Morrisons Has Put the Digital Basket on Wheels

The most useful thing about Morrisons’ new AI trolley is that it does not pretend the checkout has vanished.

At its Preston supermarket, Morrisons has become the first UK retailer to deploy Instacart’s Caper smart trolley. Cameras and sensors recognise products placed in the basket, an integrated scale handles fresh produce, a touchscreen keeps a running total, and Morrisons More members can connect their loyalty account to see savings and promotions while they shop.

At the end, however, the customer still pays. The trolley displays a barcode that is scanned at a dedicated Caper checkout lane to complete the transaction.

That detail makes the trial more interesting, not less.

The first generation of automated-store technology often framed checkout as something to eliminate. Amazon’s Just Walk Out concept pushed that idea furthest, promising that the store could identify what shoppers took and charge them without a conventional till interaction. Morrisons and Instacart are pursuing a more incremental model: keep a familiar trolley, add digital state to it, and reduce the amount of work left for the end of the trip.

In effect, Caper brings part of the online basket into the physical store.

A conventional trolley knows nothing. A Caper trolley can maintain a live record of what the customer has selected, how much it costs and which promotions or recommendations might matter next. That creates a digital surface inside the aisle rather than waiting for the shopper to pull out a phone or reach a self-checkout.

For grocery retailers, that surface could become commercially valuable. The screen can connect loyalty, promotions and product discovery to the actual contents of the basket. Instacart’s own pitch is explicit about that opportunity: the company wants to bring the personalisation retailers built online into the physical store, where the majority of grocery spending still occurs.

Morrisons’ trial will show whether shoppers see the same opportunity—or simply a heavier trolley with a screen attached.

The early feedback gathered by The Guardian at Preston was largely positive, particularly around the ability to monitor spending as the basket grows. That is a mundane benefit, but grocery technology often succeeds through mundane benefits. Knowing the total before reaching the till can be more useful to a family on a budget than another layer of generative-AI recommendations.

There are also practical frictions to test. Smart trolleys need to be charged, maintained and recovered. The sensing has to cope with children, bags, produce and the kind of improvised behaviour that makes real supermarkets harder than demonstrations. Customers still need a route for exceptions. And if the screen becomes an advertising surface, retailers will have to decide how much commercial messaging a shopper will tolerate while trying to buy milk.

Instacart says Caper is already used in more than 100 cities across the US and Australia. The Preston deployment is therefore not a laboratory prototype. What remains unproven is whether the economics and customer behaviour transfer cleanly into the British supermarket environment.

The strongest version of the product is not a robot trolley that replaces checkout staff. It is a persistent digital basket that happens to have wheels.

That is a smaller promise than checkout-free retail. It may also be a more deployable one.

There is a second competitive angle. Smart trolleys offer retailers a way to digitise the in-store journey without asking every customer to keep a phone in hand. Scan-and-go apps can be cheaper to deploy because the shopper supplies the hardware, but adoption depends on app installation, login, battery life and a willingness to scan deliberately. Caper shifts that interaction into equipment the store already expects customers to use.

That also changes the data available during the trip. A phone-based loyalty app may know who the shopper is, but it does not necessarily know the complete state of the physical basket. A sensor-equipped trolley can, in principle, connect basket composition to recommendations in real time. That is why Instacart’s ambitions around Caper extend into advertising and merchandising as well as checkout.

The retail-media opportunity needs restraint. A screen attached to every active trolley is attractive inventory, but it is also attached to a task. Grocery shoppers are navigating aisles, comparing prices, handling children and trying to remember what they came for. A recommendation that saves time can feel useful; an irrelevant sponsored interruption will feel like clutter. The distinction will determine whether the trolley becomes an assistant or a moving billboard.

There is a labour dimension too. Caper does not eliminate the need for staff or a payment control point, but it can redistribute work. If item recognition and basket state are sufficiently reliable, colleagues may spend less time on routine scanning and more time resolving exceptions or helping shoppers. That is a more modest labour story than “cashierless,” and probably a more realistic one.

The trial should therefore be judged on a bundle of metrics: repeat usage, time to complete the shop, error rates, shrink, average basket value, promotion redemption and the amount of colleague intervention still required. Without those figures, positive customer comments are encouraging but not evidence of a scalable business case.