How we check what AI says about you
AI assistants and answer engines increasingly decide what a buyer hears about a business. When they get it wrong — or say nothing — the business has no way to see it or fix it. This is how Nymrel checks, and the open format that lets agents read the correct facts.
How Nymrel checks what AI assistants and answer engines say about a business against its verifiable facts, and the open, fail-closed Verified Claims Record format that lets agents read the correct facts directly.
What we check
We assemble a branded query set — the real questions a buyer or an autonomous agent would ask about the business — and run it across AI assistants and answer engines. Each answer is recorded verbatim, with its date, so the check is a dated snapshot, not an impression.
How we decide what's true
Every fact must trace to something published or a primary document. We fail closed: a claim is only marked verifiedwhen it carries a source, a method, and a check date. A claim we can't confirm is self-asserted or unverified — never asserted as true. We never write plausible-sounding filler, because AI models read it and repeat it as fact.
The Verified Claims Record
The output isn't a promise — it's a machine-readable, dated, source-backed record published at a stable path on the business's own domain, so agents and AI answer engines read the correct facts directly. The format is open (version 0.1.0) and every claim carries its status, source, method, and date.
{
"specVersion": "0.1.0",
"recordType": "VerifiedClaimsRecord",
"subject": { "name": "Nymrel", "url": "https://nymrel.com" },
"claims": [
{
"id": "clm-founder",
"statement": "Nymrel was founded by Jalen Johnson, Senior product engineer.",
"value": "Jalen Johnson — Senior product engineer",
"status": "self-asserted",
"verification": {
"method": "published-source",
"sources": [
{ "url": "https://jalenbuilds.com/resume", "checkedOn": "2026-07-16" }
]
}
}
],
"integrity": { "hashAlgorithm": "sha256", "signature": null }
}Full field reference and versioning are in the format specification. The record at https://nymrel.com/.well-known/verified-claims.json is Nymrel's own, generated live from its published facts.
How an agent uses it
Fetch and validate
Fetch the record and validate it against the published JSON Schema.
Check freshness
If the record is past its valid-until date, treat it as stale.
Verify integrity
Confirm the content hash matches the claims to detect accidental corruption. v0.1 is unsigned and does not authenticate the publisher.
Use only verified claims
Trust only claims marked verified with a source; surface the rest as unconfirmed.
Questions
How does Nymrel check what AI says about a business?
We build a branded query set — the questions a buyer or agent would actually ask about the business — and run it across AI assistants and answer engines, recording each answer verbatim. We then compare every answer against the business's verifiable, published facts and flag anything wrong, missing, outdated, or misattributed, with the correct fact.
How do you decide what is true?
Every fact must trace to something published or a primary document. We fail closed: a claim is only marked verified when it carries a source, a method, and a check date. A claim we cannot confirm is marked self-asserted or unverified — never asserted as true. We never invent plausible-sounding facts.
What is the Verified Claims Record?
A machine-readable, dated, source-backed file published at a stable path on the business's own domain, listing its verified facts so autonomous agents and AI answer engines can read the correct information instead of guessing. The format is open and documented; only claims backed by a source are marked verified.
Do you promise a business will rank or appear in AI answers?
No. We measure accuracy and machine-readability and give the business a source of truth agents can read. We never promise rankings, citations, traffic, or that any engine will repeat a specific message.