AgentReadyAI visibility appCaffeine & CommerceShopify agency
Caffeine and Commerce

Proof of practice

Last updated July 6, 2026

We practice what we check

It would be easy to publish a wall of screenshots claiming an assistant loves us. We don't, because we can't prove most of it, and neither can anyone else. So this page does the one thing that is real and verifiable: it runs our own sites through our own AI-readiness engine and shows the results, gaps and all, with a button to re-run the scan live right now.

No invented citations. No fabricated ranking lift. No borrowed vendor percentages dressed up as ours. Where we have no evidence yet, this page says so plainly. What we can stand behind is that our data is clean and machine-readable, and you can check that yourself in seconds.

Our own sites, scored by our own engine

These are real results from the same code behind the free checker, captured on July 6, 2026. Not idealized: where a check warns or fails on our own site, it shows. Hit “Re-scan now” on any card to run it live and confirm the numbers for yourself.

0/ 100 · B

caffeinecommerce.com

Magento
17passing5warnings2failing
Brand identity
100
Products & catalog
14
AI discovery files
92
Answer-engine basics
100
Trust & policies
67

Two honest notes on our own result. Our platform sniffer guessed “Magento” from a string on the page. We run on Next.js, so that heuristic is imperfect, and we keep it visible rather than special-case our own domain. And our Products score is low on purpose: we are an agency site with no storefront, so the Product and Offer checks have nothing to pass. The score reflects the site we actually are.

Show all 24 checks
  • Pass: Brand identity (Organization / WebSite)Caffeine and Commerce
  • Pass: Logo in structured datamark-512.png
  • Pass: Linked social profiles (sameAs)
  • Pass: Sitelinks search (SearchAction)
  • Fail: Product structured data
  • Fail: Price & availability (Offer)
  • Warn: Ratings & reviews (AggregateRating)
  • Warn: Breadcrumbs (BreadcrumbList)
  • Pass: AI search crawlers allowedall allowed
  • Pass: All crawlers reachable (robots.txt)
  • Pass: llms.txt index56KB
  • Pass: agents.md guide3KB
  • Pass: XML sitemap
  • Pass: Sitemap freshness (lastmod)
  • Pass: Served over HTTPShttps
  • Warn: Freshness signal (dateModified)
  • Pass: Page title
  • Pass: Meta description223 chars
  • Pass: Canonical URL
  • Pass: Open Graph (image + title)
  • Pass: Language declared
  • Pass: Return policy signal
  • Warn: Shipping details
  • Warn: FAQ structured data
Runs the live public checker on caffeinecommerce.com. Baseline: 75.
0/ 100 · B

app.caffeinecommerce.com

Shopify
17passing6warnings1failing
Brand identity
90
Products & catalog
43
AI discovery files
88
Answer-engine basics
100
Trust & policies
83

The AgentReady app's own marketing surface. It publishes FAQ schema and an Offer (our pricing) that the scan reads, so Trust and Products score higher here than on the agency site. The one hard fail is Product schema: the app markets software, not physical products, so there is no product feed to carry it.

Show all 24 checks
  • Pass: Brand identity (Organization / WebSite)Caffeine and Commerce
  • Pass: Logo in structured datamark-512.png
  • Pass: Linked social profiles (sameAs)
  • Warn: Sitelinks search (SearchAction)
  • Fail: Product structured data
  • Pass: Price & availability (Offer)USD 0.00
  • Warn: Ratings & reviews (AggregateRating)
  • Warn: Breadcrumbs (BreadcrumbList)
  • Pass: AI search crawlers allowedall allowed
  • Pass: All crawlers reachable (robots.txt)
  • Pass: llms.txt index7KB
  • Warn: agents.md guide
  • Pass: XML sitemap
  • Pass: Sitemap freshness (lastmod)
  • Pass: Served over HTTPShttps
  • Warn: Freshness signal (dateModified)
  • Pass: Page title
  • Pass: Meta description293 chars
  • Pass: Canonical URL
  • Pass: Open Graph (image + title)
  • Pass: Language declared
  • Pass: Return policy signal
  • Warn: Shipping details
  • Pass: FAQ structured data
Runs the live public checker on app.caffeinecommerce.com. Baseline: 80.

docs.caffeinecommerce.com

No result

The scan couldn't resolve this host on the capture date, so there is no result to show. We list it here rather than quietly drop it: an honest empty is better than a borrowed number. When the docs subdomain is live, its real scan appears here.

2 of 3 properties returned a live result on the capture date. Each card links its score to the exact checks behind it. Want the weights? Read the methodology.

Reproducible evidence

Assistant citations, logged as they happen

When an AI assistant genuinely cites one of our pages, we record it here with the date and the exact prompt, so it can be reproduced. Until then, this space stays honestly empty.

No assistant citations logged yet

This is deliberately empty. We log real, dated citations here as they happen, each with the verbatim prompt that produced it, so anyone can reproduce it. We will not fill this space with invented screenshots or borrowed numbers. When an assistant genuinely cites one of our pages, the record shows up here with its date and prompt.

Being cited by an assistant is not something a store can guarantee, and no file makes it happen on command. What we can do, and show below, is keep our own data clean and machine-readable, then log any citations honestly when they arrive.

What this actually buys you

Clean data is hygiene, not a magic ranking lever

We are careful about what structured data, llms.txt, and agents.md do and don't do. Here is the honest version.

What it does

  • Lets an agent read who you are, what you sell, and at what price.
  • Makes your pages parseable instead of a guess from raw HTML.
  • Future-proofs you as agent conventions (agents.md, UCP) standardize.

What it doesn't do

  • It does not force any assistant to cite you. Nobody can promise that.
  • An llms.txt file has no proven ranking or citation lift on its own.
  • It is not a substitute for actually being allowed in robots.txt.

External industry findings (cited by direction, not as our results)

  • Independent studies have found no measurable ranking or citation lift from publishing an llms.txt file. We treat it as machine-readable hygiene and future-proofing, not a ranking lever. (Reported industry testing of llms.txt, 2024–2025)
  • Complete product identifiers and structured data are reported to improve how often products are matched and surfaced. We cite the direction only; we do not claim a specific percentage as our own result. (General industry reporting on product structured data)

Changelog

Every change to this page's evidence, dated

This page is a living record. When we re-capture a scan or log a real citation, it goes here with its date.

  • Published the self-proof page: real scans of caffeinecommerce.com and app.caffeinecommerce.com from our own engine, plus the empty-by-design assistant-citations log.

Now do it to your own store

The same free checker, the same published method. See where you stand in a few seconds, then watch the number climb as you fix things.