Claim registry
Reading a website is not proof the website is honest. Anything we derive from your own marketing is born unproven, and has to earn its way up.
Three states, no shortcuts
A claim starts unproven. With evidence attached it becomes asserted. With a source that stands up it becomes verified. A claim cannot skip from unproven to verified, and cannot reach verified without a source. These are guarded transitions, not a status field someone can set.
Derived claims start at the bottom
When the engine reads your site and derives that you do something, that derivation is evidence about your marketing copy, not evidence about your business. It enters the registry as unproven and stays there until a human attaches a source.
Why derived claims cannot be trusted by default
Reading a sentence on a website tells you the website says it. It does not tell you the sentence is true. So a claim extracted from a client site is born unproven no matter how confidently the page stated it, and it stays there until someone attaches something that stands up on its own.
Only two states can ship
Asserted and verified. Unproven claims are blocked at the gate, which means the worst case is content that does not publish, rather than a claim you cannot defend.
Questions
Who decides what counts as a source?
You do. We enforce that one exists and record what it was.
More in Never ship a bad claim
Validation gate
Seven validators run before anything ships and six of them block: unproven claims, unsourced numbers, fabricated quotes, missing AI disclosure and more.
Read moreGap reporting
When a check cannot answer something, it records a gap with the reason and what would fill it, rather than reporting a pass it did not earn.
Read more
Run it against your own site.
No signup, no card. It reports what it could not check as well as what it found.