Picklist Assistant + Loss Prevention

One camera,
two jobs.

A single camera at the self-checkout produce area recognises non-barcoded items to speed up picking — and catches wrong-item selections before the receipt prints. Faster lanes and loss prevention from one recognition layer.

Request a demo →

Retellect picklist on a live self-checkout — the screen suggests the matching produce while an item sits on the scale.

Live in stores since 2020

~98% recognition accuracy

Easy integration via an open API

Runs on existing checkout hardware

The challenge

Today, choosing a non-barcoded item at self-checkout is a manual search on the screen — slow enough that shoppers buying weighed goods avoid self-checkout, and open enough for the wrong code to be picked. Both happen at the same step, and the checkout has no way to catch either.

Operational problems

Speed

14–40s to pick an item

Finding a non-barcoded product on the screen takes 14–40 seconds depending on category — every fresh item slows the lane.

Friction

Help calls pile up

Shoppers who can't find an item press for help — one attendant covering several lanes becomes the bottleneck the whole bank waits on.

Adoption

Shoppers walk past self-checkout

Anyone with loose produce learns to avoid SCO — so the lanes you invested in sit idle and store productivity drops.

Loss-prevention problems

Blind spot

Invisible to the till

The weight and the price line up, so existing controls never see it.

Deliberate

Substitution

An expensive item on the scale, a cheaper look-alike picked (organic apple → conventional) — or an honest mis-pick of look-alike produce. Either way the cheaper code wins.

Disguise

Non-picklist items

A bottle of liquor in an opaque bag, placed on the scale and selected as potatoes. The camera sees what the scale can't — the picked code doesn't fit the frame.

The solution

AI picklist with Loss Prevention

One camera at the self-checkout produce screen does both jobs: it makes item picking faster, and it detects the losses the till cannot see.

Checkout efficiency

The camera ranks the most likely picks on the screen — the shopper or cashier just taps the right one.

Step 1

Place the item

A loose item — fruit, vegetable or bakery — is placed on the scale.

Step 2

The camera recognises it

In under a second, the most likely item appears on the screen.

Step 3

Tap to confirm

One tap confirms the pick — no menu search, no code to remember.

Faster transactions
Produce and other items without a barcode are recognised in about one second. Lanes move faster, and customers spend less time looking through menus.

Higher productivity
Today, shoppers buying loose items prefer staffed lanes. AI picklist moves them to self-checkout — so staffed lanes handle fewer transactions and need less staff.

Better shopping experience
Shoppers find the right item in one tap — no menu hunting, no waiting for help. Checkout feels quick and easy, so more customers use self-checkout.

~15s → ~3s

In a live grocery installation, average item-selection time fell about 5× — from roughly 15 seconds to 3 seconds per item.

Loss prevention

When the wrong item is picked, the system reacts in real time.

Step 1

Wrong item picked

The wrong product is selected — by mistake or on purpose.

Step 2

Instant alert

Right away, the system sends an event to the store's software.

Step 3

It gets fixed

Based on your rules, it asks the shopper to fix it, calls the attendant, or alerts security.

Eliminate unintentional customer errors
Customers often pick the wrong item by accident — one look-alike fruit instead of another. The system spots the mismatch before the receipt prints, so you stop losing margin to honest mistakes.

Eliminate intentional theft
Some swaps are on purpose — an expensive item rung up as a cheap one, or alcohol hidden as potatoes. The camera sees what the till cannot and flags it at the lane.

Solution highlights

Adaptive recognition
Handles the hard cases — hands in frame, look-alike produce, unusual lighting, new packaging.

Self-learning
Accuracy improves with use — new and seasonal items are picked up automatically, no manual retraining.

One engine, every checkout type
Self-checkout, manned lanes and self-service scales through one integration.

Open API
Easy integration with your existing self-checkout, POS and scale systems.

Edge deployment
Runs on your existing store infrastructure — no new hardware, no internet dependency.

Data & privacy

The camera looks at the product zone only — not the shopper. No biometric data, no payment-card data. Frames are kept no longer than needed for recognition and review, then discarded.

Analytics · Included

Numbers you can act on.

Every pilot reports its results in one place — so you measure what works and decide on the numbers, not a sales pitch.

See it in one of your stores.