AI X-Ray Experience

Your store has two realities

Humans see the storefront. AI sees the signals behind it. Most stores were never built for that second reader. Scroll down. Watch the X-ray run on a real product page.

The AI X-Ray

What AI actually sees when it looks at your store

This is a real product page, X-rayed. Scroll it into view and watch the scan run. Then click any trace below to inspect exactly what AI extracted, and what it could not.

X-RAY ACTIVATED
lumine-skincare.com/products/recovery-serum
LUMINE SKINCARE
Recovery Serum
$48
01
★★★★★(128)
02
Bestseller
03
01 PRICEOffer.price found in schemaREADABLE
Source<span class="price">$48</span> + schema
Extractability100%
ActionabilityHigh: AI can quote the exact price
Recommendation impact+12 pts confidence
02 REVIEWSStar icons, no aggregateRating schemaUNCERTAIN
Source5 star icons + "(128)" as plain text
Extractability40%
ActionabilityLow: AI must guess a rating exists
Recommendation impact−6 pts confidence
03 BADGE"Bestseller" baked into hero imageINVISIBLE
SourcePixels only: no alt text, no schema
Extractability0%
ActionabilityNone: does not exist to AI
Recommendation impact−9 pts confidence
PRODUCT
├─ name: "Recovery Serum"
├─ price: $48.00 (readable)
├─ availability: unknown
├─ rating: unconfirmed
└─ badge: invisible
recommendation_output: null
AI Confidence
0%
Why the gap exists

The store did not change. The reader did.

Across the stores Atom Foundry has scanned, the pattern repeats: a store that looks complete to a human eye is frequently missing half its signals to an AI reader. Nothing on the page is broken. It was simply never built for this second reader.
What a human sees
What AI actually extracts
Sees a clear product photo
Alt text is empty or generic: the image itself is invisible
Reads the "Bestseller" badge on the image
Text baked into pixels cannot be extracted at all
Sees five gold stars and a review count
No aggregateRating schema: AI cannot confirm a real rating exists
Sees the price rendered clearly on the page
Explicit in the DOM: trustworthy to AI only if it also exists in Offer.price schema
Infers "in stock" from an active Add to Cart button
No explicit availability value: AI cannot confirm stock status
Recognizes the brand from a logo image
No Organization or brand schema: the logo means nothing to AI
Reads the full return policy on a linked page
Hidden behind a click or JS modal: AI does not follow links to verify
Ask AI

Ask the same question a real buyer would

Click run. This is the same evaluation a shopping AI performs in real time, on the same product the X-ray just scanned.

Buyer prompt
"What's a good recovery serum for sensitive skin under $60?"
Intent understood
recovery serum sensitive skin under $60
Confidence per signal
Price clarity
91
Product understanding
44
Trust signal strength
29
Recommendation readiness
35
AI confidence
AI response
  • Price fits the budget: confirmed.
  • Fit for "sensitive skin" cannot be verified: no ingredient or claim data.
  • Trust signals (rating, brand) are not machine-readable.
Verdict: Not confident enough to recommend over alternatives it can verify.
Make AI see it

Check a box. Watch the score move.

These are the exact fixes the X-ray above exposed. Each one closes a gap between what a human sees and what AI can extract.

31
AI Commerce Score
AI CANNOT RECOMMEND
recommendation_output: null
Before / after

Same store. Same photos. Different reader.

Before
What a human sees
Recovery Serum · ★★★★★ · $48 · In stock · Bestseller badge. Looks complete.
What AI can confirm
Price
Availability
Rating
Brand
31 / 100
After
What a human sees
Recovery Serum · ★★★★★ · $48 · In stock · Bestseller badge. Identical page.
What AI can confirm
Price
Availability
Rating
Brand
78 / 100
Nothing about the storefront changed. Only what AI could read did.
Why this is different

Most GEO advice says optimize your content. We say something else.

Generic advice
"Optimize your content for AI."
Atom Foundry
"Watch what AI actually does with your store."

Advice is a guess. An X-ray is a measurement. Everything above ran on a real store, not a hypothetical one, because the only way to know what AI sees is to look.

Where stores land

The gap, scored

Every store the X-ray reveals lands somewhere on this scale. Most stores that look polished to a human still sit in the bottom two zones.

85 to 100
High Confidence
AI agents confidently read and trust this store.
70 to 84
Moderate Confidence
AI reads this with moderate confidence; stronger signals would raise it.
50 to 69
Low Rec. Confidence
AI sees the store but misses key signals.
0 to 49
AI Visibility Risk
Excluded from AI recommendations entirely.
FAQ

Human view vs AI view: common questions

Why does my store look fine to me but score low with AI?
Human eyes and AI readers extract information completely differently. You see a photo, a badge, five stars. AI needs that same information in structured, machine-readable form: schema markup, explicit values, crawlable text. A store can be visually polished and structurally invisible at the same time.
What is AI extractability?
AI extractability is how completely an AI system can pull verifiable facts out of a page: price, availability, rating, brand, policies. High extractability means AI can quote your store with confidence. Low extractability means AI has to guess, hedge, or skip the product entirely.
Do I need to redesign my store to fix this?
No. Almost none of this is visual. The fixes are structural: schema markup, server-rendered prices, real alt text, crawlable policy pages. A store can look identical to a human before and after and score completely differently to AI.
How is this different from generic "optimize for AI" content advice?
Most advice describes what to write. This measures what AI can actually see and extract from what already exists. It is the difference between guessing what might help and watching, signal by signal, what does.
How do I check my own store?
Run a free AI Commerce Score scan at atomfoundry.dev. It runs the same kind of extraction check shown above across your real product pages in about 30 seconds.
Humans will keep seeing the storefront.
The question is what AI sees.
Find out in 30 seconds.
Run My AI X-Ray Free
Free · No credit card · Results in 30 seconds