What if you didn’t actually decide to buy that last thing in your cart?
A report from Visa released on Thursday suggests that, in some cases, you might not have.
According to a survey from the financial services company, artificial intelligence is no longer just helping people shop. In many cases, AI is starting to shape what people buy, and in some cases, even act on their behalf.
The research is based on surveys of both U.S. consumers and business decision-makers. It shows that AI systems are moving from assistants to participants in commerce.
That influence is already showing up in everyday behavior.
Nearly 40% of Americans say they have made a purchase they would not have otherwise considered because of an AI tool. The impact tends to happen early in the shopping process, where AI surfaces options, compares products, and narrows choices before a person checks out.
AI becomes a new kind of customer
For Visa, this shift represents a new type of customer.
“AI is not just technology. It’s a new customer segment. Your next cancelation or your next review or your next booking of travel will increasingly be made by an agent, or at least with the help of an agent,” Visa CMO Frank Cooper III tells Fast Company.
As that dynamic takes hold, companies are no longer just competing for human attention. They are also competing for machine selection.
The report refers to this next phase as “B2AI,” where companies are selling not just to people but also to AI agents acting on their behalf.
Businesses are already preparing
More than half of business leaders surveyed say they would allow AI systems to negotiate prices or terms directly with other AI systems. Many are also willing to share data, such as pricing and inventory, to support those interactions.
At the same time, most organizations say they are already using AI in some capacity or testing it across functions like marketing, e-commerce, and payments.
The shift is also changing how companies think about persuasion. Traditional approaches that rely on emotion or limited attention don’t translate the same way when the “customer” is a machine.
“Well, the machine has no emotions, right?” says Cooper.
Instead, companies are focusing more on making information clear, structured, and easy for algorithms to process.
Consumers are on board, but cautious
Consumers are open to using AI in shopping, but with limits.
According to the report, most people are comfortable letting AI handle tasks like comparing prices or applying discounts. More than half say they are okay with those uses.
That comfort level drops pretty dramatically when it comes to AI actually spending money. Only about a third are comfortable with AI completing a purchase, and fewer are willing to let it spend without approval.
That gap points to a broader issue: trust.
“Agentic commerce does not scale unless consumers cross that trust threshold,” says Cooper.
While 53% of consumers are already using AI to help them shop, only 27% are comfortable with fully autonomous behavior. The hesitation often comes down to control, visibility, and risk.
Trust is the sticking point
Trust is likely to determine how quickly this type of commerce grows.
“That trust is the scarce commodity in this new AI agentic commerce environment,” Cooper says.
Similar patterns have played out in earlier shifts in financial technology, where adoption picked up once systems became more familiar and widely used.
“And so we think something similar is happening here,” he says.
Consumers say they want the ability to intervene, reverse transactions, and control spending. They also tend to trust AI systems tied to financial institutions or payment networks more than independent tools.
Brand still plays a role in that environment. “Brand is always important because it’s shorthand for something meaningful to people,” says Cooper.
That matters especially when people are unsure how systems work or what happens if something goes wrong.
Humans still shape the experience
The report also suggests AI is unlikely to fully replace human-driven shopping—at least not for a while.
“I do not believe, and we’ve seen no evidence, that people, consumers, want exclusively an experience where AI agents are buying everything,” says Cooper.
Instead, different types of purchases may be handled differently. Routine or low-stakes items may be delegated to AI more often, while more personal decisions remain hands-on.
“The more it’s just a basic commodity with no emotional energy in it . . . the more the machine or the agent will buy it,” he says.
So while you might use AI to buy your next bottle of laundry detergent, you might be less likely to have it buy your summer wardrobe without okaying the purchase.
For now, businesses are building toward a future where AI plays a larger role in transactions, while consumers are gradually deciding how much control they are comfortable handing over.