Business payments have a bookkeeping problem that’s been keeping them manual for centuries. Millennia, even.
In 2026, after one company approves an invoice, someone often must authenticate into a supplier portal, enter payment credentials and complete the transaction. Even two digitally native, forward-thinking CFOs doing business with one another can find it challenging to streamline and scale the need to authenticate and reconcile high-value and sensitive B2B payments.
Agentic AI, however, might just hold a long-awaited solution. One that the payments and FinTech marketplace is rapidly working toward. On Monday (Nov. 9), for example, payments technology company Clink added support for Mastercard Agent Pay, adding to its existing Visa Intelligent Commerce capabilities.
The development follows an October Visa initiative involving both BILL and Billtrust which allowed authenticated AI agents to complete previously approved invoice payments through supplier portals.
Both initiatives address some of B2B’s most stubbornly manual elements. And while AI can eliminate many final interventions across payments, doing so introduces more consequential questions, such as who authenticates the agent, defines its spending authority and assumes responsibility when an automated payment goes wrong?
Card networks like Visa and Mastercard are moving to establish the infrastructure that answers those questions.
Read more: Why the Fed Thinks B2B Payments Are Already Built for Agentic
Authorization Becomes the New B2B Payment Battleground
For decades, payment networks have competed over acceptance, transaction volume and the economics of moving money. Agentic commerce adds a new competitive layer: establishing whether a machine has permission to initiate a transaction.
“Authentication remains foundational in an AI landscape,” Chris Trainor, head of platform strategy, innovation and AI at Paymentus, told PYMNTS in an earlier discussion of agentic AI and payments. “The mistake that many have is expecting that model itself to become the system of record or a transaction processing engine. The future is not unconstrained artificial intelligence. It is governed agency.”
The operational advantage of card networks isn’t that they process payments but that, in doing so, they already operate the credentials, authentication systems and commercial rules connecting banks, buyers and suppliers. The important distinction is that the agent isn’t deciding whether the invoice should be paid. It’s executing a payment decision already made by the business.
Automating execution allows companies to preserve established approval processes. Delegating payment decisions introduces substantially more complicated questions about corporate authority, fraud and liability. For now, the networks are concentrating on the first opportunity of making existing financial instructions executable by trusted software.
Read more: AI Agents Need Permissioned Funding Sources. Not Company Bank Accounts.
Clink’s integration with both Mastercard and Visa highlights an emerging competitive dynamic. Mastercard’s Agent Pay initiative, similar to Visa’s, combine tokenized credentials, authentication and transaction controls to support payments initiated by AI agents.
The PYMNTS Intelligence report “How Agentic AI Went From Zero to CFO Test Runs in 90 Days“ found in September 2025 that nearly 7% of enterprise CFOs in the U.S. had deployed agentic AI in live finance workflows, while an additional 5% were running pilots.
Software developers want the ability to build purchasing experiences without being confined to one payment network. The networks, meanwhile, want their credentials and authorization capabilities embedded wherever those experiences operate.
An enterprise software provider might determine when an invoice is ready for settlement. A bank might provide the funding. A card network might authenticate the agent and enable the transaction. A supplier platform might determine whether the payment can be accepted. Each participant has an opportunity to control a different part of the workflow.
The networks are attempting to make authorization their indispensable contribution.
Read more: Agentic AI Could Make Net 30 Obsolete
Banks Could Win Even as Payment Decisions Move Into Software
The early advantage of card networks doesn’t mean banks are losing relevance. In fact, agentic payments could make banks more valuable, provided their products and controls are embedded in automated workflows.
Banks already supply the credit, operating accounts, liquidity services and financial relationships that underpin corporate payments. As agents begin selecting payment methods, those capabilities could become inputs.
“Payment execution alone is increasingly commoditized. The durable value as I see it today is in making the right payment decisions,” Tungsten Automation Head of Payments and Embedded Finance Andrew Ng told PYMNTS for the September edition of the “What’s Next in Payments Series: The Fall Draft.”
An agent might eventually choose between a commercial card and an account-to-account transfer based on supplier acceptance costs, available credit, cash flow requirements or payment terms.
Of course, making agents recognizable to payment networks solves only part of the problem.
A properly authenticated agent can still execute an incorrect payment. For CFOs, that means the economic case for agentic payments must also account for exception handling, payment recovery, dispute resolution and the cost of maintaining controls across multiple software and financial providers.
The companies that solve those problems could establish a competitive advantage that extends beyond transaction processing.
For all PYMNTS B2B and AI coverage, subscribe to the daily B2B and AI newsletters.
