Smart & Cognitive Checkout
Self-checkout loss prevention and barcodeless cognitive checkout — detecting non-scanned items, ticket switching, and enabling frictionless product recognition.
Two related capabilities at the point of sale: continuous self-checkout shrink detection for existing barcode-based lanes, and cognitive, barcodeless product recognition for a frictionless checkout experience.
Self-checkout shrink is not one behavior — non-scanned items, ticket/barcode switching, and bottom-of-basket misses each have a distinct visual signature, and treating them as one metric makes it hard to know which intervention is working.
Human oversight at self-checkout does not scale to watching every lane continuously, and barcode-dependent checkout struggles with produce and unlabeled items.
Approach
How Proglint Works
- 01
Overhead and scan-plane cameras monitor each self-checkout lane continuously.
- 02
Non-scanned items, ticket switching, and bottom-of-basket (BOB) misses are each detected as distinct events.
- 03
Cognitive Checkout uses automatic product recognition to identify items without a barcode scan, with a Product Acquisition Box for rapid multi-angle SKU onboarding.
Included
Core AI Capabilities
End to End
Workflow
Monitor Scan Plane
Continuous vision coverage of the scan plane and basket across each lane.
Classify Discrepancy
Distinguish non-scanned items, switching, and BOB misses as separate event types.
Prompt or Route
Trigger an on-screen prompt or route to an attendant based on the specific discrepancy.
Outcome
Business Impact
ROI Framework
The Business Case
A qualitative framework for evaluating impact — no figures are invented; this is how to think about the case until deployment-specific numbers are measured.
Shrink & Revenue Leakage Exposure
"How much of our loss is currently invisible to us?"
Unrecorded voids, un-billed food prep, and self-checkout ticket swapping share a common trait: none of them leave a record in the systems built to catch loss. A POS system can't flag a transaction that never happened. Exposure exists whether or not it is measured — continuous correlation between visual activity and transaction records is what makes it visible and actionable instead of a line item nobody can explain.
- POS-visual discrepancies are surfaced as they occur, not discovered during a reconciliation cycle.
- Loss patterns that don't generate a system record become detectable for the first time.
- Investigation starts from a specific flagged event instead of a general shrink number.
Speed of Incident Response
"How long does it take us to go from an incident happening to someone acting on it?"
Without evidence-linked alerts, investigating an incident means someone pulling footage, reviewing it manually, and cross-referencing it against whatever records exist — a process that can take days for something that took seconds to happen. When an alert already arrives paired with the relevant video clip and the matching POS or sensor record, that investigation collapses to a pre-packaged case instead of a search.
- Alerts arrive with evidence attached, not as a prompt to go looking for it.
- Investigation time shifts from a multi-day footage review to reviewing a flagged case.
- Response can happen while the situation is still active, not after the fact.
Under the Hood
Technology & Integrations
Explore Further
Where This Solution Applies
Related Industries
Related Products
DISCUSS YOUR SMART & COGNITIVE CHECKOUT REQUIREMENTS
Schedule a customized demonstration with our enterprise AI engineering team. Discover how Proglint deploys across 2,500+ locations with zero camera teardowns.