Much of the innovation in payments over the past decade has focused on making it easier for people to pay.
Contactless cards reduced friction at the point of sale. Digital wallets changed how people interact with cards and accounts. Real-time payments accelerated the movement of money. Embedded finance made the payment itself less visible within the customer journey.
Agentic payments introduce a different question:
What happens when software starts making some of those decisions on our behalf?
Cloudflare’s move into wallets for AI agents is one example of this idea moving closer to real-world infrastructure.
An AI agent could potentially be given a task, a budget and a set of rules, then interact with different services and complete a transaction for its user.
That creates new possibilities. But it does not remove the fundamentals of payments.
The payment still has to work
An agent may decide what to buy or which provider to use, but the transaction still has to move through a payment system.
There still needs to be a payment method. The transaction needs to be authenticated, processed and settled. Fraud and compliance controls still apply.
There also needs to be clarity around who is authorised to spend, what limits apply and who is responsible if something goes wrong.
This is why agentic payments are not simply an AI topic. They are also a payments topic.
The infrastructure underneath the agent matters
When a new technology arrives, much of the attention naturally goes to what is new.
In this case, that is the agent.
But understanding what the agent changes requires an understanding of the infrastructure it is interacting with.
For payments professionals, that includes digital wallets, payment rails, real-time payments, settlement, fraud, regulation, compliance and newer developments such as stablecoins.
Without that context, it becomes difficult to separate what is genuinely new about agentic payments from what still depends on existing payments infrastructure.
This is one of the areas covered in CFTE’s Payments in Digital Finance Specialisation, which gives professionals a broader understanding of the payments ecosystem and the technologies reshaping it.
Explore Payments in Digital Finance
What does AI actually change?
The more significant change may happen around the decisions made before and during a payment.
An AI agent could potentially compare providers, choose between payment options, determine whether a transaction falls within an allocated budget, initiate a payment or respond when a transaction fails.
That introduces a different set of questions.
How much authority should an agent have?
Which decisions can safely be automated?
When should human approval still be required?
How do fraud controls need to adapt when more decisions are automated?
And what happens when an agent makes the wrong decision?
These questions sit at the intersection of AI and payments.
For payments professionals, the challenge is not simply understanding what AI agents can do. It is understanding how those capabilities interact with the infrastructure, controls and responsibilities that already sit behind a payment.
