Live Commerce
Visa is betting the whole agentic economy on one word: trust
When software starts buying for you, the hard problem isn't payment rails. It's proving the agent had permission to spend. Visa wants to own that proof.
Why it matters
Anyone building AI agents that spend money faces a verification problem before a payment problem: how does a merchant know the agent was authorized and stayed within limits? If Visa becomes the default authority layer for that, every agentic commerce startup builds on top of it rather than around it. That shifts where value accrues as this market forms, from the agent makers to the network sitting underneath them.
What changes next
Watch whether Visa's near-50 production AI partners actually double in the coming months as CEO Ryan McInerney forecast, and whether the first live agentic use case is the boring B2B one he named: automated bill pay and invoice settlement. Consumer-facing agentic checkout arriving before business settlement would signal the trust problem was easier than Visa claimed. The 2027 guidance of low-double-digit revenue growth tells you how fast management expects any of this to show up in actual volume.
If you are building an AI agent that buys things on a user's behalf, your first engineering headache is not the checkout button. It is proving to the seller that the agent was allowed to spend, spent within its limits, and belongs to a real customer. On Visa's fiscal fourth-quarter earnings call, reported by PYMNTS, CEO Ryan McInerney put it bluntly: in agentic commerce, trust is the biggest barrier to scale. That single framing matters more than any number Visa posted, because it defines where the company intends to sit in the stack.
Why the authority layer is the valuable one
There is a land grab forming around agentic commerce, and most of the attention goes to the agents themselves. Visa is going after something less visible and harder to dislodge: the mechanism that confirms an agent's authority. McInerney described the work as identifying an agent, tying it to a customer, and verifying its spending matches that customer's instructions. Whoever owns that verification becomes a dependency for everyone else. An agent builder can swap model providers in an afternoon. Swapping the system that merchants trust to confirm authorization is a different order of commitment.
Visa already has the raw material. More than half its 5.2 billion credentials are tokenized, and over 60% of its eCommerce transactions run through tokens across 200 countries. Tokenization was built to let a transaction happen without exposing the underlying card, which is close to the shape of the problem an agent creates: a third party acting on your account without being you. Visa is adapting infrastructure it already runs rather than inventing a category from scratch, and that is a real head start.
The internal tell
The most revealing detail is not customer-facing. McInerney said 12 people shipped stablecoin-platform functionality in seven weeks, and that over 60% of Visa's product and engineering teams now use agentic tooling. Read that as a company that believes agents compress real work, not just a company selling the idea to analysts. If agents cut Visa's own delivery timelines this sharply, its confidence that businesses will delegate payments to software is grounded in something it has tested on itself. That is more persuasive than the partner-count forecast, which is a projection and could slip.
The catch
Trust as a strategy has a trap built in: Visa has to be right that trust is the bottleneck. If the actual friction turns out to be merchant acceptance, or agent reliability, or consumer willingness to hand a bot a spending limit, then the authority layer solves a problem that was not the one holding things back. The B2B starting point McInerney named, automated bill payment and invoice settlement, is telling. It is low-drama, rules-heavy spending where authorization is already structured, which is exactly where a trust layer is easiest to prove and least likely to embarrass anyone.
There is also a timing gap. Visa's 2027 guidance is low-double-digit adjusted revenue growth, with agentic commerce named as one investment area among several alongside stablecoins and business payments. That is the language of a long build, not an imminent surge. The credential growth of 6% and the firm consumer spending are today's business. Agentic commerce is a bet on who controls the plumbing of a market that barely exists yet. Visa is positioning early, which is the right move, but early positioning and realized revenue are years apart.
Source: PYMNTS.com