Retail Technology Show 2026: AI talk turns practical

Retailers focus less on AI hype and more on the data, infrastructure and operational change needed to make it work, writes Miya Knights, Retail Technology Publisher

Artificial intelligence (AI) dominated the agenda at the Retail Technology Show 2026. But the most useful conversations at ExCeL London were not about novelty. They were about execution.

Across diverse sessions featuring New Look and Currys and on the show floor, with vendors including RetailNext, eLocker, SAI and 4POS, the core message was clear: AI will not deliver value unless retailers have the data, systems, processes, infrastructure and change-management capability to move beyond pilots.

This context was apparent in the mix of technology products represented at the annual UK trade tech show. At the subcategory level, the largest clusters of vendors offered enterprise and cloud platforms, as well as electronic point-of-sale (EPoS) and checkout solutions, including self-checkout.

These core retail systems were out in force compared with AI and machine learning offerings, suggesting that AI was less a standalone theme than a part of the broader infrastructure and operations needed to make stores, fulfilment and customer engagement smarter.

Data readiness enabling AI

New Look chief operating officer Lynda Petherick placed data foundations at the heart of the argument. She told conference delegates that the UK fashion retailer had invested in enterprise and customer data platforms to support more advanced personalisation, forecasting and customer insights.

“[How] you have organised your data sitting underneath it to enable true data centricity is absolutely essential,” she said. New Look, she added, had stitched around 9.2 million customer data points into a single customer record “in an extraordinarily short time horizon” and scaled its loyalty programme to one million customers in around three-and-a-half months.

That work now supports more advanced AI modelling. Petherick said New Look is building the ability to forecast through customer cohorts and segments, rather than only through a traditional finance lens. It also prepares the retailer for the arrival of agentic commerce.

Moving towards agentic

“Agentic commerce is on our doorstep,” she cautioned. “If we think about that as an extra channel in which customers and consumers can engage with us, then clearly how ready we are in terms of our data, particularly around product mastery and understanding the attributes of a product, is going to be essential.”

But Petherick warned retailers against mistaking AI experimentation for transformation. “There’s a lot of shiny objects out there, but ask yourself: how demonstrably useful are they?” she asked.

“Are you actually changing the way you operate? If you think about really, truly transforming a business, really moving cost-to-serve ratios and really moving that top line, it requires a level of radical thinking, not just taking your legacy process, applying some agents on top of it and hoping for the best.”

That discipline extends into resilience. Petherick said operational technology must be treated differently from experimental tools. “If it’s truly operational technology and people are relying on it to do their business, it has to work, and it has to work all the time,” she said, adding that cyber resilience and executive ownership could not be delegated.

Removing process friction

Managing director of AI and monetisation for consumer electronics retailer Currys, Ryan den Rooijen, took a similarly practical line. He said his interest was not in “technology for technology’s sake,” but in technology that creates “genuine value,” particularly through better consumer experiences and measurable response.

For Currys, that includes using AI to support colleagues in stores and repair operations. Den Rooijen said the strongest adoption comes when AI removes friction from daily work.

“If you can come in and say, ‘This task that you’re currently doing several times a day, and it takes you 30 minutes of time, we’ve got a solution that we think will cut that down to five minutes,’ you get a huge amount of uptake,” he said. “The key here really is not so much the technology experience, it’s having that empathy.”

Building new revenue streams

Currys is also exploring AI as part of a broader monetisation strategy. Den Rooijen described three revenue buckets: selling products, selling services such as care and repair plans, and using aggregated market insight to build new commercial products and services.

“We, as Currys, have phenomenal reach,” he said. “We’ve got a deep understanding of the market, of where technology is going. What do we do with that? How do we monetise it?” He cited demand signals as one example, where aggregated browsing and buying patterns could help manufacturers understand market trends.

Retail media is part of that opportunity, although he was frank about its maturity. Asked how far down the retail media route Currys is, he said: “Honest answer, I’d say we’re probably a two out of five,” while pointing to work underway to develop privacy-aware, anonymised customer insight.

Practical deployment focus

On the show floor, the same practical deployment theme was visible in the vendor propositions. RetailNext positions its platform around in-store analytics, using traffic, dwell time and conversion data to support merchandising, store design and marketing decisions.

RetailNext UK retail engagement manager, Deli Carter, told Retail Technology magazine that retailers are investing in its computer vision AI-based systems to provide operational visibility in-store. With only 30% of stores heat-mapped, for example, Carter explained that its technology could provide a baseline against which to measure potential AI returns.

The eLocker retail proposition focuses on click-and-collect lockers, asset management and staff lockers in a single management platform, designed to automate collections, device accountability and locker management.

Representatives of eLocker explained that the proposition specifically targets retailers with mid-sized store estates who often avoid relying on third-party carriers or locker networks, such as InPost, because they want to create and own a seamless, end-to-end omnichannel customer journey.

Meanwhile, 4POS, a German EPoS supplier launching in the UK, emphasised the importance of sustainable and modular point-of-service and point-of-sale systems, and self-checkout and self-order terminals to keep pace with the heavy compute resource requirements of AI systems, for fraud detection, for example.

Computer vision intelligence

Storewide Active Intelligence (SAI) provided the clearest example of AI moving from detection to store execution. Its patented computer vision platform is designed to turn existing cameras into “in-store AI agents,” covering loss prevention, store safety, operations and customer journey use cases.

SAI founder and chief executive Som Sinha said the company’s early work focused on theft alerts, but retailers soon asked for queue management, violence alerts and other use cases. “We ended up building 45 such products,” he said. “When we turned all of them on, we were disrupting ourselves.”

The lesson, he said, was that retailers do not need more disconnected alerts. “We realised that we are on the wrong journey. Alerting is not the use case here. What we need to do is understand the store’s context and then tell them something that makes their life a little bit better than it is today,” he said.

That distinction matters. A queue alert may be useless if no colleague is available to respond. A theft alert may be the wrong priority if a safety incident is unfolding. Sinha said SAI was therefore developed to understand context, availability and action priorities before recommending what store teams should do next.

Visibility needs context

“Underneath that hood, we have got an encoder that is taking the image and changing it into language, and then we have got our own large language model that’s bringing the two together and then making a judgment continuously on what is the right thing to do,” he said.

The strongest message from the show was therefore not that AI is coming to retail. It is already here. The harder question is whether retailers are ready to operationalise it.

For New Look, that means clean data, product mastery, cyber resilience and horizontal business change. For Currys, it means colleague enablement, privacy-aware insight and new commercial models. For vendors, it means proving that AI can integrate with real retail hardware, software and workflows, not just generate more alerts.

As Petherick put it, the challenge is bigger than technology selection. “This is a revolution. It’s not an evolution,” she said. “And to do that, you need those people who are bold and want to be on this journey with you.”