like artificial intelligence When agents start making purchases on behalf of consumers and businesses, the payments infrastructure is asked to solve a problem it was never designed to address.
“The first is intent. Is the agent acting properly on behalf of the user?” xubai Chief Technology Officer Jill Willard He said to PYMNTS during a conversation for PYMNTS’ 2026 Original Series “Summer School.”
When AI agents begin searching for products, comparing offers, negotiating prices, and ultimately completing purchases without direct human intervention, payments systems inherit a different responsibility. The challenge is no longer just determining whether payment credentials are valid. It is to determine whether autonomous programs operate within the authority that the human intends to delegate.
“There is nothing in today’s payments package that thinks about intent,” Willard said. “The question has to become, not just is the agent good, is the agent’s intention right, is the agent itself good? And then the fourth thing is, does the agent act in a way that I expect the agent to behave?”
This shift is forcing payment providers to rethink everything from format and coding to compliance and fraud management. The current infrastructure has failed; It is designed for another participant in the trade.
Proxy commerce transfers payments to coordinating systems
Standalone software does not eliminate payments infrastructure; It introduces another participant for the infrastructure to coordinate. While identity has long been the organizing principle of Digital paymentsProxy trading adds several new layers.
“I kind of liken it to your neighbor,” Willard said. “You may know your neighbor, you may wave to him…but if you go to your house and he’s in your living room unexpectedly, you may become suspicious of his behavior.”
Knowing the existence of an AI agent is not enough. Payments systems must determine whether their actions remain consistent with the authority originally granted to them. Identity becomes persistent rather than binary, and the payment stack in a proxy environment is like an operating layer responsible for synchronizing identities, credentials, network rules, fraud signals, and the AI agents running across it all.
“A lot of people think coordination is guidance, but guidance is a component of coordination,” Willard said. “It’s tokenization, it’s network tokens, it’s digital wallets, it’s 3DS, it’s Authentication Redundancies, fraud controls, failover strategies, operational visibility, the list goes on and on.”
Once merchants start working across multiple payment providers, marketplaces, and payment methods, routing becomes just one decision among many.
Tokenization carries more than just credentials for agent transactions
When independent software starts making purchasing decisions, payments no longer exist at the end of the trade. They become one of the systems that govern it. This has expanded the role of tokenization, a technology that replaces sensitive card numbers with alternative values that reduce exposure to fraud.
The purpose of the tokenization was to protect credentials while allowing transactions to continue. AI trading is asking for more of it. Instead of just representing payment credentials, future tokens may also preserve the context surrounding a transaction.
“You’ll have an agent code that will have an identity,” Willard said. “You might have a network token that not only has card data, but also carries verifiable intent… You’ll need the ability to audit the transaction.”
The symbol becomes a guide and not just a mandate. The AI agent executing a purchase may identify, retain the customer’s delegated instructions, record which network credentials were used, and create an auditable record explaining how the transaction was executed.
“Having a plan that says exactly how you’re going to be able to take a transaction and replay it from start to finish and understand who initiated it from a human perspective, what agent made the transaction, what the intent of the human was…all of that needs to be encapsulated somewhere,” Willard said, adding that she expects auditability to become key to compliance in the future.
Automation increases the value of vision
The promise of proxy trading also reduces human intervention. Paradoxically, this makes vision more valuable, not less. Merchants have spent years filtering automated traffic through CAPTCHAs, bot detection, and identity verification. Now they must distinguish between malicious automation and legitimate software working for customers.
“The visibility is more than ever,” Willard said. “Gone are the days when you could get your pay package and then not be monitored.”
She said she expects merchants will increasingly rely on payments information that is able to identify unusual authorization rates, geographic anomalies, unexpected transaction patterns, and abnormal agent behavior before these issues turn into financial losses.
The fewer people approve individual purchases, the more closely organizations need to monitor the systems that make these decisions.
Independent trading does not eliminate censorship. It changes where supervision occurs.
Watch the full PYMNTS Summer School interview With Ixopay CTO Jill Willard to hear more about:
- Why AI agents are changing what payments infrastructure is designed to do.
- How the coordination of payments became the control layer in proxy commerce.
- Why might visibility be more important than automation?
For all of our PYMNTS AI coverage, subscribe to our daily newsletter Amnesty International newsletter.
