Imagine asking an AI assistant to find a carry-on suitcase under $300, compare the best options, check delivery before Friday and recommend the one most likely to survive years of travel. The customer has expressed the need, but they may never open ten browser tabs or work through every product page themselves. The agent can do much of that work first.

That possibility is moving quickly from demonstration to real commerce. In September, Mastercard announced new capabilities aimed at helping merchants make product information available to AI-powered shopping experiences and support agent-led discovery, research and consumer-authorized purchases. Its guidance to merchants is telling: product details, pricing, availability and fulfilment information need to be accessible in ways that agents can find, understand and act on.

For brand and marketing leaders, that creates a new question. What happens when the first interpreter of your brand is not the customer, but a system acting on the customer’s behalf?

The first encounter may happen before the customer arrives

For years, digital customer journeys have been designed around a fairly familiar assumption. A person searches, lands on a page, scans the content, compares options and decides what deserves more attention. We have optimized headlines, navigation, imagery, product detail pages and calls to action around that behaviour.

AI agents add another participant. A person may still define the goal and make the final decision, but part of the discovery and comparison process can happen elsewhere. The agent may read structured product information, policies, availability, reviews and other signals before deciding what to surface.

That does not make the brand experience less human. It changes where part of the interpretation happens.

If an agent summarized your offer in two sentences, would it accurately represent your value? If it compares your product with a competitor’s, are the differences clear enough to survive that comparison? If a policy, warranty or availability detail is inconsistent across channels, which version will it trust?

These are not future-facing questions in the abstract. They are extensions of problems organizations already have with fragmented information and unclear value propositions. Agentic commerce simply raises the stakes because the customer may not be doing the reconstruction themselves.

Clarity has to survive translation

The temptation may be to think about this as a new form of optimization: how do we market to the machine? That is probably the wrong starting point.

The better question is whether the substance of the offer is organized well enough to travel. AI agents cannot compensate for a brand that has not decided what matters, a product catalogue that contradicts itself or a website that hides essential information behind vague language. They can only work with the signals available to them.

This is where brand strategy, content structure and digital experience begin to overlap in a new way. A strong brand still needs a clear promise. Product information still needs to be accurate. The relationship among offers still needs to make sense. What changes is that some of those signals may now be interpreted by a system before they are experienced by a person.

There is a useful connection here to a lesson we have seen in brand architecture work: customers should not have to organize your brand for you. The same principle becomes even more important when an intermediary is doing part of the organizing. If the relationships among products, services and platforms are unclear, an agent can only infer what the organization has failed to make explicit.

Trust still belongs to the human

The other side of this shift is trust. Visa’s September 2026 Trust Index found that AI use is already common among surveyed U.S. consumers, but only 23% said they trusted GenAI to handle payment transactions on their behalf. Visa’s broader research across the U.S., Australia and New Zealand also found that people place high value on transparency and control when agents are involved in shopping decisions.

That gap matters. An agent may be able to search faster than a person, but the person still cares about whether the recommendation is accurate, whether the purchase reflects their intent and whether they can intervene when something feels wrong.

This is why Human2Human thinking still belongs in the conversation. The agent is not the person we are ultimately trying to serve. Behind the prompt is someone trying to solve a problem, make a choice or reduce uncertainty. The technology is becoming part of the route between the need and the decision.

Brands should prepare for that route without losing sight of the person at the end of it.

A practical test for the brand

There is a simple way to pressure-test this idea. Imagine a customer gives an agent a goal rather than a brand name. The agent has to decide whether your organization belongs in the consideration set and how to explain why.

Would the information it finds make the distinction clear? Could it verify the claims being made? Would product names and categories help it understand the relationships among your offers, or create more ambiguity? If the customer asked why your option costs more, is there a credible answer available in the information you publish?

Those questions are useful even if agentic commerce develops more slowly than predicted. They expose the same gaps that already make digital experiences harder for people. 

The customer journey may begin before the customer arrives

AI agents are unlikely to remove people from commerce. They may remove some of the searching, sorting and comparing that people currently do themselves. That makes the quality of the information surrounding the brand more important because someone – or something – will interpret it before the decision reaches the person.

The opportunity is not to make brands more machine-like. It is to make the meaning easier to carry.

When the customer finally arrives, the brand should still feel recognizable, credible and worth trusting. The journey may have started somewhere else, but the human decision is still the one that matters.