Shopify now lets browser AI agents complete checkout. Your store's data is about to be read by machines first

Shopify just made it easier for AI agents to buy from its merchants, while Amazon and others are blocking them.

What happened

TechCrunch reports that Shopify expanded WebMCP support to checkout, including Shop Pay. Browser based AI agents could already search a store and add to cart. Now three new tools, get_checkout, update_checkout and complete_checkout, let an agent inspect the checkout, change things like the address or delivery option, and place the order after the buyer authorizes it. Shopify says this works without screenshots or scraping, and it is rolling out to all eligible merchants.

Shopify's developer docs explain the setup. WebMCP is a proposed web standard where the storefront registers tools with the browser, and the agent calls them with structured inputs. Shopify provides these tools on every Liquid storefront with nothing to install, though agent support is currently limited to Chromium based browsers. Storefront tools include search_catalog, get_product, update_cart and search_shop_policies_and_faqs. Cart changes run through the same standard actions apps use, so an agent's changes behave like the shopper's own. Shopify also runs a hosted MCP server for agents that work server to server, and both paths sit on its Universal Commerce Protocol.

My take

For merchants the question is no longer whether agents will shop your store. It is whether they will read it correctly.

An agent answering "what is your return window" calls search_shop_policies_and_faqs. If your policy page is vague or out of date, the agent repeats that to the buyer. Same with variants and stock: bad product data now fails silently inside someone else's AI.

What I would check this week:

1. Policies and FAQs: clear, current, one answer per question.
2. Variant names and inventory: consistent enough that "the blue one in medium" resolves.
3. Order tagging: can you tell which orders came through an agent, and route them in your CRM?
4. Post purchase flows: confirmation and support messages still land when a human never saw the checkout page.

This is the kind of system I build for teams: clean product and policy data flowing into the store, and agent driven orders tracked and followed up in the CRM like any other. More at romielwillautomate.dev.