The consumer version of agent commerce imagines an agent who knows what a person wants to buy. The B2B version has a bigger prize and a harder task. You need to know which invoice can be paid, through which portal, on what terms, at what cost, and whether the supplier on the other end will take the card without charging a fee. Then he must implement it without making a mistake.
This is a different problem than recommending a jacket, which is a much bigger problem. The spending that moves through accounts payable dwarfs the consumer checkout process, which is why the largest TAM in agent commerce isn’t hiding in the shopping app. It exists within supplier portals, AP workflows, ERP handoffs and card acceptance rules that vary by buyer, supplier, geography and context.
In a recent conversation with PYMNTS CEO Karen Webster, Guy fakeco-founder and CEO of sunlightmade a claim that reframes the entire B2B card conversation: suppliers that accept cards without additional fees actually exist in far greater numbers than buyers assume. The market has never had a way to see them.
The real problem was fragmentation, not acceptance
For years, the decline in B2B card adoption has been blamed on suppliers who won’t accept cards. Ziff believes the industry is diagnosing the wrong disease.
“We think the acceptance rate for the card is actually very high,” he said. “The problem is implementation.”
This changes where the value is. If the constraint is the supplier’s desire, the solution is sales and incentives. If the ecosystem is fragmented and hides acceptance behind thousands of non-standard gateways, the solution is software that can see, decide and act. By Ziv’s reading, most suppliers already accept cards, and many accept them without additional fees. Buyers cannot find one portal at a time.
This fragmentation is the point. Acceptance is not a clear yes or no stamped on the supplier. The same supplier can treat buyers differently.
“Supplier “And we have that intelligence by design.”
The AI agents will independently determine that there is a portal associated with that specific invoice or bill payment, and they will go and navigate that portal in real time, determine whether the card was accepted, whether there is a fee associated with that payment and confirm it back to our customers,” Ziff said.
“Determining whether there is a gateway and accepting the card is not enough,” he added. “Some suppliers out there are adding fees or surcharges to the card payment. So being able to intelligently determine whether or not there’s a fee, using AI, basically opens up this whole process to be fully autonomous and automated.”
There are two reasons to access the card, and they are not the same thing
Finding acceptance without additional fees is important because the card is asked to do two very different functions in B2B, and the economics of each are worth separating.
The first is the classic method: pay on day 60. Use the card to extend the terms and keep the cash. The buyer wants the float and the discount, but someone has to absorb the exchange, and it’s usually the supplier who is asking him to take it. This is a tough sell. The supplier has little reason to swallow the fee so the buyer can pay later (or late), which is why the industry has spent years trying to convince suppliers to accept cards at all. This is where free discovery has its place. If the agent can find suppliers who actually accept no fees, the 60-day card ceases to be a negotiation and becomes a routing decision.
The second is more interesting. Pay on the 5th (or 15th, but you get the point). Here the card is not a stalling tactic. It is a working capital tool. The buyer pays early, the supplier gets cash faster, and the cost of the card becomes a cheap form of financing versus alternatives. For suppliers who lack easy access to credit, early payment financed through a card can be the cheapest capital, and everyone in the chain has a reason to say yes. The fee ceases to be a tax that the supplier resents and becomes a liquidity price that the supplier wants.
They’re opposing value propositions wearing the same plastic, so to speak. One asks the supplier to give up something. The other hands the supplier something he needs. Proxy execution is important because it can differentiate them from one another at scale, direct each invoice to the path it actually specifies, and show the no-additional-fee suppliers that make the 60-day case viable in the first place.
Trust architecture, not better models
None of this works without control. Autonomy in regulated financial workflow is only valuable when accuracy and compliance are strong enough to be trusted.
“Consumer proxy commerce faces a behavioral challenge,” Ziv said. “You have to trust the agent first that his taste closely matches yours and he will make the right decision, and then you will get paid.”
He pointed out that the arrangement reverses in the B2B field. Payment does not simply follow the decision. It is part of the decision infrastructure, holding working capital, supplier relationships, compliance, and revenue for AP platforms and banks.
“What AI actually enables is accuracy, speed and scale, and then you’re not building integrations. You’re building an intelligent agent that kind of thinks like a human, understands what it sees and makes some intelligent decisions,” Ziff said.
Ziff talks about keeping AI “on a leash.” Not being allowed to pay the wrong bill and putting the right controls in place was a major milestone for Sunlight.
The invoice becomes the shopping cart
The retreat and the card are one piece of the larger transformation. As supplier onboarding, procurement, invoicing, and payment move online, the traditional AP process collapses into something that looks more like a B2B shopping cart. Order, approval, invoice and payment cease to be a series of separate deliverables and become a single workflow.
This breakdown is important because B2B is full of artificial layers. Purchases fall apart from invoices. AP is located far from AR. Payment is implemented following previously made decisions. Every delivery results in delays, lost information, and compromises. An invoice, in this context, is nothing more than a request for payment awaiting verification.
“An invoice is a payment request model in a B2B workflow,” Ziff said. “Building that bridge, that layer of communication between the two sides, it’s an exciting space to be in.”
Once the invoice behaves like a shopping cart, the agent becomes the layer that connects the parts at scale, revealing payment opportunities buried in the long tail. Large vendors and big ticket payouts have received attention because they justify manual enablement. Mid-market sellers and segmented portals sat outside the economics of the old model.
“If you think about the middle market and the long tails, the beauty there is that until today, the manual processes to enable supply have been very expensive,” Ziff said, noting that proxy execution shifts the cost curve of monetizing that segment.
So, the future of the agent business may not start with an assistant buying a pair of shoes. It might start with an invoice, a supplier portal that already accepts no-fee cards, a virtual card, and an agent trusted enough to pay the invoice on the most reasonable terms.
a witness FULL TV INTERVIEW WITH PYMNTS With Sunlight co-founder Guy Ziv to hear more about:
- Why do more suppliers accept cards without additional fees than buyers think? Ziff says acceptance is high, fragmentation is the real bottleneck, and that agents can map acceptance at no extra charge across thousands of non-standard gateways.
- Why pay on day 60 and pay on day 5 are two different actions. The card, as a tool for extending terms, asks the supplier to absorb the fees. As a working capital tool, it gives cash-starved suppliers the cheap form of financing they really want.
- Why AP is the largest TAM in proxy trading. The spending that goes through accounts payable dwarfs the consumer payment process, and smart agents eventually make it possible to access the long tail of that spending economically.
- Why confidence and control are the real tests Deploying AI in B2B payments requires accuracy, compliance, and safeguards, because an agent cannot pay the wrong seller, invoice, or amount.
