Something is happening to the way we buy things. Not the usual incremental shift, not a new payment method or a slightly faster checkout. Something structural. The transaction is learning to complete itself.
Mastercard launched Agent Pay: AI agents that initiate and complete purchases on your behalf. Not a chatbot. Not a recommendation engine. A buyer. Software that holds your card details, understands your preferences, and executes transactions without you lifting a finger. In early 2026, Banco Santander and Mastercard completed Europe's first live end-to-end payment executed by an AI agent. Not a demo. Not a pilot. A real transaction inside a regulated banking framework. The infrastructure for autonomous commerce is no longer theoretical. It shipped.
Which means the question that has defined retail for the past two decades, "how do I get them to the checkout?", is about to become irrelevant. When software handles the buying, the entire visual layer of e-commerce, the homepage, the product grid, the add-to-cart button, loses its purpose. All of it was built for human attention. Agents do not have attention. They have instructions.
The storefront is becoming irrelevant
Think about what an online shop actually is. It is a persuasion architecture. Colours chosen to trigger urgency. Copy written to reduce friction. Layouts optimised for the scroll. Every pixel exists to guide a human being towards a purchase. When the buyer is software, none of that matters. The agent never sees the hero image. It never reads the reviews. It never gets distracted by a sale banner. It simply matches a set of requirements against a catalogue and executes.
This is not a minor UX challenge. This is the entire front end of commerce becoming decorative. The question is no longer "how do we get them to the checkout?" It is "how do we get chosen by the algorithm that never visits our site?"
Taste as the new input layer
For an agent to buy well, it needs instructions. Not just "buy milk" but "buy the oat milk from the brand I actually like, in the size that fits my fridge shelf, from the shop that is not out of my way." That level of specificity is taste. And most people cannot articulate it. We know what we want when we see it, but translating that into parameters a machine can act on is a different skill entirely. The gap between what you want and what you can express to a machine is where the next generation of products will live.
Mastercard is already building for this. They are developing a proprietary Large Tabular Model, a generative AI engine trained on structured financial data: 175 billion anonymised transactions, merchant location data, fraud patterns, loyalty behaviour. They now operate both the advertising layer and the agent layer. Commerce Media uses transaction data to power smarter ads. Agent Pay uses transaction data to power autonomous buying. The company that processes the payment is becoming the company that decides what gets purchased.
The brands that survive agentic commerce will not be the cheapest or the fastest. They will be the most legible. The ones whose taste profile is so well defined that an agent can represent them faithfully. Every brand will need a taste brief. Not for the creative team. For the agent.
Culture is the last moat
Price can be compared instantly. Availability can be checked in milliseconds. Delivery can be optimised algorithmically. The only thing an agent cannot replicate is the cultural meaning of a choice. Why this, not that. Why now, not later. Why here, not there. Cultural context is the one input that resists automation.
This is what makes the shift so interesting for anyone who works in brand strategy. For 15+ years I have watched the discipline slowly cede ground to performance marketing, to conversion optimisation, to whatever could be measured in a dashboard. Agentic commerce reverses that entirely. When the machine handles the transaction, the only competitive surface left is meaning. The curation premium returns, not as a luxury add-on, but as the primary mechanism by which a brand gets chosen at all.
Mastercard is investing $200M and building partnerships with 85+ organisations to make this work. $1.8 billion in further commitments signal that this is not an experiment. They are constructing an entirely new trust architecture, one where machines vouch for machines on behalf of people. Identity confirmation, fraud scoring, biometric authentication. The agent needs to be trusted before it can act. Trust becomes infrastructure.
Who briefs the agent?
This is the strategic question nobody is asking yet. Someone will define the taste parameters that guide agentic purchasing. It could be the consumer, but that requires a level of self-knowledge and articulation that most people simply do not have. It could be the platform, but that creates a dangerous conflict of interest: the entity choosing what you buy is also the entity profiting from the choice. Or it could be the brand. That is the opportunity.
The brands that build the richest, most coherent taste profiles will be the ones agents default to. Not because they gamed a ranking. Not because they bought placement. Because their cultural signal was clear enough for a machine to read it. The brand becomes a set of instructions an agent can follow with confidence.
This changes what strategy actually is. It is no longer about persuading a person standing in an aisle or scrolling through a feed. It is about encoding a point of view so precisely that software can act on it. The brief does not go to the creative department. It goes to the model.
The taste engine
We are heading towards a world where commerce happens without browsing, where transactions complete themselves, and where the storefront as we know it fades into the background. In that world, the brands that win will not be the ones with the best checkout flow or the loudest campaign. They will be the ones that understood, earlier than everyone else, that taste is not a nice-to-have. It is the product.
The next billion-dollar brand will not be a shop. It will be a taste engine.
