Independent
Local AI Health Companion
PrototypeProduct thinking
Health questions can be hard to organise, and the notes behind them are personal. This prototype explores whether an on-device companion can help someone keep track of symptoms and prepare a clearer handoff without making cloud processing the default. The product value is in organising the questions and context, not presenting an AI response as a diagnosis. A symptom journal and a shareable summary serve that narrower job.
What I built
Gagan built a browser-based health-information prototype with local model inference, a symptom journal, side-by-side responses and a PDF handoff.
A Web Worker runs a quantized model through WebGPU or WASM; messages and journal entries are stored in IndexedDB, and the PDF is generated in the browser.