Patrick Collison predicts checkout pages will disappear for human shoppers, not just AI agents

The checkout page has been the terminal node of e-commerce for roughly three decades. Patrick Collison, co-founder of Stripe, says it will not stay that way. His prediction is brief and unqualified: “They will go away for human users too.”

That framing matters because the conventional expectation in commerce circles has been more modest. Most accounts of agentic commerce treat the disappearance of checkout pages as a phenomenon specific to AI agents buying on behalf of users. An agent that already holds authorization credentials, payment details, and purchase parameters has no reason to land on a form designed to collect those things. The page becomes friction without a function. Collison’s call goes further: the same logic applies to human shoppers.

The prediction has a specific shape that makes it checkable over time. Collison is not saying checkout becomes faster, or more embedded, or less visible. He is saying the page itself goes away. That is a harder claim than the agentic-commerce variant, because human consumers have three decades of muscle memory organized around the checkout ritual. Removing the page requires not only a technical substitution but a behavioral one.

They will go away for human users too. Patrick Collison

The practical stakes of the call being correct are substantial. Checkout pages are not merely a product category. They are the site of an enormous industry: conversion optimization, cart abandonment recovery, A/B testing on form design, payment flow sequencing. Merchants have organized significant engineering and marketing resources around the assumption that some version of a checkout page will always exist and will always need to be improved. If Collison is right, that assumption fails, and the work organized around it will need to be redirected.

Worth noting is that Collison is not a disinterested observer predicting someone else’s product category. Stripe builds and maintains the payments infrastructure that would have to support any world in which checkout pages do not exist for human users. He is describing a destination that his own company’s rails would have to reach. That proximity to the infrastructure gives the prediction a different character than an analyst forecast.

What the prediction does not supply is a timeline. Collison does not specify when human-facing checkout pages disappear, whether the transition happens across all retail categories simultaneously, or which segments lead. Those are the variables that determine whether the prediction lands as a five-year disruption or a decade-long evolution. Without that granularity, the call is directional rather than precisely dated, which means it is harder to falsify at any specific point along the way.

That ambiguity does not reduce the prediction’s significance. The question it raises is not whether checkout pages will eventually give way to something more embedded and less visible. The evidence from agentic commerce alone suggests some version of that shift is already underway. The question Collison’s call poses is whether the human shopping experience converges on the same endpoint as the agent experience, or whether the two diverge permanently because consumers prefer the ritual of review and confirmation. If he is right, a large amount of the optimization work currently applied to checkout will eventually be redirected toward whatever replaces it. If he is wrong, checkout pages persist as a human interface even after they vanish from the agent layer. Either outcome reshapes how merchants and payment infrastructure providers allocate resources. The prediction, for now, is a specific and traceable bet from someone positioned to see the infrastructure side of that question more clearly than most.