Publish a catalog AI assistants can read, and track whether they mention you. This is the AEO half of the app.
On this page
Answer engine optimisation is search optimisation aimed at an assistant rather than a results page. The app’s framing: “Shoppers now ask ChatGPT and Claude what to buy as often as they search Google.”
The difference in practice is who is reading. A search engine ranks pages; an assistant needs facts it can state confidently and attribute. That means it wants structured, unambiguous, machine-readable information — which is why structured data, GTINs, product types, brands, descriptions and FAQs all count toward AI readiness as well as SEO.
The dashboard checklist carries AI catalog is published — “llms.txt and agents.md tell AI assistants what you sell” — with a Publish button.
It is one of the three setup items, alongside the app embed and structured data, and the one most stores have never heard of.
| File | What it is for |
|---|---|
| llms.txt | A short, structured summary of your store for AI assistants — who you are, what you sell, where the important pages are. Your store summary is read from here. |
| llms-full.txt | The longer version, with more of the catalogue. |
| agents.md | Guidance aimed at AI agents browsing or acting on your store. |
These are emerging conventions rather than settled standards, and not every assistant reads them today. The cost of publishing them is a single button; the cost of not having them when an assistant does look is a store it cannot describe.
Written during setup, and the thing assistants quote most directly: “AI assistants read this summary in your llms.txt file, and we use it to suggest the shopper questions to track. Keep it factual.”
| Field | Guidance |
|---|---|
| What do you sell, in a few words? | A plain category phrase. This is how an assistant will classify you. |
| Store summary | “Two or three sentences: what you sell, who it is for, what makes you different.” |
Keep it factual is the operative instruction, and it is not stylistic advice. An assistant repeating your marketing superlatives back to a shopper reads as either noise or a claim you are making — and a claim an assistant attributes to you is one you own.
What belongs: the category, the audience, specifics that genuinely differentiate (sizes carried, fitting service, certifications, where you ship). What does not: “best”, “leading”, “world-class”, and anything you could not evidence.
Draft summary with AI writes a first version from your catalogue. Edit it before saving.
Scored separately from SEO, and checking different things: “GTINs, brand, product type, FAQs, reviews, and whether your catalog is published for AI.”
| Signal | Why an assistant wants it |
|---|---|
| GTINs | Identity. It is how an assistant knows your product is the same one it has seen elsewhere. Filled in on the variant barcode field in Shopify. |
| Brand | Who makes it, as distinct from who sells it. |
| Product type | Classification. An assistant answering “walking shoes for wide feet” needs to know this is a walking shoe. |
| FAQs | Already in question-and-answer shape — the easiest thing in your store for an assistant to quote. Generate them from Meta tags. |
| Reviews | Third-party corroboration, read from your reviews app. |
| Published catalog | llms.txt and agents.md. |
Three of the six — GTINs, brand and product type — are Shopify data fields, not app settings. The app can tell you they are missing; filling them in is catalogue work.
“Add the questions your customers ask and we will check the answers weekly.” The app then watches whether assistants mention your store in response.
Your plan sets how many you can track — 3 on Starter, 15 on Mid-market, 50 on Enterprise, all checked weekly.
Choosing them well matters more than the number:
Alerts will tell you about “a big drop in AI assistant mentions”, so you do not have to check manually.
No app can make ChatGPT or Claude recommend your store. Assistants decide what to say from their training, their search results and their own judgement, and none of that is purchasable. What this app does is make sure that when an assistant does look at your store, it finds facts it can state rather than a page it cannot parse.
Treat AI readiness the way you would treat crawlability: necessary groundwork, not a growth channel on its own. The visibility questions tell you whether it is working, which is more than most stores can currently say.
Still stuck? Email [email protected] — we reply within 24 hours, every day of the year.
NIVI Solutions builds and supports the app. Contact our team through the app or our contact page for help.
Yes. We can configure the app, import your data and connect it to your ERP as part of a Shopify project.
Yes. NIVI Solutions is a Shopify Partner with 15+ years of B2B commerce experience.