How it works
From scan to measured, in one afternoon.
SchemaX adds one layer to your existing site. No rebuild, no SDK, no migration. You set the rules, SchemaX does the rest, and shows you every exception. Here is the whole flow, end to end.
step 01
Scan your site, in seconds.
Paste a URL. SchemaX crawls the way an agent would: pages, products, the schema already there, contact details, and the gaps between them. You get an honest report of what an agent can read today.
- Crawl budget is small, polite, and identifies itself
- Read only. No JS executed against your site
- Result is yours. Not stored, not shared
scanning https://acme.com crawling pages ▓▓▓▓▓▓▓▓▓▓ 142/142 parsing semantics ▓▓▓▓▓▓▓▓▓▓ done grounding entities ▓▓▓▓▓▓▓▓▓▓ done found 28 entities 19 grounded 12 gaps → ready to review at app.schemax.io/r/x9k2
step 02
Set the policy.
SchemaX grounds every fact against your live pages, with the source paragraph alongside. You set the policy once: grounded facts publish automatically, and anything uncertain or sensitive is held for you in Claim Review. You spend attention only on the exceptions.
- Every value carries the spot on the page it came from
- Grounded facts auto-publish; uncertain ones wait for you
- Nothing is invented. No borrowed prices, no guessed facts
{
"@type": "Product",
"name": "Wireless Lamp",
✓ "price": "49.00",
✓ "availability": "InStock",
⏸ "shippingDetails": { … }
}
// ✓ auto-published: grounded on the page
// ⏸ held for you: source unclearstep 03
Publish within your rules.
Add one script tag. Your grounded facts ship as JSON-LD in your pages, and your agent-commerce profile is served at /.well-known/ucp. Every agent-facing surface stays off until you turn it on, and the manifest declares only what you have grounded. No payment handlers, no checkout, by construction.
- Works with any stack: Next, Astro, WordPress, static
- Serves /.well-known/ucp from your grounded identity
- Surfaces are owner-activated, off by default
{
"ucp.version": "2026-04-08",
"capabilities": {
"ai.schemax.schema.manifest": { … }
},
"_schemax_identity": {
"name": "Acme Studio",
"url": "https://acme.com"
}
}
// no payment handlers · no checkoutstep 04
Answer agents from verified facts.
Turn on your agent app and any agent can connect to your site's hosted MCP server. It answers about your business from your grounded pages only, and refuses anything that is not there. Visitors can send you an inquiry without leaving the chat. It can never transact or change your site.
- Read-only plus inquiry capture, no agent-writable routes
- Off until you turn it on; you approve the widget version
- Off-graph questions get exactly: Not stated on the site.
> about
{
"name": "Acme Studio",
"url": "https://acme.com",
"source": "site content, verified 2026-06-12"
}
> ask "do you ship to Canada?"
Not stated on the site.step 05
Re-verified daily, measured continuously.
SchemaX re-checks every published layer against your live page daily: presence, version, canonical alignment, and grounding. If anything drifts, you find out before agents do. And it measures what legibility earns: AI-referred visits by surface, agent reads, and the inquiries that follow.
- Daily drift re-verification, flagged before an agent sees it
- Drift alerts to Slack, email, or webhook
- AI-referred visits, agent reads, inquiries on your dashboard
AI-referred visits ▓▓▓▓▓▓▓▓░░ trending up agent reads ▓▓▓▓▓▓░░░░ ChatGPT · Claude · Perplexity inquiries captured ▓▓▓░░░░░░░ in your lead inbox
Open the door for the agents.
Set once, approved by you. Every agent on every platform reads the same verified facts: in your pages, in your /.well-known/ucp manifest, and through your site's agent app.