At the Retail Technology Show 2026 keynote panel, the shift towards agentic AI was discussed in a way that felt far more grounded than the language around it usually suggests. Framed around efficiency and customer engagement, the conversation moved quickly away from big claims and into how retailers are actually applying these systems now, where they are seeing value, and where the gaps still are. Less about shiny tech, more about results.
The panel brought together four distinct perspectives. Iain Robertson from Fortnum & Mason, Amine Mekouar from Aroma-Zone, Nikki Baird from Aptos Retail, and Dan Chasle from New Look. Together, they covered heritage retail, digital-first brand, enterprise technology and high street transformation, which is why the discussion stayed anchored in reality.
What came through early is that agentic AI is not being approached as a way to remove people from retail. Across the panel, the emphasis was on augmentation. At Fortnum & Mason, that plays out directly on the shop floor. With tens of thousands of products across the Piccadilly store, the challenge is not access to stock, it is access to knowledge. Agentic systems are being explored as a way to give staff instant, contextual information, so someone working in one department can confidently advise across others.
At New Look, the application is more operational. The focus is on transforming internal processes, particularly in product development and supplier communication. Agentic AI is being used to capture, structure and translate information from suppliers, turning what would traditionally be manual and time consuming workflows into something faster and more consistent. This is where a lot of the immediate efficiency gains are coming from, even if they are less visible to customers.
For Aroma-Zone, the opportunity sits in understanding customer behaviour at scale. With a large, highly engaged community, the business is using AI to interpret feedback across multiple channels and feed that insight back into product development and customer experience. It reflects a shift from pushing out messaging to building a more continuous feedback loop between brand and customer.
From a technology perspective, Aptos brought a useful level of realism to the discussion. There is still a tendency to assume that these systems will scale easily once they are built. In practice, cost and complexity are becoming more visible. Every interaction with an AI system carries a cost, and as adoption increases internally, those costs can rise quickly. That is forcing retailers to think more carefully about where agentic AI genuinely adds value, rather than applying it broadly.
Control and oversight were recurring themes throughout. None of the speakers suggested that agentic systems should operate independently. Instead, the focus is on building in monitoring, feedback loops and guardrails, particularly in areas where brand and customer experience are critical. For a business like Fortnum & Mason, maintaining brand tone and authenticity is as important as functional accuracy, which adds another layer of complexity to how these systems are designed and deployed.
Partnerships are also playing a key role. Rather than building everything internally, retailers are working with a mix of established technology providers and newer, more specialised partners to move faster. That has enabled rapid testing, with multiple pilots running in parallel, and an acceptance that not everything will work. That shift towards experimentation, and the willingness to learn quickly from failure, feels like one of the more significant cultural changes happening alongside the technology itself.
What the panel made clear is that agentic AI is not a single, transformative moment. It is a series of incremental changes across different parts of the retail business. Some are customer facing, many are operational, and all of them depend on how well retailers can balance speed with control.
The opportunity is real, but so are the constraints. Cost, governance, brand integrity and the complexity of implementation all sit alongside the potential gains in efficiency and engagement. For now, the retailers making progress are the ones treating agentic AI as a way to extend what their teams can do, rather than replace it. That may be less dramatic than the narrative suggests, but it feels much closer to how this shift is actually playing out.
