Self-checkout has become a staple of modern retail, but one common item is quietly wreaking havoc on your margins: produce.

Every time a shopper hesitates at the PLU screen, types in “banana” three times, or accidentally scans a Fuji as a Gala, you’re losing something—time, efficiency, even revenue. And when that friction builds up across thousands of lanes and stores, it’s not just an annoyance. It’s an operational leak.

Visual AI is here to plug that leak—and turn self-checkout into a high-performance, margin-protecting asset.

A Problem of Scale, Not Simplicity

Retailers have spent the past decade embracing self-checkout to boost efficiency and meet shopper demand. But while adoption has grown, operational friction remains—especially when it comes to fresh, unwrapped items like fruits and vegetables. Customers fumble with lookup codes. Staff intervene. Errors sneak in. Shrink follows.

According to the National Retail Federation’s 2023 National Retail Security Survey, total retail shrink reached $112.1 billion in2022, up from $93.9 billion the previous year. The average shrink rate climbed to 1.6%, with theft (both external and internal), misidentification, and process breakdowns as leading contributors.

At the same time, self-checkout now accounts for 30% of all transactions, and 96%of retailers offer SCO in some form—making it not just popular, but mission-critical to optimize.

A global study from the ECR Retail Loss Group found that up to 23% of unknown losses were linked to self-checkout systems, and 66% of retailers flagged it as a growing concern.

Put simply, self-checkout is now too important to remain inefficient.

What Visual AI Really Solves

Visual AI flips the script. Rather than placing the burden on customers to identify produce or navigate endless PLU lists, it sees what’s on the scanner scale and makes an informed, real-time suggestion.

Advanced systems like NCR Voyix Picklist Assist  use built-in computer vision and deep learning models to recognize fresh items. It then presents a short, high-confidence list of options—minimizing both guesswork and errors.

The intelligence behind it lives in the cloud, which means models are trained continuously and improved centrally. This allows for faster onboarding, no manual database building, and shared learning across all lanes and locations.

This isn’t a one-lane-at-a-time system. It’s a networked approach to self-checkout intelligence.

The Measurable Impact

Visual AI doesn’t just improve the checkout experience—it moves the needle on key metrics that matter to retail decision-makers.

In deployments of NCR Voyix Picklist Assist, checking out a produce item has shown to be up to 44% faster. Attendant interventions drop. SCO adoption increases. Throughput per lane improves, especially during peak times.

Perhaps most critically, NCR Voyix Picklist Assist helps catch misidentified or fraudulent transactions before they’re finalized—providing real-time alerts to attendants when visual input and selection don’t match. This turns visual recognition into active loss prevention.

Combined, these improvements don’t just reduce friction. They protect margins.

A Smarter Front End

What makes Visual AI transformative isn’t just what it does—but how it fits into a broader vision of store modernization. It elevates the front end from a reactive process to a proactive advantage.

And it does it without requiring a total infrastructure overhaul. Voyix Picklist Assist integrates directly into existing Voyix self-checkout solutions and supports hybrid deployment models—whether fully cloud-based or edge-connected. With no “gold” databases to manage and models that learn automatically from live usage, deployment is fast, scalable, and sustainable.

It’s not just a product upgrade. It’s a step forward in retail intelligence.

Reducing Shrink Without Slowing Down

In a time when front-end friction affects the customer experience and bottom line in equal measure, Visual AI offers something rare: a technology that simplifies operations, protects revenue, and scales without complexity.

Self-checkout isn’t going away—but the inefficiencies that come with it can. If your lanes aren’t getting smarter with every scan, they’re leaving value on the table.

Voyix Picklist Assist helps retailers close that gap—quietly protecting margins while making the checkout experience smoother for everyone.