On September 28, 2026, TechCrunch reported that Shopify has extended support for WebMCP, a protocol that lets AI agents working in the browser complete purchases on merchants' stores. Until now, those agents could search for products and fill a cart. Now they can also read, edit and submit the checkout form, Shop Pay payments included, but only after the buyer gives permission. If you run a Shopify store, this concerns you.
What we know, and what we don't
We know the new tools are three: get_checkout to read the checkout, update_checkout to change address, delivery and other details, complete_checkout to submit the order. Shopify's technical documentation lists a fourth one too, navigate_to_storefront, which sends the buyer back to the store's storefront. According to TechCrunch and The Paypers, the rollout covers all eligible merchants.
The documentation also sets a limit: before completing the order, the agent must show the buyer the order and the total and get permission. The "ready_for_complete" status does not count as authorization. The agent also cannot enter new card details: only already-saved Shop Pay cards and billing addresses are supported.
We don't know which checkouts are eligible, because the documentation mentions excluded checkouts without listing them. We don't know whether a merchant can turn the feature off, or how many orders will go through it: none of the sources we read gives adoption numbers. TechCrunch adds that Amazon and Adidas are blocking AI agents that try to buy, so opinions in the industry are split.
What it means for a small store
For buyers, a browser agent is an assistant that fills in the form for them: you say what you want, it goes all the way to payment and asks you to confirm. For you, the checkout may also be read by an assistant, which looks at delivery options and totals the way a person would, only faster and without distractions. This is our own reading, Shopify does not put it in these terms. It follows that shipping costs and conditions revealed only at the last step become a problem for everyone, person or assistant. If you run a small store you have an advantage: you can fix these things in an afternoon, with no meetings and without waiting for anyone.
What to do
- Place a test order as a customer. Look at the checkout the way someone who doesn't know the store would: costs, delivery times, required fields, discounts.
- Make costs and timing clear before the last step. What a customer has to hunt for across three different pages, an assistant may fail to find.
- Reread your returns and terms. They should say simple things, consistent with what you see in the checkout.
- Watch the orders of the coming days more closely. Addresses, totals and discounts: if something looks odd, note it down.
- Read Shopify's documentation on WebMCP checkout, or have whoever looks after your store read it, before making decisions.
- If your checkout is customized, ask whoever maintains it whether the news affects it. The sources we read have no answer.
What to avoid
- Changing the checkout in a rush. Without adoption numbers, waiting and watching costs little.
- Buying promises about "AI agent optimization". No source talks about effects on sales, so whoever guarantees them is guessing.
- Leaving inconsistent data. Different prices between the product page and the checkout confuse customers and would confuse an assistant too.
A hypothetical example
Hypothetical example: an online shop selling handmade ceramics on Shopify. A customer asks her browser assistant to buy the light blue mug and have it delivered to her mother's home by Friday. The agent picks the mug, changes the delivery address, selects shipping and, before submitting, shows the order and total and waits for the go-ahead. The potter notices that delivery times are written only on a separate page, so nobody, human or machine, can tell from the checkout whether the mug arrives by Friday. She moves the information where it's needed, and the problem disappears for everyone.
The bottom line
Shopify's checkout now also admits visitors that are AI assistants, with the buyer's permission as the brake. For a small store the useful move is the same as ever: a clear checkout, with readable costs and timing. If you have a Shopify store and want to know what to check first, or an order from an AI assistant has already reached you, write to us at info@sarabi.cloud and tell us your case.
Sources: TechCrunch, The Paypers and Shopify, Checkout WebMCP documentation.
Have a similar case or a doubt? Write to us at info@sarabi.cloud.