Albert, the Czech banner operation of Ahold Delhaize, has implemented Mettler Toledo’s AI-powered solution FreshAI at all its attended checkouts. The company is also testing the technology at self-checkout terminals in five pilot stores. Albert’s employees and customers should be able to identify products without barcodes more quickly using image recognition. In an interview with the Retail Optimiser, Aleš Kupčák, Customer Technology Director at Albert, explains that FreshAI significantly reduces the workload for store employees and noticeably reduces both the operating effort and the error rate when processing loose items. The nationwide roll-out at staffed checkouts has been underway since April and should be completed by mid-November 2025.
Albert had already installed around 1,700 FreshAI systems from Mettler Toledo at the attended checkouts in its approximately 330 stores in the Czech Republic, Aleš Kupčák says. Cameras automatically recognise loose items such as fruit, vegetables, baked goods or nuts using product recognition and significantly shorten the checkout process: Instead of 8.4 seconds, employees now only needed an average of 3.7 seconds per item, explains Aleš Kupčák. Following the positive results at the attended checkouts, Albert was now also testing the solution at self-checkouts in five pilot stores.
The camera-supported solution from the weighing technology specialist identifies products according to Aleš Kupčák within milliseconds and presents Albert’s store employees with either the exact item or a small selection of the most likely matches. In most cases, the system offered one single correct product or in the rest of the results the correct item appears among the first three suggestions with a hit rate of approximately 99 per cent.
Product recognition reduces errors and waiting times
“Our store employees used to memorise hundreds of products.” says Aleš Kupčák. “With FreshAI, all it takes is the touch of a button – the camera recognises items automatically.” The feedback from Albert’s store teams has been consistently positive. For new employees and temporary staff in particular, the system makes onboarding much easier, as it can be operated intuitively without any prior knowledge of product numbers.
Albert was able to fully integrate the hardware components for FreshAI into its existing infrastructure. Mettler Toledo’s solution is also seamlessly connected to the POS software of the retailer’s Czech banner operation: Albert uses TP.Net at the attended checkouts, TP.Scan from Diebold Nixdorf at the self-checkout terminals both of which have been significantly enhanced by the retailer over the years.
Intelligent learning across all stores
Albert manages updates and training data for Mettler Toledo’s FreshAI centrally. Every correct product selection by store employees automatically improves the recognition models across all stores. Albert and Mettler Toledo are continuously working together to further develop the product recognition solution – including complex scenarios for camera-based product recognition such as changing light conditions, different degrees of ripeness in fruit and vegetables or similar-looking products. As a result, the self-learning system is becoming ever more precise.
Mettler Toledo has recently enhanced its FreshAI solution with the new MultiRecognition feature. The system now identifies multiple products within a single transparent bag – a capability aimed primarily at the baked goods segment.
