How we make our content

Google’s guidance on AI-generated content recommends telling readers how content was created. We agree, so here it is in full.

The short version

Our articles are drafted by AI and published automatically. They are grounded in sources we cite, and they pass a set of automated checks before going live. No human reads every article before publication.

We would rather tell you that directly than let you discover it.

What actually happens

1. Topic selection. We track what people search for around sleep. A topic is only written if there is real demand for it and the existing results don’t already answer it well.

2. Research. The drafting model performs live web searches and pulls from primary sources — NIH, NHLBI, CDC, the American Academy of Sleep Medicine, and published sleep research.

3. Drafting. A large language model writes the article against a structured brief.

4. Fact-checking pass. A second model pass re-reads the draft against the sources it cited and flags claims that aren’t supported by them.

5. Automated gates. Before anything publishes, it must pass checks for: sufficient depth, no duplicate or near-duplicate content, no health claim without a citation, no absolute claims (“cures”, “guaranteed”, “100% safe”), and no page left unlinked from the rest of the site. A failed check blocks publication. There is no override.

6. Publication. The article goes live with its sources listed and the date it was last checked.

What the AI does not do

• It does not diagnose anything, and it is never asked to.
• It does not decide what is true. Every health claim has to trace to a source we link.
• It does not invent authors. Articles are attributed to the editorial team, because that is who is accountable for them.

Where this approach is weak

Being straight about the limitations, because they’re real:

• Automated fact-checking is not the same as expert review. It catches unsupported claims; it does not catch a source that is outdated, contested, or being read too generously.
• We have no clinician on staff. For health content, that is a genuine limitation, not a technicality. Where a topic needs clinical judgement, the honest answer is to see a clinician, and we try to say so rather than fill the gap with confident prose.
• Errors will happen. When you find one, tell us and we will fix it and note the correction.

Why publish this at all

Because a reader deciding whether to trust a health page deserves to know how it was made. If that costs us some trust, we would rather lose it honestly than keep it by omission.