Prices agents cannot misread
Schema.org Product JSON-LD with price, currency, availability and variants, served from your own data rather than reverse-engineered from your markup.
For e-commerce and marketplaces
Stop shopping agents guessing your prices.Serve them structured data instead.
AI shopping agents are already fetching your catalogue on behalf of buyers. When all they receive is rendered HTML, they infer price, availability and variants — and get them wrong in front of someone about to purchase. ADN serves those agents Schema.org product data derived from the page you already publish, and prices the commercial scrapers instead of blocking them into disguise.
- JSON-LD
- Schema.org Product
- MCP structured tools
- Content Signals
Sound familiar?
Where it hurts today
- 01
Hallucinated prices and stock
An agent that scrapes a variant selector will guess. The buyer sees a price you do not offer, and the mistake lands on your brand rather than on the model.
- 02
Catalogue crawls hit the database
Product pages are dynamic, numerous and expensive. A sweep across the catalogue is the worst-shaped traffic your origin can receive.
- 03
Blocking the scraper blocks the buyer
A WAF rule that stops a price scraper also stops the agent with a human waiting behind it — and you never find out which sales you lost.
What you get
What ADN does about it
Structured tools for buying agents
MCP tools let an agent look up a product, check stock and compare variants directly, instead of driving your storefront like a browser.
Scrapers priced, not blocked
Commercial scrapers get a free allowance and then metering with machine-readable terms. Priced traffic stays visible; blocked traffic just comes back wearing a residential proxy.
Googlebot untouched
Search and social crawlers pass straight through by a hard safety rule. Nothing about your organic search or your link previews changes.
Two ways in
Whatever is in front of your origin
The pitch changes with your infrastructure, not the product. On Cloudflare we install alongside your zone plan; everywhere else we are the whole agent layer.
Path 1 — Cloudflare add-on
Already on Cloudflare Pro
Pro serves one Markdown format to every machine visitor. ADN adds structured JSON-LD pricing for shopping agents, metering for commercial scrapers, and keeps your WAF and caching exactly where they are.
Pro $25 + Growth $149
Path 2 — primary stack
On Shopify, Vercel or your own stack
ADN is the whole agent layer: classification, structured product data, metering and analytics, with no Cloudflare plan underneath it.
Growth — $149/mo
Where you land
Growth$149/mo
Hosted proxy or edge connector
3 domains, 750k requests, 5k renders, per-class policy engine, commercial metering.
What this is not
If your real problem is competitors repricing against you, understand what you are buying: we meter that traffic and attach terms to it, we do not try to stop it. Detecting and blocking determined adversaries is fraud work, and the honest answer there is Cloudflare Bot Management, not us.
Free efficiency scan
See what a shopping agent gets from a product page
Scan a product URL. The results show the JSON-LD types found, what an agent would receive, and how much of the page is markup rather than content.
Questions
The ones that come up first
- Does this change what Google sees?
- No. Legacy SEO and social crawlers are passed through untouched — never Markdown, never a rewritten response. It is the one rule in the routing matrix with no exceptions.
- Where does the product data come from?
- From the pages you already publish. The pipeline extracts existing JSON-LD, and where a page has none the content pipeline derives structured fields and marks what was inferred rather than asserted.
- Won't metering scrapers just make them evade?
- Some will. The bet — and it is stated as a bet — is that a scraper offered clean structured data at a workable price behaves better than one offered a 403, because evasion costs it money too. Every response carries machine-readable terms, so the choice is at least explicit.
See what a shopping agent gets from a product page
Scan a product URL. The results show the JSON-LD types found, what an agent would receive, and how much of the page is markup rather than content.
Not quite you?