Private, offline health intelligence.
YOUR best doctor. In your pocket.
prepaihq.com · Blog · Privacy · Contact
PrepAI combines on-device AI with a curated clinical knowledge graph to give you a personal health advisor that never sends your data anywhere. It analyzes your sleep, activity, heart rate, nutrition, labs, symptoms, and medications — cross-references everything against peer-reviewed evidence — and surfaces patterns, suggests next steps, and prepares you for doctor visits.
No account. No cloud. No data leaves your device.
📚 Evidence, not opinions. Every insight carries a grade (A–D) and cites its source. When evidence is weak, the app tells you. When a pattern suggests something worth discussing with your doctor, it helps you prepare for that conversation — with the questions to ask and the data to bring.
🧬 Intelligence that knows YOU. The app learns your personal baselines — not population averages. It knows your resting heart rate runs high because of anxiety, so it doesn't flag it. It notices your sleep declining for months, even when each night looks "normal." It connects dots across sleep, heart rate, weight, mood, and labs that no single doctor has time to piece together in a 12-minute visit.
🔐 Privacy you can verify. All AI runs on-device. The knowledge graph lives on your phone. No account required. No cloud processing. No data leaves your device. Open the app, see your data, close it — nothing was ever sent anywhere. Not because we promise. Because the architecture makes it impossible.
🌍 Built for everyone. The person managing a chronic condition. The new parent on 4 hours of sleep. The athlete optimizing recovery. The caregiver tracking medications for someone else. The person who doesn't feel right and can't figure out why. The doctor who wants a second set of eyes on a complex case. Designed to meet people where they are — never judge, always help.
| Domain | Capability |
|---|---|
| 🍽️ Nutrition | AI food scanning (LFM2-VL), nutrition label OCR, 7-day meal plans with shopping lists + prep schedules, macro tracking, food catalog |
| 🏋️ Training | Workout logging (sets/reps/weight/templates), HRV-guided readiness, overtraining detection, exercise prescriptions by condition |
| 😴 Sleep | Quality scoring (efficiency + fragmentation + regularity), chronotype detection, social jetlag, sleep disorder signal detection |
| ❤️ Recovery | HRV trends, resting HR, personal baselines, cross-domain pattern detection |
| 💊 Supplements | Protocol tracking, adherence, evidence-graded suggestions, interaction checking against medications and conditions |
| 🩺 Clinical | Lab result tracking with reference ranges, symptom logging, pattern detection (33 clinical patterns), clinician-ready visit summaries |
| ⌚ Apple Watch | Workout companion (live set tracking + rest timer), quick water/weight/supplement logging, 4 complications |
| 🔒 Privacy | All on-device. No account. No cloud. HealthKit/Health Connect sync is permission-gated and optional. |
┌─────────────────────────────────────────┐
│ React Native (Expo SDK 57) │
│ TypeScript · NativeWind · Zustand │
├─────────────────────────────────────────┤
│ Rust Health Engine (5 modules) │
│ analytics · biomarker · correlation │
│ recommendation · knowledge graph │
│ 84 tests · 0 warnings · iOS + Android │
├─────────────────────────────────────────┤
│ On-Device AI (ExecuTorch) │
│ Llama 3.2 (text) · LFM2-VL (vision) │
│ OCR (nutrition labels) │
├─────────────────────────────────────────┤
│ SQLite — user data + knowledge graph │
│ HealthKit · Health Connect │
└─────────────────────────────────────────┘
YOUR DATA THE ENGINE YOUR INSIGHT
───────── ────────── ────────────
Sleep · HRV · HR 5 Rust modules Personal baselines
Activity · Weight 84 tests · 0 warnings Pattern detection
Nutrition · Labs iOS + Android Evidence-graded
Symptoms · Meds All on-device recommendations
│ │ │
└────────────────────────────────────┴──────────────────────────────┘
│
KNOWLEDGE GRAPH
197 entities · 550+ relations
60 biomarkers · 33 clinical patterns
63 interventions · 16 guideline rules
| Layer | Technology |
|---|---|
| 🖥️ Framework | Expo SDK 57, React Native 0.86, Hermes, New Architecture |
| 📝 Language | TypeScript (strict) + Rust |
| 🗄️ Database | SQLite (local, versioned migrations, FTS5 search) |
| 🧠 AI Runtime | react-native-executorch (LLM + vision + OCR, all on-device) |
| ⚙️ Knowledge Engine | Rust → C FFI → JSI → React Native |
| ❤️ Health | HealthKit (32 read, 7 write) + Health Connect (30 read, 9 write) |
| 🎨 Styling | NativeWind v4 (Tailwind CSS) |
| 🌐 Web | Expo static export → Cloudflare Pages |
| 🚀 CI/CD | GitHub Actions + EAS Build + EAS Submit |
| Phase | FRDs | Status | Focus |
|---|---|---|---|
| A — Ship | 7 | 95% | App Store submission, vision pipeline, meal planning, Apple Watch |
| B — Knowledge | 5 | 85% | Rust engine, knowledge graph (197 entities), Android parity |
| C — Analytics | 9 | ✅ 100% | 5 Rust modules, clinical KG expansion (172 seed records), 84 tests |
| D — Harden | 9 | ← Next | Accessibility, resilience, telemetry, AI fallback, security |
| E — Expand | 6 | Research | 500+ conditions, automated research ingestion, personal baselines |
| F — Personalize | 8 | Research | Trajectories, family history, "What Am I Missing?" CDS |
| G — Everyone | 8 | Research | Harm reduction, mental health, life stages, social determinants |
| H — Scale | 4 | Future | Millions of users, clinical validation, multi-language AI |
bun install
bun run dev # iOS / Android
bun run dev:web # Landing page
bun run typecheck # Zero errors
bun run test # All passing