Trust & Content Integrity
Why this page exists
AI writing tools have a fabrication problem. They invent statistics, cite studies that do not exist, and attribute quotes to people who never said them.
We built specific machinery to prevent that. Here is exactly what it does, and — just as importantly — what it does not.
What the engine will not do
These are enforced in code, not merely requested in a prompt:
Never invent a statistic, study, quote, source, person, or credential. The writer is instructed that publishing without a claim is better than publishing with a fabricated source for it.
Never publish a page that fails the quality gate. Every page is checked before it can go live: answer-block length, heading structure, meta limits, word targets, and a scan for the phrasing patterns that mark machine-generated filler. Failures block publishing outright — they are not warnings.
Never publish two pages targeting the same query. Self-cannibalisation is blocked at the database level with a unique constraint, not just checked in code.
Never generate a programmatic page without real data behind it. Thin variant pages are doorway spam and they damage the whole domain.
Never mark up content that isn't visible on the page. Structured data describes what a reader can actually see.
What we do not yet verify
Being straight with you about the gap:
We do not currently fetch every cited URL to confirm it exists and says what the page claims. The writer is instructed not to fabricate and the gate flags a page with no citations at all, but there is no independent verification step yet. It is the highest-priority item on our roadmap.
Until it ships: check the sources on anything you publish. Especially statistics.
What stays your responsibility
You are the publisher. Your name is on the site.
- Review before publishing. Every page is editable, and drafts do not go live on their own.
- For medical, legal, financial, or safety topics, get a qualified human to review. Our engine will draft these; it should not publish them unattended.
- Disclose AI assistance where your jurisdiction or platform requires it.
On AI disclosure
We think you should be honest that content is AI-assisted and human-reviewed. Every site we generate includes an editorial policy page saying so.
That is not a legal requirement everywhere. We think it is the right default anyway.
Your data
We do not train models on your content. Your business description and topics are sent to Anthropic to generate your pages; Anthropic does not train on API inputs. Details in the privacy policy.
Reporting a problem
Found a fabricated fact or a bad citation in something we generated? Tell us — it helps us fix the engine, not just that page.