route uncertain checks to a reviewer or request another view.
Close the loop
Send findings through the API to your field tools. Link the correction photo to the original installation and recheck it.
Where this fits.
Can this start from a PDF placement guide?
Yes, a guide is useful for defining the proposed checks. We also need the approved artwork files, an explicit mapping to models, and real store photos. Clean reference slides alone cannot establish performance on field images.
Can one photo prove both the outside and inside are correct?
Only if the relevant surfaces are actually visible at sufficient detail. Normally a closed-door overview and an open-door view answer different questions. If the evidence does not show a required area, the result should request that view rather than assume a pass or failure.
Does this certify the appliance or the claims printed on its stickers?
No. This checks a visible installation against the brand's approved standard. It does not test product performance, certify safety, or establish that a printed feature claim is legally or technically valid.
Is it limited to refrigerators?
The same approach can be explored for washing machines, televisions, small appliances, branded coolers, and other retail equipment. Each product family needs its own references, capture requirements, and pilot evaluation.
Same engine. Different evidence.
Let’s run it
on your evidence.
Bring your placement guide, approved artwork, and representative store photos. We will scope a pilot and agree how to measure it.
Book a demo 30 minutes. Your use case. Real possibilities.