How we use AI

Avunara is pre-launch. This page says plainly what is built today, what is planned, and what we will not do.

The workflow

A family member sends a task: a report to explain, an appointment to book, a medicine schedule to track. AI helps read the material, explain it in plain language and draft the next step. A named human care manager reviews anything that matters before it reaches the family or a provider. The AI does not diagnose, prescribe or decide.

The model

Today: the working prototype uses Google's Gemini model. Some integrations in the prototype are still simulated.

Planned: we intend to move the AI layer to Anthropic's Claude and test it against a question set reviewed by clinicians. This is a plan, not a partnership or an endorsement, and we have not completed the move.

What data goes to the model

The prototype demo uses fictional data only. We hold no family data today. Once families are onboarded, only the minimum needed for a task will be sent to a model provider, only with the parent's consent for that purpose. We will publish where data is stored and which providers process it before the first family is onboarded.

Human review

A person reviews AI output before it is acted on. Anything uncertain, urgent or medical is escalated to a human and, where needed, to the family's own doctor. In an emergency, contact local emergency services.

Questions: [email protected]. We have not yet had a legal review or audit of our data practices.