How We Built spruesentry.com for the AI Catalog Era

We built spruesentry.com to test AgentReady on a site we own. Disclosure: it is our Labs proving ground, not a customer store and not a Shopify merchant. Every number below comes from the public Labs feed or the site itself.

The AI catalog era is not a slogan. Shoppers ask assistants for products and answers. Assistants read structured catalogs, discovery files, and pages written for questions. We wanted a site we could change freely, measure in the product, and publish without anonymizing the failures.

What SprueSentry is (so the receipts make sense)

SprueSentry tracks Warhammer 40K and Age of Sigmar kit prices across eBay and Shopify listings. It is a price tracker and alert tool, not a marketplace. The public agents.md and llms.txt say that plainly so assistants do not invent a storefront that is not there.

The catalog on the site is large enough to hurt if the data is messy: the card sitemap alone lists on the order of a thousand URLs for box and kit pages, plus army hubs, paint guides, and a weekly preorders board. That density is the point. Thin marketing sites do not stress the same readiness failures a real catalog does.

What we changed for the AI catalog era

Before AgentReady, SprueSentry was already a Next.js site with SSR HTML, JSON-LD, a sitemap, and robots.txt. That was table stakes. The gap was the agent-facing layer: a clear agent guide, an llms.txt index, and pages that answer the exact questions collectors ask, not just product cards.

Here is what we shipped and measured, with dates from the Labs program (joined July 15, 2026):

1. Discovery files that tell the truth. Live agents.md and llms.txt describe what SprueSentry is, what it is not, and where to send people. Paste-and-drift still applies on Shopify; on our own stack we publish them as first-class routes and keep them aligned with the product.
2. Answer monitoring as the scoreboard. By July 18, 2026 the Labs scoreboard showed 47 answer runs (24 OpenAI, 23 Google) and 0 mentions. That is not a ranking claim. It is a dated presence count. Measured since then, not necessarily because of any single edit.
3. A recommended fix we actually published. AgentReady queued 14 recommended fixes. We published one: a comparison guide, "Which 40K boxes would you recommend?", live at /guides/which-40k-boxes. The product verified that artifact present on July 17, 2026 after an earlier same-day verify failed. One verified fix out of one published. The rest of the queue is still open on purpose; Labs is not a vanity ship-everything program.

If you want the living changelog rather than this snapshot, it is public at /labs/spruesentry. Numbers on that page are derived from the app. This article is a dated cut of the same record.

What AgentReady changed in practice

The useful change was not "AI magic." It was a loop we could not fake on a customer store without permission:

• Confirm what the site is (profile and offerings), including rejecting junk crumbs that look like products but are not.
• Recommend fixes grounded in that profile (comparison content, structured clarity, monitor questions).
• Apply on a site we own, verify by recrawl, then re-run the monitor.

For Shopify merchants the same loop maps onto the Global Catalog: eligibility, absorption, category searches, then dated re-checks. SprueSentry just happens to sit outside Shopify, which is why this case study never pretends we "won the Catalog buy box." We did not. We ran the readiness and measurement loop on a catalog-shaped site and published the receipts.

How to steal the useful parts for a Shopify store

You do not need to build a Labs site. You need the same discipline:

1. Write down what an assistant should believe about your store (policies, category language, what you sell and do not sell).
2. Put that into the surfaces assistants already read: product data that meets Shopify Catalog eligibility, plus agents.md / llms.txt if you maintain them.
3. Measure with fixed questions and dates. Presence first. Ranking talk later, or never.

If a product is missing from ChatGPT shopping answers today, start with whether Agentic Storefronts are on (they have been on by default for eligible stores since roughly March 2026) and whether organic ChatGPT product surfacing is actually reading Shopify Catalog. It is. Then run the absorption check in did the Global Catalog absorb your products.

Where this leaves us

SprueSentry is ten days into Labs as of late July 2026, with one verified content fix and zero answer mentions so far. That is an honest early read, not a before-and-after victory lap. The companion post covers the products and offerings that still do not show, the requirement that failed, and the fix we applied: why some SprueSentry products don't show.

If you are on Shopify and want the merchant path, check your store free with the Catalog Readiness checker, then read the /shopify lander for how AgentReady verifies the six catalog requirements and keeps the structured layer in sync.