Visible Is Not Buyable: What the MACH AI Exchange Revealed About Agent-Ready Commerce
After eight weeks of building agentic AI with easyJet holidays, Dr. Martens, and Your Golf Travel, the MACH Alliance community found that showing up in AI answers is only half the problem. Structured, specific data is what turns discovery into a sale.
One of the most useful sessions at MACH X Amsterdam was also one of the most practical. For eight weeks, members of the MACH Alliance community worked alongside easyJet holidays, Dr. Martens, and Your Golf Travel through the MACH AI Exchange. Their goal was to build agentic AI against each brand's real data and real constraints. The projects ranged from unifying review data and building an interactive points-of-interest map to triaging customer email with a human in the loop, building natural language search for trip planning, and creating an AI assistant that could carry a shopper all the way to a completed order. Kerrigan Baron, Co-Founder and CTO of Fidget Labs, presented the results on stage with the three brand leads.
Much of the industry conversation about AI shopping agents has centred on visibility, meaning whether a brand appears when someone asks ChatGPT or Perplexity for a recommendation. A phrase from the Dr. Martens team, represented on stage by Sezin Cagil, Head of Unified Commerce Technology, sharpened that conversation: visible is not buyable.
An agent can surface a product and still be unable to act on it, because the product data is incomplete, the inventory isn't exposed, or the agent lacks the information it needs to recommend the product with confidence. An assistant built to take a shopper through to a completed order exposes each of those gaps quickly.
easyJet holidays, represented by Alex Black, Head of Digital Product, worked on the trust that comes before a sale. The team focused on city-break holidays, unifying customer review data and pairing it with more specific location data through an interactive points-of-interest map. An agent recommending a weekend in Lisbon needs the kind of detail a well-informed friend would offer, such as how far the hotel is from the old town and what recent guests said about noise. That detail gives the agent a reason to choose one option over another, and it gives the traveller a reason to believe the choice.
And Your Golf Travel, represented by CMO Oliver Gunning, approached the problem from the side of customer context. One concept, which the team called "the Greenkeeper," triages customer email with a human in the loop and uses those exchanges to build richer context for future interactions. Another, nicknamed "Jack Nicklaus," would let a customer ask for an expert opinion, the way a golfer might seek advice from someone who has mastered the game. Each casts the agent as a knowledgeable advisor, and each is only as good as the structured knowledge the brand can give it.
Across the three brands, the hardest work sat underneath the agent. Reviews had to be unified before an agent could cite them, and locations had to be mapped before an agent could reason about distance. Customer history had to be organized before an agent could act on it, and the human review step in Your Golf Travel's email triage shows that the brands are still deciding where an agent's judgement should end and a person's should begin.
Same principles, different houses
At a partner day roundtable during the event, George FitzGibbons, Principal Sales Engineer at Vercel, offered reassurance in the form of an analogy for why projects can look so different from one another: "The architecture of houses is all different, so it's okay if our agent architecture is different too." Houses vary, but they're built from the same core materials and follow the same structural principles. In agentic commerce, those materials are structured data, capabilities exposed through defined interfaces, and clear rules about when a human steps in. How a business assembles them depends on what it sells, how its customers buy, and what its teams can realistically support.
JD Sports, which was a featured keynote presenter at MACH X, is building its own version. Antonia Hansen, Group Director of Product, described a roadmap that moves from existing channels to standalone LLM channels and eventually to instant purchase from anywhere a customer happens to be, social platforms included. She called agent discoverability critical to winning that last stage, and much of JD's work since replacing its monolith has gone into structuring its data so future channels can use it. JD's design will look nothing like a golf travel company's email triage or a footwear brand's shopping assistant, but the foundation underneath is formed from the same materials. Brands that lay those foundations now will have more options when buyable starts to matter as much as visible.
Leigh Bryant
Editorial Director, Composable.com
Leigh Bryant is a seasoned content and brand strategist with over a decade of experience in digital storytelling. Starting in retail before shifting to the technology space, she has spent the past ten years crafting compelling narratives as a writer, editor, and strategist.