A privacy-conscious mobile product concept we designed to organize health signals, summaries, forecasting experiments, and AI-assisted insight.
Context
Health information is often fragmented across sensors, exports, apps, and manual logs. We designed Glucose AI to connect those signals without turning the experience into a clinical dashboard.
Responsible scope
We frame the product around organization, explanation, and personal insight. It does not present model output as diagnosis, treatment advice, or clinical validation.
Challenge
We needed to normalize different sources, handle time and device constraints, communicate uncertainty, and keep sensitive information private.
Approach
We designed the application around local-first data, understandable trends, and clear separation between measured data, calculated summaries, model predictions, and AI-generated explanation.
Highlights
- Health data ingestion
- Local storage
- Weekly glucose summaries
- Sleep and heart-rate context
- Meal and nutrition analysis
- Forecasting experiments
- Structured AI insights
- Cross-platform mobile UI
Outcome
We established a product architecture that combines personal health signals, derived metrics, forecasting, and conversational insight while keeping each layer's limitations visible.