Skip to content

AI Transparency

Last updated 8 November 2026 · Adam Richard Turner, trading as Sleep Coach

Sleep Coach uses generative artificial intelligence in two optional features. This notice explains what is generated, what data is shared, and how to report a problem. It forms part of the Terms of Use and the Privacy Policy.

1. What is generated

Two features use generative AI: Chat (ask questions about your sleep habits) and Sleep review (a short written read of recent nights). Both produce new text from a large language model. A person does not review each reply before you see it.

Everything else — the diary, charts, night verdicts, sleep window maths, weekly check-in, reminders, worksheets and educational articles — is ordinary software and pre-written copy, not live AI.

2. The AI provider

Replies are generated by Claude models provided by Anthropic, PBC (United States) through its commercial API. Under Anthropic’s commercial terms, inputs and outputs are not used to train its models. Anthropic acts as my processor and may retain data for a limited period for trust and safety purposes under its terms.

4. What the model can see

When you send a message or run a review, the App may include: recent diary nights (times, durations, efficiency, stages, ratings and factors you logged), your wake time and sleep window, and the words you typed. Your name, email address and account identifiers are not sent.

Avoid entering other people’s information, full medical records, or anything you would not want a processor to see.

5. Safeguards and limits

  • The model is instructed that it is not a doctor, therapist or clinician; to stay on everyday sleep habits; to never diagnose, prescribe, treat or tell anyone to change medication; and to never invent numbers.
  • AI screens tell you that replies are generated by AI, can be wrong, and are not medical advice.
  • Clearly off-topic requests are filtered out before they reach the model.

The model can still make mistakes, sound confident about something untrue, or miss a safety issue. Check numbers against your diary and ignore anything that conflicts with advice from a qualified clinician.

6. What the AI is not

The assistant is not a doctor, therapist or CBT-I practitioner and does not provide a regulated health service. Its replies are educational suggestions about habits such as a regular wake time, winding down, or using the bed only for sleep.

7. Reporting a problem

If an AI reply is harmful, offensive, inaccurate or unsafe, please report it to hello@adamrichardturner.dev with the date and time of the message (a screenshot helps). I review every report, and I use them to improve the instructions and safeguards. Reports are not used to train any AI model.

8. Transparency obligations

This notice is intended to meet the App Store Review Guidelines on disclosing data shared with third-party AI, UK GDPR transparency requirements, and the transparency duties for AI systems that interact with people under the EU Artificial Intelligence Act (Article 50), where applicable.

Questions about this document? Email hello@adamrichardturner.dev.