Trust & policies
AI Usage & Disclosure
What is sent, when external AI may be used, and the controls around optional feedback.
When AI is used
Opening or writing in Reflective Signal or Mentor does not automatically send content to an external model. A request occurs only when the learner explicitly activates feedback. If no API key is configured—or Mentor local-only is selected—the application uses a deterministic/local path.
What may be sent
Reflective Signal sends the active response and bounded scenario/rubric/context needed for feedback. Mentor sends the text the learner pastes plus its stated document type, audience and objective. The application does not intentionally attach unrelated learning history, raw workplace files, billing or payment information, security logs, or secrets.
Provider
The only external model provider implemented is the OpenAI API. No external request is made unless it is configured. The live project, model, region and retention controls must be verified before public activation.
Outputs and limitations
AI feedback is generated, may be wrong and is not a factual authority. It is educational and qualitative—not a certification, employment evaluation, legal/medical advice, psychological diagnosis or guarantee. Learners remain responsible for reviewing and applying it.
Privacy choices
Use fictional or anonymized material. Do not submit confidential, proprietary, client-identifying or sensitive personal information. Mentor supports a local-only request mode. The authenticated account-wide local-only preference is enforced for both Coach and Mentor.
Failures
Timeouts, invalid output, provider outages and unsafe responses return bounded errors or local fallbacks. Failed requests do not authorize invention or silent provider switching.