Stripe says the checkout page is dying. Our take: not yet.
Stripe president Will Gaybrick said on the a16z podcast in August that checkout pages “will just go away” once agents are doing the buying. The argument is that the purchase should complete straight from the product page: you see something, you say “buy it”, and the transaction happens machine to machine. No forms, no card entry, no checkout flow.
Stripe is backing it with products rather than talk. Instant Checkout inside ChatGPT, a protocol for agent purchasing built with OpenAI, agentic buying in Google’s Gemini.
So is the checkout page dying?
Our honest take, after ten years of building ecommerce: not yet. But the people building the replacement are not messing about.
Three reasons not to change anything this week
Consumers want to stay in the decision. Logica Research surveyed 3,000 online shoppers for Commerce and PayPal in March and April 2026 — a thousand each in the US, the UK and Australia. Roughly two-thirds were interested in trying agentic shopping. But the interest stops at letting the agent research and suggest; letting it complete the purchase without approval is a different proposition. That is the gap between skipping the address fields and letting a model decide what ends up in the basket.
The technology works; the behaviour has not caught up. Gaybrick himself points to Link and Shop Pay as the closest thing we have today. That is worth pausing on, because the same two products read both ways: he sees them as the precursor, we see them as evidence of how slowly this moves. They have been laying the groundwork for years and adoption has been steady, not explosive. Removing friction from checkout is one thing. Removing the human from checkout is another order of magnitude.
For complex purchases this is years away. Nobody is letting an agent buy something expensive without reviewing the order. This lands first where the stakes are low and the purchase repeats — replenishment, consumables, the things you already buy monthly without thinking.
Why it is worth caring anyway
Because “the behaviour has not caught up” holds right up until it does, and because the change will not show up as a moved checkout button. It will show up as one store appearing in answers where others do not.
That is the uncomfortable part. If an agent picks between three suppliers on the customer’s behalf, the choice happens before any checkout page is involved. Your store no longer competes on how the checkout feels. It competes on whether your data can be answered with.
And that part you can start on today, whatever you believe about the timeline. We wrote about what an agent actually asks for in an earlier post.
What the series covers
Four posts, none of which claims you should rebuild anything this autumn:
- This one. Why the checkout page is not dying yet, and why it matters regardless.
- Checkout does not disappear — it moves. Where the logic goes when there is no page.
- B2B already has agentic commerce. It is called punchout and EDI and it is twenty years old.
- What to actually do. A short list. Shorter than you expect.
The conclusion up front, so you can stop reading here if you like: the checkout page is far from dead, but the companies building its replacement really are not messing about.
Forty-five minutes with someone who knows the platform, not a sales deck. The migration assessment is free.