Joe White's working days can begin as early as 5am. By the time most people are getting ready for work, the Everseen chief executive may already be through his first customer meeting. Some days stretch until 10pm, with calls spanning time zones from Europe to Australia.

"Over my 30-year career, I've always supported a global footprint, and I've gotten used to it," he says. White considers himself a morning person. "I find I get a lot more done in the morning than I do in the evenings." Still, the routine takes its toll. "I work too hard, probably," he admits, adding that he sometimes works through weekends.

White is accustomed to adjusting his working day to other people's schedules. In supermarkets, he argues, customers are increasingly expected to adapt to unfamiliar technology. Take self-checkout, for example. Shoppers who have never been trained as cashiers are expected to scan their purchases without making mistakes.

"We expect those customers to be 100% perfect in what they do," he points out. "The reality is, we know that humans make mistakes."

Not every missed scan is theft. Some are honest errors that become costly when repeated across millions of transactions.

Inside the Morrisons rollout

"Every retailer has a standard operating procedure," White says, but "we live in the real world and reality never reflects what should be on paper."

Everseen develops vision AI systems that can detect potential checkout errors and prompt shoppers to correct them. In August, Morrisons announced plans to introduce its Evercheck technology in an initial 200 UK stores after a trial. As of early October, the rollout is "roughly halfway through," White said. He reports encouraging results in recovering losses, although he offers no figures. His account dwells more on the work involved in getting the technology into stores.

Everseen is working with checkout technology partner NCR and an installer responsible for hardware and networking.

"There's always a learning element of how those parties all work together," he says. The next challenge is preparing the employees who must use it.

"Store associates are trained differently from one retailer to another," White says. "There's a learning curve because there's people in the middle." He believes Everseen needs to help train staff from the outset.

"It's a change management process," he says. The system can flag an error, but employees still need to know how to respond.

More work on the shop floor

"The jobs of a store associate have only become more complicated as time and technology have moved on," White says. He remembers when shop-floor work was divided among cashiers, stock attendants and sales staff in individual departments. Today, stores fulfill online orders, prepare collections and manage wider product ranges alongside their traditional duties.

"We put a lot of complexity on the same number of people inside a retail store," he says.

White argues that AI can relieve some of that pressure. Instead of walking the aisles looking for stockouts, employees could receive alerts directing them to shelves that need attention.

"We're just assisting the associates," he maintains, arguing that the aim is to help rather than replace them.

Whether automation eventually changes staffing levels will be up to individual retailers. White's immediate concern is whether employees can deal with problems faster. He imagines a retailer with 700 stores, each expected to have its bakery shelves replenished by 7am.

"How do you monitor that to see is that really happening or not?" he asks. The example is hypothetical, but the lost sale is easy to imagine when a customer finds an empty shelf.

A career spent learning

White knows something about tackling unfamiliar technology. As he mentioned in a previous interview with Machine Design, he studied accounting at the University of Maryland before teaching himself programming and attending night school. When he joined RFID developer Matrics, he was asked to help develop the silicon inside identification tags, despite knowing little about it at the time.

He went on to hold senior roles at Motorola and Zebra Technologies, becoming Zebra's chief product and solutions officer in 2023. The position put him in charge of product strategy and investment across handheld scanners, mobile computers, software and automation. Zebra's regulatory filings credit him with 10 RFID patents.

"RFID took a long time to take off in the industry," he recalls. Retailers such as H&M now use it to track goods, but White is interested in what can be done with the information those systems produce.

"Providing more data and more information just creates more noise for the retailer environment," he says. He sees more value in using technology to act on that information, whether that means reordering goods or redirecting a shipment.

For all the technological change White has seen, he believes the customer's basic expectations have barely moved. Shoppers still arrive with a list and expect to find what they came for.

"That really hasn't changed," he says. And for the customer walking into a supermarket at 7am, the most useful technology may be the kind they never notice: the bread is there when they reach for it.

The post The learning curve behind retail AI appeared first on Raconteur.