← Back to blog

Shopify Now Lets Stores Approve What AI Shopping Agents Say About Them

⭐ Featured

When an AI agent shops for someone, it does not visit ten stores and read their return policies. It reads a feed. Shopify's Fall '26 Edition, published on October 28, makes that feed something a merchant can see and edit.

The release is long. Three of them change how a store gets picked when the buyer is an agent: a Knowledge Base for agents, a Catalog eligibility check, and Shop.md, a skill that turns a general assistant into a personal shopper.

A page of facts that only agents read

The new Knowledge Base sits in the admin under the agentic sales channel. It holds the facts an agent gets about a store: return windows, shipping destinations, gift services. Shopify drafts it from the storefront and suggests updates when the site changes.

The important line in Shopify's announcement is the control. Nothing Shopify drafts reaches an agent without the merchant's sign-off.

That is new. Until now, an agent learned a store's return policy from whatever it found: a footer link, an old FAQ, a review that complained about a refund. Now a store has one approved answer, and Shopify hands it to the agent directly. A store that never reviews the page leaves the agent with nothing approved to say.

Eligibility decides whether an agent can see you at all

Shopify says ChatGPT, Copilot, Meta, Instinct and other agents get product information from Shopify Catalog. The new admin view tells a merchant whether the store is eligible for that catalog, and links to fixes when it is not.

Read that the other way round. A store that fails the check is not ranked low by these agents. It is absent. There is no answer to optimize when the agent's source list does not contain you.

Shopify also says products with Shop Promise badges, its delivery and returns guarantee, can rank higher in the catalog data that shopping agents use. Eligibility for the badge has grown twentyfold. So there is a second layer: once you are in the feed, verifiable delivery promises move you up it.

Any agent can now shop the whole catalog

Shop.md is a skill file. A buyer points ChatGPT, Claude or another agent at it, and the agent can search more than a billion products, recommend items from preferences and order history, check out with Shop Pay and track the order.

Checkout itself runs inside three personal agents first: Muse from Meta, Claude Duet and Instinct. Shopify says the buyer stays in control and the card is never shared with the agent. More than 400 million people already use Shop Pay, so the payment step is not where these purchases will stall.

On the merchant side, connectors now link a store to nine agents, including ChatGPT, Claude, Microsoft Copilot, Google Gemini, Grok and Perplexity. That half is about running the store from a chat window, not about being found.

Discovery is the part that matters

Vanessa Lee, Shopify's VP of Product, put the priority plainly in coverage of the launch: discovery is "the bulk of what people want" when they shop through an agent.

That matches what we see in AI answers outside retail. The buying step is easy to build. The hard part is the shortlist: which sellers the agent considers, what it believes about each, and which one it puts first. Shopify has now turned each of those into something a merchant can check. Are you in the catalog? Are your facts approved? Do your guarantees carry a badge the ranking reads?

Two weeks ago we wrote about Amazon blocking outside shopping agents while Shopify opened its catalog to them. This release goes one step further. It gives sellers the tools to manage what agents see, not only whether they see it.

What a B2B company should take from a retail release

Most B2B companies will never have a Knowledge Base page inside an agent's admin. ChatGPT and Claude do not ask a software vendor to approve the facts they repeat. They build them from whatever the open web says: your pricing page, a two-year-old comparison post, a Reddit thread, a directory listing with your old plan names.

So the same three questions apply, without the convenient admin screen. Is your company in the sources the engines read when a buyer asks for a vendor? Are the facts there current and consistent, or is the agent quoting the stale version? Is there anything third-party and checkable, the B2B equivalent of a delivery badge, that makes an engine more confident naming you?

Shopify merchants now get a checklist for all three. Everyone else has to build their own.

The bigger picture

The question behind every AI shopping feature is the same: when a machine makes the shortlist, what does it believe about you? Shopify has answered it for its stores with a page the merchant signs. For a B2B startup, the answer is spread across the pages the engines can read and the questions buyers actually ask. You have to measure it before you can fix it.

ClawWorld runs AI visibility work for B2B startups. Start with a free baseline report: 10 buyer questions, who ChatGPT and Claude name instead of you, and the pages they cite.

Get your free baseline report →