Marketers have spent generations learning how to influence human shoppers. The next challenge is persuading something that does not have feelings, preferences, or a favourite colour. The customer of the future may be an artificial intelligence agent acting on behalf of a person, and it judges brands by an entirely different set of rules.
Tati Lindenberg, global chief marketing officer of Unilever's homecare division, frames the shift in retail terms. "The store, online or offline, has always been where the decision is made. The AI answer is becoming a new kind of shelf," she says. In that environment, a brand must be recognised, trusted, and preferred by both people and machines.
Adoption is accelerating
Consumer use of AI for shopping is no longer experimental. In the United Kingdom, 52% of shoppers have used AI for at least one shopping task in the past three months, according to McKinsey research. In China, where adoption is more advanced, the figure rises to 80%.
Most consumers still use AI only for research and make the final purchase decision themselves. That is expected to change quickly. Mastercard forecasts that one in ten people will routinely use AI agents to shop and pay by 2030. Meta is preparing to launch its personal agent, Muse, in the United Kingdom, with capabilities that include researching products and completing purchases.
Mintel data suggest household products are especially vulnerable to automation. Forty-one percent of consumers say they would feel comfortable delegating regular purchases in this category to an agent. The reason is practical: these are repeat buys with lower emotional stakes, making them ideal for hands-off commerce.
Why brand size no longer guarantees visibility
AI agents operate differently from search engines. Their recommendation algorithms are often opaque, and the signals they value are not the same ones that push a brand to the top of a Google results page. Fredrik Kridahl, founder of AI search consultancy Hyper Agency, notes that only 38% of URLs cited by AI rank in the top ten on conventional search, while 31% sit outside the top 100.
"AI does not care about how big your company is," Kridahl says. "It cares about the facts and giving the right advice." That levels the playing field in unexpected ways. A smaller competitor with cleaner data, stronger reviews, and clearer product attributes can outrank a household name when an agent is choosing what to recommend.
Queries are also becoming longer and more specific. A typical Google search averages 3.2 words, Kridahl says, while an AI prompt averages 91 words. Consumers are no longer typing "best detergent." They are describing their entire situation and asking for a tailored recommendation.
What marketing to machines looks like
The response from brands is a new discipline sometimes called business-to-agent, or B2A, marketing. The goal is to make product information accurate, detailed, and structured so AI systems can understand, compare, and recommend it.
JD Sports has built technology that lets U.S. shoppers buy through Microsoft Copilot, Google's Gemini, and ChatGPT in a single click. John Lewis has announced a major investment in AI-powered shopping to ensure its products surface on those same platforms. Very Group digital director Sarah Hanna says AI search currently accounts for less than 2% of traffic to its site, but she expects that figure to grow rapidly.
"Getting in front of AI agents is set to become just as important as traditional SEO," Hanna says. "Retailers need to ensure their product information is accurate, detailed and structured so AI can understand, recommend and compare their products effectively."
Trust and emotion still matter
Marketing to machines does not mean abandoning human connection. EY research found that 47% of consumer products executives believe the ability to influence AI recommendations will be essential to competitiveness over the next five years, yet only 20% feel ready to do so today. The gap matters because even small shifts in consumer behaviour can have outsized revenue effects.
Experts argue that emotional branding becomes more important, not less. Consumers who trust a brand are more likely to let an agent automate purchases from it. Agents, in turn, may feel more confident recommending products with strong reviews and positive social signals. The challenge is translating emotion into machine-readable evidence without losing the creativity that makes a brand distinctive.
"The danger is optimizing so hard for algorithms that we lose the distinctiveness that makes a brand worth choosing," Lindenberg says. As AI makes content easier to produce, she adds, taste, originality, and human imagination become even greater competitive advantages.
The practical agenda
For now, the work is largely operational. Brands are cleaning up product feeds, improving metadata, monitoring Reddit and review platforms, and making sure pricing and availability data are current. They are also learning how to describe products in ways that capture both functional details and emotional context.
Expedia is applying the same thinking to travel. A hotel might feel right to a person browsing photos, but an agent needs different evidence, such as guest descriptions of atmosphere and amenities that match a quiet weekend or a family trip. Turning experience into structured data is becoming a core skill.
The brands that succeed will be those that can speak fluently to both audiences: the human who sets the goal and the agent that executes it. The shelf is changing. The question is whether the product on it can be understood by the shopper of tomorrow.