Orlivas — AI Transparency Notice
Version 1.0.0 · Effective 2026-07-11 · Last updated 2026-07-11
This notice explains how Orlivas uses AI, what data an AI feature may process, where that
data goes, and how you stay in control. It supplements the Privacy Policy and the Health
& Wellness Disclaimer.
1. Which features use AI
| Feature | What it does | Input it uses |
|---|---|---|
| Visual Wellness | Scores a selfie for appearance-based wellness signals and writes a short non-diagnostic note | The wellness selfie image + the enabled metric list |
| AI Insights | Daily/weekly/monthly summaries of a member's or family's logged activity | A compact health-activity summary (member display name + logged entries/values) |
| AI Assistant | Answers your questions about a family member using their logged data | The same health-activity context + your question |
| Translation (internal) | Drafts translations of app/content strings | UI/content strings only — no health data |
2. Where AI runs, and what leaves Orlivas
- By default, AI runs on Orlivas' own servers using a built-in engine (the
"simulated" provider). In this default mode, **no health data is sent to any third
party** for AI purposes.
- Orlivas can be configured to use an external AI provider. The provider categories
are large-language-model providers and vision-capable model providers. The
providers integrated in the system are Google (Gemini), OpenAI, and
DeepSeek (text only).
- When an external provider is used:
- Visual Wellness sends the selfie image to a vision provider.
- AI Insights / AI Assistant send a **text summary of the member's logged health
activity, including the member's name**, (and, for the Assistant, your question) to a
text provider.
- Orlivas keeps an internal execution log of AI calls (a preview of the request and
response, timing, token counts and the related family/member id) so operators can
audit and improve the features.
3. Your control and consent
- The AI features are optional and are offered on certain paid plans.
- **We will not send your family's health data (text or photos) to an external AI
provider without your separate, explicit consent.** Our intended model is
default-off / opt-in for external AI on health data, with the ability to decline or
turn it off at any time.
4. Limits of AI output
- AI output is non-diagnostic and may be wrong, incomplete or out of date.
- Do not rely on AI for medical, safety, financial or legal decisions. The AI never
diagnoses, never names diseases, and always defers to a qualified professional.
- AI does not make automated decisions that produce legal or similarly significant
effects about anyone. A human (you, the caregiver) decides what to do with any output.
5. Provider data use and training
Whether an enabled provider retains or uses inputs to train its models depends on that
provider's contract, which is set at configuration time:
(available on request). Our commitment is to not allow third parties to train
models on your family's records, and to use zero-retention / no-training API tiers and a
data-processing agreement for any provider enabled on health data.
6. Feedback and correction
If an AI result seems wrong or inappropriate, you can delete it and contact us at
support@orlivas.com. Your feedback helps us tune prompts and provider choices.
7. Contact
AI and privacy questions: privacy@orlivas.com