How we use AI
What AI does here
We use AI to research and draft our entries and answers. The AI is Claude, made by Anthropic, and the founder directs it. We say so openly because you should know how this site is made.
Our rule on sources
Every claim must carry a source you can open. No source, no claim.
How we check the writing
- A second, separate AI check reads each written answer against the exact lines of the entry it cites. It did not write the answer.
- Only answers that pass are published. Answers that fail are fixed and checked again, or held back.
- Automated checks run on every build of the site. They look at sources, grades, wording and country labels.
- The founder reads a random sample of answers from every batch.
When we cannot answer
Sometimes an entry cannot honestly answer a question. We do not publish an answer. The question goes on a list for a future entry.
Clinical review
No clinician has reviewed these pages yet. Pages say "not yet clinically reviewed" until a named clinician has reviewed them. Named clinical review is planned, starting with pages that carry a harm flag.
Mistakes
AI makes mistakes. If you see one, tell us. Please use our corrections page.
AI assistants can read us too
The site is built so that people and AI assistants can both look things up.
This is information, not medical advice. If you are worried, see a doctor or pharmacist.