AIHealthIQ

Your smartwatch sees patterns in your sleep, heart rate and activity that nobody else notices. Built an AI app that reads your watch data. It actually tells you when something's off, in plain everyday words.

• 01User surface

  The whole experience the user sees

  This frame shows the live product: ai reads your wearable data, spots changes, recommends treatment and lifestyle tweaks. Every component is ours · scope, design, code, deploy.

• 02Stack behind the screen

  What's powering it: Express, React-Redux, SQLite

  4 stack components run behind this frame · Express, React-Redux, SQLite drive the visible UI; the rest sit in the data layer. All studio-owned.

• 03What we shipped

  Wearable data ingestion

  Someone always watching your data

• 04Status

  Private deploy · under NDA.

  Per the client's request the URL stays private · the build, architecture, and lessons can be shared in a scoping call.

• 01 · BRIEF

  Map the wearable signals worth pulling.

  Workshop with the founders to pin down which four signals (HR, sleep, steps, SpO2) actually drive the recommendations · everything else dropped from v1 scope.

• 02 · ARCHITECTURE

  Stack decisions before any code.

  Decision doc captured the data flow, Express, React-Redux, SQLite, OpenRouter role split, and the failure modes we'd handle in v1 vs defer. Cross-service boundaries (where AI ends and the web app begins) were drawn here so neither side leaked into the other later.

• 02 · BUILD

  Express ingest + LLM correlation + Redux UI.

  Express backend handles per-vendor wearable feeds, normalises, persists in SQLite, then OpenRouter routes to the model best-suited per signal type. Redux drives the React + mobile UIs off the same store shape.

• 04 · POLISH

  Performance, accessibility, and observability.

  PSI / a11y / coverage budgets enforced as launch gates. Logging + metrics wired before cut-over · the team can answer 'is it working?' from a dashboard, not a Slack thread. Threat-model checklist signed off before traffic hits the box.

• 03 · SHIP

  Web, iOS, Android, plus the public API.

  Three end-user surfaces shipped on the same correlation engine, plus a HealthApplication-typed JSON-LD'd API for clinic integrations.

• 01Ingest

  Per-vendor wearable feeds, normalised

  Apple Health, Garmin, Fitbit, Oura · normalised to a single internal schema before correlation.

• 02AI

  OpenRouter-routed correlation engine

  Different signals route to the model that handles them best · cost-aware fallback per request.

• 03UI

  Web + iOS + Android off one store

  Redux store shape shared across React web and React-native mobile · one source of truth for trend cards.

• 04API

  Clinic-grade JSON API

  HealthApplication-typed schema · clinics can pull patient summaries on demand.

• 
  01Frame

  Cardiovascular dashboard · risk stratified

  Anatomical view + biological age (47) vs chronological age (54) + six risk cards (atrial fibrillation, stroke, coronary artery disease, hypertension, heart failure) · the cardio system reads itself.

• 
  02Frame

  Body view · cholesterol cluster highlighted

  While 'we're checking your Cholesterol' progresses, the heart highlight pulses · the user sees the AI's attention land on the right organ in real time.

• 
  03Frame

  Lifestyle prescriptions · concrete actions, with cart

  Quit smoking, train zone 2, low-GI diet, manage stress, fruits + veggies, reduce sugar · each prescription is a card with the rationale and an 'Add to cart' that links to a supplement or device, no vague 'consult a doctor'.

• 
  04Frame

  Action plan · supplements + lifestyle + escape hatch to physician

  Beetroot, turmeric, omega-3 land in supplements; lifestyle steps follow. Bottom CTA 'Check-in with our Physician first' is the safety valve when the user wants a human in the loop.