In July, Morrisons unveiled a futuristic shopping assistant – an AI-powered trolley that could recognise items, total a basket in real time and even process payment without a traditional checkout. The project, branded as the Caper Cart, was part of a joint trial with US-based Instacart, a company that supplies grocery-delivery technology to retailers worldwide.
The experiment was confined to a single Morrisons store in Preston, often referred to as the Riversway or Ashton-on-Ribble location. Inside the cart, a suite of sensors, scales and a small touchscreen worked together to identify each product as it was placed inside, displaying a running total that shoppers could monitor while they moved through the aisles.
What the Caper Cart promised
The trolley’s hardware bundle included computer-vision cameras that captured each item, weight sensors that verified produce, and a touchscreen interface that showed the accumulating cost. When the basket was complete, the system generated a unique barcode; a dedicated lane at the checkout allowed customers to scan that code and settle the bill instantly. Anti-theft features, such as a red-light alert, and automatic battery recharging when carts were stacked together, were also built into the design.
Beyond the customer-facing experience, the Caper Cart was billed as a data-rich platform that could feed purchase information back to the retailer, potentially improving inventory management and personalising promotions via the customer’s More Card.
Why the pilot fell short
After barely three months of operation, Instacart announced that it would pull the plug on the entire Caper Cart programme, both in the UK and in the United States where the carts had appeared in retailers such as Schnucks and Wegmans. The firm cited “retailer adoption has not scaled quickly enough to justify our continued investment.” In plain terms, supermarkets were not embracing the technology at a pace that made the heavy-handed R&D spend worthwhile.
Consumer behavior contributed to the stall. Shoppers were required to learn a new way of scanning and paying, a habit change that proved too demanding for many. While computer-vision can recognise items in seconds on a phone, the added step of handling a high-tech trolley and navigating a separate checkout lane diluted the promised speed advantage. The novelty of the carts also sparked online jokes about theft and the carts ending up in canals, highlighting a broader scepticism.
AI that works better behind the scenes
Although the customer-facing experiment faltered, AI continues to thrive in less visible roles within grocery stores. Morrisons, for example, has deployed AI-enhanced shelf cameras that monitor stock levels, trigger automatic re-orders and flag potential issues for staff. These systems have earned industry awards and demonstrably reduce out-of-stock events.
Other retailers are leveraging machine-learning demand forecasts to cut waste and optimise supply chains. Such behind-the-scenes applications deliver tangible ROI without requiring shoppers to alter their routine, a contrast to the Caper Cart’s front-end approach.
Instacart expects to complete the global wind-down of Caper Cart operations by the end of June 2027. Morrisons has indicated that the trial’s termination will not affect its finances or broader store operations, and the retailer plans to keep exploring AI-driven solutions that better align with shopper habits.
The brief life of the AI trolley underscores a key lesson for the retail sector: cutting-edge hardware must deliver a clear, frictionless benefit to consumers, or else the technology is likely to retreat to the background where it can still add value without demanding a behavioural shift.
