Aurora is an AI-driven well-being app, fosters an empathetic connection by learning from users without replacing professionals. Serving as an emotional companion, it encourages daily sharing through voice messages. The app organizes information based on eight parameters, suggesting personalized activities for reflection and self-care. The historical record is also a valuable tool for psychologists to optimize session time by providing an overview of the patient’s emotional state. Aurora ensures user privacy by allowing explicit data sharing with professionals.
How it works
Every entry starts as a voice message, because talking is lower friction than typing when you feel bad. What the user says is turned into structure: each entry is scored against eight well-being parameters, and it is that scoring, accumulated over weeks, that lets Aurora suggest reflection and self-care activities matched to how someone has actually been rather than to a generic programme.
The app runs on a serverless AWS stack (API Gateway in front of Lambda, DynamoDB for storage, Cognito for identity) with a React Native client built in Expo. Serverless was the right fit for a usage pattern that is a handful of short interactions per user per day, and it keeps the surface holding sensitive data small.
The privacy decision
Aurora is explicitly not a replacement for a professional, and the design follows from that. The emotional history it builds is genuinely useful to a psychologist, who can see the shape of the weeks between sessions instead of reconstructing them from memory, but nothing is shared unless the user explicitly shares it. The default is closed, and the handover is an action the user takes, not a setting a clinician can flip.