Boots Is Building for an AI Discovery World. Data Is the Foundation.

The rules of retail discoverability are being rewritten. Not gradually, not eventually: the shift is already underway, and the retailers who are still optimising for the last era of search are building on ground that is moving beneath them.

Amish Mehta, Director of AI and Agentic Commerce at Boots, made this case plainly at Shoptalk Europe 2026 in Barcelona. The session, titled “Getting Seen in the Age of AI Commerce,” was not a horizon-scanning exercise. It was a practitioner’s account of what one of the UK’s largest health and beauty retailers is doing right now to position itself across an ecosystem it cannot fully predict.

Boots is within 10 minutes of most of the UK population. At that scale, questions about how customers find you, and how you show up when they are not looking for you by name, are not abstract. They are operational.

The cognitive load problem has already been solved by AI

Mehta’s answer to whether the industry is genuinely moving into a post-traditional-search era was unequivocal. “I think we’re already there,” he said. The old search journey, Google query to website click to sidebar filter to internal site search, carries a cognitive load that consumers are increasingly unwilling to absorb. AI assistants compress that into a single exchange. “For myself personally, I’m using a multitude of AI assistants to help me find the information I’m seeking, and no doubt consumers are doing the same.”

For a retailer with over 50,000 products and a substantial services proposition, this is a significant shift. The question is no longer purely how you rank in a list of ten blue links. It is whether you are present, authoritative, and trusted enough to be surfaced when a customer asks an AI what they should buy for a specific health condition, beauty routine, or wellness goal.

Authority, relevancy, and trust are the new ranking signals

The signals that drive AI recommendation, as Mehta described them, are authority, relevancy, and trust. These are not new concepts. What is new is the mechanism by which they are expressed and rewarded. LLMs are trained on publisher sites, crowd-sourced platforms, and community content. A retailer that has historically relied on traditional SEO and paid search to drive traffic has to think differently about how it earns presence in that training environment.

Boots’ response has been to treat its own workforce as a content asset. Pharmacists in stores, commercial teams at head office: these people carry specialist knowledge about health, beauty, and wellness that no publisher site can replicate at scale. Getting them to contribute to a bank of content that guides customers, whether those customers arrive at boots.com or are asking OpenAI for a recommendation, is the practical expression of an authority-building strategy. “We’re starting to see a large proportion of customers who would have typically done a lot of that research on publisher or health sites now coming to Boots as a primary source,” Mehta said. That is an early signal worth watching.

[SPONSOR ANGLE: The Fospha positioning around data infrastructure as the foundation for everything downstream connects naturally here: Mehta’s argument that data is the path of no regrets mirrors the Fospha argument that measurement infrastructure built now is what AI-driven decisioning will run on tomorrow.]

Agentic commerce is not a future problem

The concept of agentic commerce, where AI agents complete transactions on behalf of customers rather than simply surfacing recommendations, has not yet fully landed in the UK market. Mehta was candid about that. But the framing he used was instructive: rather than waiting for a dominant platform to emerge, Boots is building to be platform agnostic.

The specific challenge is interoperability. An agentic commerce infrastructure that works only within one ecosystem, whether that is Google UCP, OpenAI’s ACP, or a proprietary in-app experience, creates a strategic dependency. What Mehta described is an approach that asks: how do we take what is genuinely valuable about Boots and make it available across whatever ecosystem ends up driving consumer behaviour? That means clean data, accessible product inventory, pricing information, and contextual content, structured so that any ecosystem can ingest it.

On-site, Boots is also thinking about what an AI assistant experience looks like for customers who arrive at boots.com. The principle is the same: guide the customer through the experience rather than presenting them with the cognitive load of navigating it themselves.

Change management is the capability most people underestimate

Mehta was direct about where the real difficulty lies. Technology is not the binding constraint. Culture is. “Where you really get value from AI is by empowering your teams to test, learn, validate, and explore,” he said. The approach at Boots has been to identify change agents within the organisation and double down on enabling them, rather than issuing top-down mandates.

This matters because Boots sits at an unusual intersection: it operates both as a retailer and as a brand owner. The AI capability required to serve those two roles is not identical, and the teams involved are different. Building the kind of distributed, workflow-embedded AI capability that produces lasting change requires champions at every level of the organisation.

What this means for senior marketers

Mehta’s closing argument was also his most durable: in a landscape where models, protocols, and ecosystems are changing at speed, the path of no regrets is data. Businesses that scrambled to integrate into a specific AI platform at the end of last year found that the protocol had shifted by the time the integration was complete. Businesses that invested in clean, accessible, well-structured data have something that transfers regardless of which platform wins.

In an agentic world, the LLMs are looking for product inventory, pricing, and contextual information about why a specific retailer is the right answer to a specific question. Retailers with fragmented systems, inconsistent data definitions, and siloed infrastructure will struggle to supply that. Retailers who have done the unglamorous work of getting their data in order will find that the AI era is built on foundations they already have.

The brands that get seen in the age of AI commerce will not necessarily be the ones who moved fastest. They will be the ones who built correctly.