I have just published my latest deep dive where I break down why the current payments stack is structurally incompatible with a world where AI agents transact on behalf of users.

What most people are missing is that this is not another checkout upgrade or UX improvement, it is a fundamental shift in how liability, trust, and control are distributed across the system.

As soon as an AI agent enters the flow, the assumptions the industry has relied on for decades start to break down. There is no human behavior to analyze, no device fingerprint you can trust, and increasingly no checkout happening inside your own environment, which means the merchant is left exposed to decisions made by systems they do not control.

The part that stands out to me is how quickly a new stack is forming in parallel. You have ACP shaping how agents interact with merchants, AP2 introducing enforceable mandates and audit trails, UCP standardizing discovery and distribution, while Visa and Mastercard are both trying to solve the identity layer, and x402 is quietly redefining settlement for machine-to-machine payments.

These are not competing solutions. They are different layers of the same system being rebuilt in real time.

The shift underneath all of this is from transaction validation to intent verification. If you cannot prove what the agent was actually authorized to do, you cannot defend the payment, and that becomes a direct revenue risk.

Most teams are still focused on enabling agentic checkout and improving conversion. I think that is the wrong layer to optimize.

The real battleground is trust, and more specifically, who controls the audit trail behind every autonomous transaction.