The Agent Knocked. Santander Just Built the Door.

The most important agentic payments story this week was not a protocol or a keynote. It was an acquirer quietly turning on the other side of the transaction.

Santander's Getnet, the bank's global merchant payments platform, launched new AI-powered capabilities that let businesses accept payments initiated by agents. For a year the agentic commerce conversation has been dominated by the buy side: the assistants that shop, the wallets that pay, the protocols that carry intent. This is the sell side starting to arm. And the sell side is where agentic commerce has been stuck.

An agent can be brilliant at reaching checkout. None of that matters until the merchant on the other side is willing to take the payment it presents.

Why the merchant side is the hard side

Picture the flow. An agent fills a cart, reaches a merchant it has never transacted with, and presents a payment. The merchant's systems were built for two cases: a human at a browser, or a known recurring biller. The agent is neither. It looks, to a fraud engine tuned over a decade, like exactly the thing that engine was built to stop.

So the merchant has two bad options. Wave the agent through and absorb the fraud risk of an actor it cannot identify, or challenge it and break the automated flow the customer wanted in the first place. Friction up, or fraud up. That is the wall agent-led checkout has been hitting, and it is the same wall an operating system runs into the moment its agent tries to act on infrastructure it does not own.

Getnet's move is an acquirer saying it will take responsibility for that decision on the merchant's behalf. The merchant does not have to learn what an agent is. The acquirer does, and it absorbs the question of whether this particular agent-initiated payment should clear.

This is a land grab, not a feature

Acquirers do not usually race to be first on an unproven flow. When one does, it is staking a position.

The agent economy needs a small number of trusted intermediaries who can vouch for an agent-initiated payment at the moment it lands. Whoever occupies that role at scale becomes infrastructure, the way the card networks became infrastructure for the human web. Getnet is betting that merchant-side trust is a durable position and moving before the standard is settled, the same logic driving the protocol authors racing to define agentic commerce from the buy side.

The open question, and the one to press Getnet and everyone who follows on, is what the acceptance actually rests on. Is the agent verified through a network credential, a protocol attestation, or Getnet's own risk models reading behavioral signals? A merchant accepting agent payments on a proprietary risk score is a very different thing from a merchant accepting a cryptographically verifiable proof that an agent acts on a named user's authority. The first is a bank extending its fraud appetite. The second is the trust layer the whole ecosystem has been waiting for. Only one of them scales past a single acquirer.

What it signals

Three things follow from this, and they will repeat.

First, the competitive front in agentic commerce just widened from wallets and protocols to acquirers. Expect the other global acquirers and the networks to answer, because being the place merchants go to accept agent payments is too valuable to cede.

Second, the merchant-side build is where the trust layer stops being a slide and becomes a product. The buy side can demo an agent shopping. The sell side has to decide, in milliseconds and with money on the line, whether to trust it. That decision is where verifiable agent identity goes from nice to necessary.

Third, the institutions positioned to win are the ones that already sit at the trust boundary of every transaction and already issue credentials at scale. That description fits the acquirers and the networks far better than it fits the operating systems. Apple built the agent. Santander is building the door. The question for the networks is whether they supply the lock, or watch acquirers improvise one merchant relationship at a time.

If accepting an agent's payment becomes a service a merchant buys, who do they buy it from, and what exactly are they trusting?

Charlie Major is a Product Development Manager at Mastercard. The views and opinions expressed in Major Matters are his own and do not represent those of Mastercard.