What the AI sees, and what it never sees.
Written so you can decide whether to use it. It is entirely optional.
About AI-assisted reflection
AI-assisted reflection can help you explore your thoughts and experiences. It is not a replacement for professional mental health care and should not be used for emergencies.
- AI responses may not always be accurate or appropriate.
- The AI does not diagnose, and it is not a therapist, doctor or clinician.
- Please seek professional or urgent support when you need it.
What is sent
- The text of the message you choose to send.
- Up to a small number of recent turns from the same conversation, so the reply makes sense.
- Audio you record for transcription, if you use dictation and the on-device engine is unavailable.
What is never sent
- Your name or any profile details.
- Your journal entries.
- Saved reflections, insights or letters.
- Your settings or preferences.
- Anything you have written but not sent.
This is enforced where the request is built, not only as a policy statement.
Why it is sent
To generate a reflective response to what you have written, and to transcribe audio into text when you use dictation.
What we can and cannot confirm
Where something has not been independently verified, it says so rather than sounding more certain than it is.
We do not store your AI conversations on a server of our own. Your side of the exchange stays on this device. The AI gateway that carries the request does keep a record of every request — always summary information (status, model, size, cost, timing), and, if request visibility is switched on for this app, the full text of the request and the reply for up to 90 days. Request visibility is not switched on for this app.
Confirmed
We do not use your content to train any model. The gateway operator's own terms allow customer data to be used for model training on some account plans unless the operator is asked to opt the account out, and state that raw or identifying personal information is not used for training. The exact position for this app's account has not yet been confirmed in writing, so treat this as unresolved.
Not independently verified
Requests are processed by the AI gateway's infrastructure and by the underlying model providers it routes to. The processing location has not been independently verified by us and may involve transfers outside the UK.
Not independently verified
Contractual safeguards with the AI provider have not yet been reviewed. This will be completed before commercial launch.
Not independently verified
What the service itself keeps
SOLISURE holds almost nothing about you. This is the exception, described exactly.
SOLISURE is designed to be local-first: your journal, reflections and settings stay on this device unless you deliberately use a feature that sends something elsewhere. There is one exception, and this is it. To stop the AI features being abused and to keep AI costs within a fixed daily ceiling, the service keeps a small anonymous usage record: a scrambled random identifier for this browser, the date, and counters for how many AI reflections and transcriptions were requested and how many seconds of audio were transcribed. These records are deleted after about 30 days. They contain no journal content, no prompts, no AI replies, no audio, no transcripts and no crisis information, and the identifier is random rather than derived from you or your hardware.
Confirmed
Using SOLISURE without AI
AI-assisted reflection is never required. Every other part of the app — journalling, CBT-informed reflections, DBT-informed skills, grounding and crisis information — works without it. You can withdraw your acknowledgement at any time in Profile → Privacy & data.
Provider and modelsLovable AI Gateway
AI-assisted reflection uses google/gemini-3.6-flash. Dictation, when the on-device engine is unavailable, is transcribed by openai/gpt-4o-mini-transcribe. Both are reached through the same gateway, and everything on this page applies to both.
Requests are made from the app's own server, so the provider credential is never present in the app or in your browser. A model may change as better ones become available; this page and the privacy policy are updated when it does.