Install a salesperson on every product page.
One script tag turns every product page into a knowledgeable salesperson — verified specs, smart negatives, living FAQ, AI advisor. Built from your verified data.
See it live on a sample productSchema.org markup and the LLM Data Layer ship today. The interactive 3-layer surface is rolling out next.
60–70% of visitors leave to research.
They open 15 tabs. They go to Reddit. They ask ChatGPT. By the time they come back — if they come back — they've forgotten which shop they were on.
That's not a UX problem. That's a content problem. Your product page can't answer their questions, so they ask elsewhere.
Hook → Details → Deep Dive.
Lighter than 78% of over-ear ANC headphones.
30-hour battery beats 81% of wireless ANC. Quiet, scannable, fast.
Full spec sheet · Smart Negatives · Living FAQ · Trust provenance.
Every spec carries a source list. Smart Negatives surface what your product isn't for. 50–100 FAQs auto-generated.
Ask anything. Answers come from your verified data only.
Physically incapable of inventing claims. If we don't have the source, the Advisor says so.
Five things no static product page can do.
Contextual specs
Not just weighs 250 g — “lighter than 78% of over-ear ANC headphones.” Percentile context turns every spec into a buying decision.
Smart Negatives
Tell the customer what your product isn't for. Counter-intuitive trust signal that reduces returns and reviews.
Living FAQ
50–100 questions per product, auto-generated from review themes and verified data. Marked up as Schema.org FAQPage for rich snippets.
AI Product Advisor
RAG-based deep dive. Reads your verified data only. Physically incapable of inventing answers — if it doesn't have the source, it says so.
Trust provenance
Every spec links to its source. “Confirmed by 4 independent sources.” Footnotes for e-commerce.
Drop one tag.
Plain HTML. Shadow DOM means zero theme conflicts. < 40 KB gzipped, async-loaded.
<script src="https://widget.central.to/v1.js"
data-shop="<your-shop-id>"
data-product="<product-sku>"
async></script>
Works on every platform.
- ✓ Shopware 6
- ✓ WooCommerce
- ✓ Shopify
- ✓ Magento 2 / Adobe Commerce
- ✓ Custom HTML
AI agents read this. Without you writing JSON-LD.
{
"@context": "https://schema.org",
"@type": "Product",
"name": "Sony WH-1000XM5",
"brand": { "@type": "Brand", "name": "Sony" },
"weight": { "@type": "QuantitativeValue", "value": 250, "unitCode": "GRM" },
"x-central:data_confidence": 0.94,
"x-central:sourceCount": 16,
"x-central:negativeNotes": [
"No IP rating — not for gym use",
"Flat-fold only, larger carrying case than XM4"
]
}
From 2–3% to 4–6%.
Industry average sits at 2–3%. Shops with rich, contextual product intelligence routinely report 4–6%. The widget is the difference between a customer who leaves to open 15 tabs and a customer who buys.
The catalog gets smarter while you sleep.
Every search query, every FAQ click, every layer expansion is a signal that flows back into enrichment. Unanswered questions become new fields. Click heatmaps re-rank FAQs. Smart Negatives are extracted from review themes — automatically.
Things people ask before they install.
Will it conflict with my theme?
No. Shadow DOM isolates the widget from your theme's CSS.
Does it slow down my page?
Under 40 KB gzipped, async-loaded. Layer 1 first paint under half a second on 4G. CLS = 0.
Can I customize the look?
Yes — colors, typography, layout density. Brand voice in the AI Advisor follows your project's voice config.
What about Schema.org markup I already have?
Central detects existing markup and either enriches it or defers, depending on your preference.
What if a question can't be answered from my data?
The AI Advisor says so. It refuses to invent an answer. That's the whole point.
Which platforms does it support today?
Shopware 6, WooCommerce, Shopify, Magento 2, custom HTML.