Is my store visible to AI shopping agents?

Your store is visible to an AI shopping agent when it is permitted to fetch your pages, able to enumerate your catalogue, and able to read your product data once it arrives. All three have to hold. Any one of them failing makes you absent rather than merely worse placed.

The reason this catches people is that none of the three is what makes you visible in search, so search performance is not evidence either way.

Permission: are you letting it in

The first question is whether an agent is allowed to fetch your pages at all. Two things decide this, and they are usually owned by different people.

Your robots.txt is the one you control directly. The common failure is a rule written for a different purpose: a disallow aimed at training crawlers that also removes assistant traffic, because the tokens were grouped together in whatever guide was being followed at the time. Search, training and agent traffic are three separate decisions with three separate consequences.

Your CDN is the one that changes without you. Cloudflare wrote on 1 July 2026 that from 15 September 2026, for new domains onboarding to Cloudflare, bots in its Training and Agent categories will be blocked by default on pages that display ads, that the Search category stays allowed, and that customers can opt out in their Security settings before that date. Most storefronts do not display ads, so most are not directly affected. The point is that these categories exist, somebody is going to set them, and it should be you rather than a default.

Enumeration: can it find everything you sell

Being fetchable is not the same as being findable. An agent has to get from your domain to your individual products, and it will not crawl your store the way a search engine does over weeks.

• A sitemap that lists product URLs, not only collections. See how to structure collections for agent discovery.
• A product feed, so your catalogue can be read in one request rather than reconstructed page by page. See how to publish a product feed for AI agents.
• Category structure that means something, so a request for a kind of product resolves to a set of yours rather than to your homepage.

Being expensive to enumerate is a soft form of invisibility. You are technically present and practically skipped.

Legibility: can it use what it finds

The third condition is that the data survives the trip. A product page an agent can fetch, listed in a sitemap it can read, is still useless if the price is rendered by script and the returns policy is an image.

This is where product structured data does the work: price, availability and an identifier such as a GTIN, published as fields rather than inferred from layout. Signs your store is invisible to AI agents lists the symptoms you can spot without tooling.

Visibility is per assistant, until you decide the category

There is no single switch. Assistants identify themselves with different user agents and discover products in different ways, so a rule that names one token by hand permits exactly that one. Every new assistant then arrives blocked by default, and you find out slowly or not at all.

Deciding at the category level, rather than maintaining a list of names, is what makes this stable. Blocking is a legitimate decision to take, as long as it is taken deliberately rather than inherited from a rule written for training crawlers. Robots blocks agent covers what the rule actually does.