Resilient Minds includes S'AI, an AI coach for wellbeing and performance. S'AI is a coaching tool. It is not a medical device, a mental health service, a crisis service, or a substitute for professional advice. S'AI suggests; you decide.
1. What S'AI is for
S'AI helps you notice your state, regulate stress, build habits and reflect. It offers one suggestion at a time, such as a breathing exercise or a learning module from the app.
S'AI does not diagnose or treat, and does not give medical, legal or financial advice. It can be wrong. Treat what it says as a coach's suggestion, not as fact.
2. Which AI we use
- Standard service: S'AI runs on Claude, a large language model from Anthropic, through Anthropic's commercial service.
- Private model: for organisations that need it, S'AI runs on a private model on infrastructure we control. For those users, messages are never sent to Anthropic or any public AI service, and chat history is kept on the user's device rather than in their account.
- Voice: when S'AI speaks aloud, the text of S'AI's reply is sent to ElevenLabs to be turned into speech. In the app, dictation is turned into text on your device and your voice is not sent. On the web page for a shared conversation, if you use the microphone, your voice is streamed to ElevenLabs while you speak, so the words can appear as you say them. We don't store the recording; ElevenLabs may keep it briefly under their own terms.
We do not build or train our own AI models.
3. What happens to what you type
- Your messages and check-in responses are sent to the AI model to generate a reply, and stored in your account so the conversation can continue.
- Nothing you type is used to train AI models, by us or by our suppliers. Anthropic's commercial terms prohibit it from training on customer content.
- Anthropic deletes what we send within 30 days, except content flagged under its usage policy, which it may keep longer.
- You are identified to the AI by your username only. We do not send real names, email addresses, location or device identifiers.
- You can delete your account and everything linked to it at any time in Settings.
4. Your agreement and your choice
Before your first conversation, the app shows what is sent, who receives it and how it is protected, and asks you to agree. We record that decision with the date and the version of the wording you saw. If the wording changes, we ask again.
You can decline. The rest of the app, including exercises, learning modules and check-ins, works without S'AI. You can change your mind at any time in Settings.
5. Safety
- S'AI is instructed to put safety first when someone appears distressed, to keep its replies simple and calming, and to encourage human or emergency support when the situation calls for it.
- The app has a Human Support screen, available from the menu at any time: a safety check with one-tap 999, and one-tap contact for Samaritans, Shout, Mind and other independent UK services. S'AI can open this screen itself when it judges you should speak to a person.
- S'AI cannot contact anyone, report on you, or change anything outside the conversation. We do not monitor conversations, so we cannot step in. The app's job is to make human help immediate.
If you are in immediate danger, call 999.
6. What your organisation can and cannot see
If you were given the app by your employer or another organisation, they do not have access to your conversations, check-ins or survey answers. There are no administrator screens showing any individual's information.
7. Human oversight
- S'AI's behaviour is defined by written instructions authored and reviewed by our team, including a practising coach.
- Changes to those instructions are version-controlled and tested before release.
- We review known issues and feedback, and correct S'AI's behaviour through those instructions.
- We do not read individual users' conversations as part of running the service.
8. Security
Information is encrypted when it travels (TLS 1.2 or higher) and when it is stored (AES-256), and hosted on Google Cloud in London. Each account can reach only its own records, which we test automatically, and we commission independent security testing. A security summary is available to client organisations on request.
9. Our suppliers
Anthropic provides the AI model under its commercial terms, data processing agreement and usage policy. ElevenLabs provides text-to-speech, and speech-to-text for the shared-conversation web page, under its terms and data processing agreement. The full list of suppliers that handle user information is in our privacy policy.
10. Changes and contact
We will update this page when the way we use AI changes, and tell client organisations about changes that matter. Questions: contact@resilientmindsmatter.com.