Visa says AI agents may soon negotiate business deals

Visa has published research on the use of artificial intelligence in commerce, suggesting many businesses are preparing to let AI agents negotiate and transact on their behalf.

The survey, conducted with Morning Consult, found that 53% of US business decision-makers would allow AI agents to negotiate prices or terms directly with other AI agents. It also found that 71% would tailor products, offers and experiences for AI agents, while 77% are already using or piloting AI in their operations.

Visa describes the trend as business-to-AI, or B2AI: a model in which AI agents play a more active role in commercial decisions and execution, while people remain responsible for outcomes. The data suggests a shift beyond AI as a back-office tool or recommendation engine toward its use as a participant in transactions.

Among business respondents, 88% said they were willing to provide pricing or inventory data to enterprise AI systems. More than half (55%) said they were already familiar with the concept of B2AI commerce.

Consumer findings were more mixed. The survey found that 58% of Americans are comfortable with AI comparing prices, 55% with AI applying discounts, but only 38% with AI completing a purchase.

There was stronger resistance to letting AI spend freely. Just 27% said they were comfortable allowing AI to spend money autonomously without limits, while 60% said they would not allow AI to spend any amount without approval.

Visa's data also suggests AI tools are already influencing shopping behaviour. Nearly 40% of Americans surveyed said they had made a purchase they would not normally have considered because of an AI agent or tool.

Trust factor

Trust emerged as a central theme in the research. Consumers were more likely to trust AI systems linked to established financial institutions than independent agents.

The survey found that 36% trust bank-backed AI systems and 35% trust payment network-enabled AI systems, compared with 28% who trust independent AI agents. That gap may prove significant as companies assess whether customers will let software handle more of the buying process.

Frank Cooper III, Visa's chief marketing officer, said the industry is entering a new phase of machine-led interaction in commerce.

"Commerce is moving from market-to-human to market-to-machine," Cooper said.

He added that trust would determine how far and how quickly the model develops.

"B2AI describes what happens next as AI agents begin evaluating, negotiating and transacting on behalf of people. In that world, as always, trust becomes the critical infrastructure. If we don't build it into machine-mediated commerce, adoption stalls," Cooper said.

Younger users

The results point to a generational divide in acceptance of AI in payments and shopping. Younger consumers appear more comfortable with AI systems linked to payment networks than older groups.

Nearly half of Gen Z respondents (48%) said they trust payment network-enabled AI systems, compared with 20% of Boomers. Among Gen Z and Millennials who use AI shopping assistants, nearly half reported buying items they would not otherwise have considered because of AI recommendations.

The findings come as payments groups, banks and technology companies examine how generative AI and agent-based software could reshape online shopping, customer service and procurement. A system in which software tools search, compare, negotiate, and complete transactions would shift some commercial influence away from traditional marketing channels and toward the rules and preferences embedded in AI models.

That raises questions for merchants and payment providers about how products are presented to machines, how prices are disclosed and how customer approval is managed. The survey suggests businesses are already thinking about those issues, even as consumers remain cautious about giving AI direct control over spending.

"The message is unmistakable: people are open to AI acting for them, not instead of them," Cooper said. "Our findings show that trust is the adoption switch for agentic commerce. Consumers are willing to let AI act on their behalf, but only when they retain visibility, control and the ability to intervene."