Retail AI scales up: Shelf-edge systems and digital twins
As business futurists, over the past couple of years we’ve spent a lot of time talking about advances in AI and the potential this technology has to deliver huge change to the retail industry. While it all looks and sounds very interesting and forward-thinking, one of the responses we often hear from our clients is that it’s difficult to imagine how it can be deployed at scale in a way that provides real value to businesses.
Yet innovative retailers are already demonstrating that AI is no longer just theoretical – it’s already robust enough to be a practical reality in brick-and-mortar environments. From computer vision systems that reduce theft and optimise shelf space, to sophisticated recommendation engines that empower store associates with better real-time information about their store, to inventory management systems that predict demand with unprecedented accuracy – behind the scenes, AI is quietly revolutionising physical retail.
In this feature, we’ll explore how leading retailers are successfully implementing AI solutions that deliver measurable ROI. We take a look at technologies that are ready for prime time today and how they are being deployed at scale.
Morrisons x Focal Systems
UK supermarket Morrisons has partnered with San Francisco-based tech company Focal Systems, using the company’s AI store system to improve its on-shelf availability by 2%, a number that translates to significant sales increases. We spoke to Simon Shankster, Vice President Sales & Customer Success EMEA at Focal Systems, about the capabilities of the company’s shelf-edge camera technology and how it uses AI to deliver tangible benefits to the retailers it works with.
Founded in 2015, Focal Systems specialises in shelf-edge camera technology that enhances retail inventory accuracy. In its early days as a start-up, Focal set about exploring the possibilities of computer vision in retail. Launched within the Stanford Computer Vision Lab back in 2015, after five years of experimentation the business pivoted to focus purely on shelf-edge camera technology in order to translate the images gathered by its system of cameras back into availability for customers.
The system, deployed in hundreds of cameras across stores, captures images once an hour to monitor product availability. Based on the data picked up in the images, the system creates a list of tasks for workers to complete to maximise availability – prioritising tasks based on value and top-selling items, significantly reducing the number of hours store associates need to spend walking up and down aisles manually scanning. Focal Systems claims a five to ten times return on investment and has rapidly expanded its reach globally – with Morrisons having now outfitted all of its UK stores with the system.
So how does the system actually work? Each store is outfitted with around 500 thumbnail-sized cameras, with more or less cameras required depending on the size of the store. The cameras are battery powered, so don’t require lots of maintenance. Dormant most of the time, they capture a single image each hour and go back to sleep. This provides a view of the store environment that is as close to real-time as is actually necessary, without draining the camera’s battery life or creating too many tasks for store staff to deal with in a timely way.
A recent study from Prosper Analytics found that 32% of US adults are ‘extremely concerned’ about the use of AI violating their privacy, which means retailers will need to take special care to keep customer data secure when rolling out new technology in stores. Focal says that because its images are captured only once an hour, privacy issues are avoided. If someone happens to be standing in the way when the photo gets taken, the system automatically drops all the pixels, leaving a black silhouette containing no identifiable personal information.
Optimised images are sent from the cameras via Wi-Fi where they are processed in the cloud. Focal’s AI analyses the images received and translates that data into tasks that appear on store colleagues’ handheld devices. It allows stores to keep on top of stock availability without needing to have employees constantly walking up and down the aisles scanning and monitoring. This saves retailers time, automatically completing a lot of the legwork that would otherwise have to be done manually.
Lowe’s x Nvidia
Other brands are also rolling out AI systems at scale in their stores. US retailer Lowe’s is using Nvidia’s Omniverse technology to create digital twins for its stores. With this advanced spatial modelling, Lowe’s has created accurate digital replicas of its 100,000 square foot locations, allowing it to apply e-commerce style analytics to bricks-and-mortar retail.
The company says that the use of the digital twin technology dramatically increases planogramming efficiency, as it allows teams to collaborate in real-time to visualise and optimise store layouts. The system uses AI avatars developed in Lowe’s Innovation Labs to simulate different store layouts before implementation, predicting how changes will impact shopper behaviour. This virtual testing is helping to eliminate costly trial-and-error approaches to store design. The company claims that this data-driven approach to layout optimisation has reduced customer walking distances by 22% and increased complementary purchases by 18%.
The system’s real-time capabilities also help prevent common issues with self-checkout, from picking up on accidental oversights and preventing potential theft. All this tech is supported by Nvidia’s powerful Tesla P4 GPUs, essentially giving them access to mini data centres in each of its 1,700 stores.
This technology represents the convergence of physical and digital retail strategies, pointing to a future where stores evolve continuously based on real-time data rather than periodic remodels. As customers expect increasingly seamless shopping experiences, Lowe’s demonstrates how traditional retailers can apply sophisticated digital tools to enhance their core physical advantage.
John Lewis x Yoti
As well as being deployed in stores, AI is being used to address common speed bumps while shopping online. In January 2025, UK retailer John Lewis announced it had partnered with British tech company Yoti to implement facial age estimation technology for online knife purchases, creating a frictionless yet robust age verification system that’s accurate to within 1.3 years for teens.
The system requires shoppers to consent to a facial scan, which is then analysed behind the scenes by Yoti’s AI to determine age without storing personal data. For added safety, the system includes a buffer beyond the legal limit of 18, ensuring underage customers cannot complete purchases. Anti-spoof features also ensure the system can’t be tricked by using a photo.
Tests showed a 96% verification completion rate, significantly higher than traditional ID methods, while successfully preventing underage purchase attempts. The technology operates entirely on the user’s device, addressing privacy concerns that often accompany facial recognition systems. The partnership represents a growing trend of retailers using AI to comply with regulations without compromising the customer experience, as well as potentially establishing a new standard for age-restricted products across UK e-commerce as knife crime prevention measures intensify.
Read more like this and subscribe to our newsletter here.
Like this and want more?
Get in touch to subscribe, or request a demo for GDR's The Futures Vault for on demand insights and foresight:
Think Smarter Faster
Impressive deployment and results, Morrisons and Focal Systems!