AI Agents Now Pay Suppliers. Your Vendor Data Decides Who Gets Paid.
Payment agents move faster than the vendor records behind them
Most organizations believe supplier payment risk lives at the approval step. If the right person signs off on the right invoice, the money goes to the right place.
The larger problem is the vendor record behind that approval. Last week, Visa, BILL and Billtrust launched a live pilot in which an AI agent completes approved supplier invoice payments end to end through supplier portals, using tokenized credentials instead of a human login, according to Visa's announcement. The agent does not make payment decisions. It executes what buyers have already approved, The Neuron noted.
That moves a recurring question to the front of the queue. When an approved invoice points at "this supplier," which supplier record did the approval actually rely on: the ERP vendor, the procurement profile, or the copy someone updated after an email last Tuesday?
And it gets worse as payments speed up. Every step a human used to perform by hand was also a chance to notice something odd. An agent removes that step, so the vendor data has to be right before anyone presses approve.
Agents that pay suppliers will need to find, verify and trust other agents, not just log in to portals. That is what we are building at OpenMap: The Agent Network, the connective tissue for the world's AI agents, starting with every person and every local business. We are raising a £5M seed. Meet us at openmap.ai/investors.
Why Does an Approved Invoice Still Go to the Wrong Place?
Because approval checks the invoice, not the supplier record it points to.
Picture an accounts payable team at a mid-sized manufacturer. A long-standing supplier exists twice in the ERP, once from before an acquisition and once after. A month ago, an email asked for a change to payment details, and someone updated one of the two records.
The invoice is approved. The agent pays it, quickly and exactly as instructed.
The payment is on time. It is just going to whatever that record now says.
This is not a payments automation problem. It is a vendor master data problem.
The Risk Isn't New Fraud. It's Old Fraud at Machine Speed
The attacks that target payment instructions are already common. In the 2026 AFP Payments Fraud and Control Survey, 76% of organizations reported attempted or actual payments fraud in 2025.
Business email compromise affected 74% of organizations, with fraudsters mainly impersonating vendors or executives to request changes to payment instructions, the report's highlights show.
An agent does not make that attack more likely. It makes the gap between a bad vendor change and a completed payment shorter.
Before an agent can safely pay suppliers on your behalf, someone has to:
Find every system that holds a version of each supplier
Merge duplicate suppliers created by acquisitions, regions or one-off purchases
Decide which source wins for remit-to and payment details when two disagree
Assign an owner who must confirm any change to payment instructions
Track which agents, approval flows and reports depend on each supplier record
None of these tasks are particularly difficult. The problem is that they are usually done once, during an ERP migration or an audit, while new supplier records keep appearing every week.
How to Give Your Payment Agents One Trusted Supplier in 30 Days?
You do not need to replace your ERP or pause your accounts payable automation to do this.
Within the first month, focus on the suppliers an agent will pay first: your highest-volume vendors, anyone with a recent change to payment details, and anyone who appears in more than one system.
Start by discovering every system that holds supplier records and resolving duplicate and conflicting records into one trusted version of each supplier. Then attach an owner, and the policy for changing payment instructions, to that record.
Next, expose this trusted record to the same approval flows and agents that already handle invoices, so a payment is checked against it before it goes out.
DataManagement.AI accelerates this process by unifying master data, governance and operational trust into one continuously maintained system. It resolves duplicate and conflicting records, maps lineage across the modern data stack and identifies downstream dependencies, and ownership and policy travel with every entity, so every agent pays the same supplier your team approved.
Instead of chasing a misdirected payment after it clears, your teams can let agents pay at speed, knowing every approval points at one verified supplier.
Agentic Payments Are Actually a Master Data Problem
Pilots like this one will spread. Supplier portals, payment platforms and agents are learning to trust each other, and that removes real friction from accounts payable.
But an agent can only pay the supplier your records say is the right one. If two records disagree, the agent does not pause to ask. It pays the one it was given.
Organizations that scale agentic payments successfully do not just automate the last click. They make sure every record behind that click is owned, current and singular.
Warm regards,
Shen and Team