APPZCART  /  Investor Handbook
CHAPTER 04 · PRODUCT

The build.

Product roadmap, tech architecture, clinical credibility framework, customer personas, and the growth operations playbook that turns downloads into trusted family vaults.

10 — PRODUCT & TECHNICAL

The engineering story.

Build depth investors test for
ROADMAP

12 / 24 / 36 month product roadmap

Y1: Vault, nutrition (IFCT/ICMR), AI insights, doctor connect, ABHA pilot. Y2: MediaPipe workout AI, family health score, insurance dashboard, API marketplace. Y3: Predictive health AI, chronic care modules, regional expansion features, B2B white-label.

API STRATEGY

Licensing & developer ecosystem

Nutrition macro API, AI health-insight API, vault API. Tiered pricing (free dev tier, growth, enterprise). Documentation portal, sandbox, OAuth-secured. Target: 50 paying API customers by Year 2.

PLATFORM DECISION

Mobile-first, web-companion

React Native for cross-platform mobile (Android-first in India), Next.js for the web companion and doctor dashboard. Native modules for camera (MediaPipe), biometric auth, and notifications.

CONNECTIVITY

Offline-first architecture

Tier 2/3 connectivity is patchy. Local-first data with conflict-free replicated sync (CRDT or similar). Vault entries, prescription uploads, AI prompts queue locally and sync when online. Critical for trust.

LOCALIZATION

Full UX in 6+ Indian languages

Not just translation — vernacular UX. Hindi, Tamil, Telugu, Kannada, Marathi, Bengali. Voice input in regional languages. Text-to-speech for low-literacy users. Cultural icons, family-structure semantics, festival-aware reminders.

DATA ARCHITECTURE

Family permissions model

Role-based: Head, Spouse, Adult Child, Minor, Elderly Parent, Doctor (limited). Per-record visibility. Consent flows logged. Minors' data has guardian-only access. Audit trail on every read/write for ABDM compliance.

AI EVALUATION

How we measure AI insight accuracy

Quarterly clinical review board reviews 500 random AI insights against medical guidelines. Target: 95% safe + appropriate, <1% potentially harmful. Continuous human-in-the-loop refinement. Published accuracy reports build trust.

TECH STACK

Production-grade foundation

Spring Boot modular monolith (Java 21), Spring AI + Anthropic Claude, PostgreSQL + pgvector, Redis, MinIO for vault files, React Native + Next.js, AWS Mumbai (India data residency). Stripe + Razorpay for payments.

11 — CLINICAL CREDIBILITY

Medical trust is the deepest moat.

The silent risk most founders miss
MEDICAL ADVISORY BOARD

4 named clinical advisors

(1) Practicing physician with chronic-care expertise. (2) Registered dietitian / nutritionist. (3) Preventive care specialist or public health doctor. (4) Pediatrician (family vaccines, child health). Each gets equity + retainer. They review AI logic and sign off on clinical guidelines quarterly.

VALIDATION PROTOCOL

Clinical validation framework

Every AI feature passes: (1) Literature review against ICMR/WHO guidelines. (2) Sample-output review by medical board. (3) Beta rollout with safety telemetry. (4) Quarterly audit against real outcomes. Failure mode protocol — escalate or disable feature.

LIABILITY & DISCLAIMERS

Legal safety architecture

Every AI insight carries a clear "not medical advice" disclaimer plus "consult a doctor" CTA. Emergency keywords (chest pain, severe bleeding) trigger immediate doctor escalation, not AI response. Indemnity insurance — professional liability + cyber + product.

INSTITUTIONAL PARTNERSHIPS

Research & credibility tie-ups

Target a medical college (CMC Vellore, AIIMS, JIPMER) or research institution for a joint research MoU. Publish a peer-reviewed paper on family-health vault outcomes by Year 2. Builds investor credibility and unlocks AYUSH/ICMR endorsement pathways.

ENDORSEMENT PATHWAY

ICMR / AYUSH / state recognition

ICMR endorsement for nutrition tooling (we already use IFCT 2017 + ICMR 2020 RDA). AYUSH integration for traditional medicine modules. State health department pilot agreements — one state by Year 1 end becomes a powerful reference for the next 10.

SAFETY OPERATIONS

Adverse event & complaint protocol

Documented SOP: user reports an issue → 4-hour triage → clinical board review → user response within 24 hours → if pattern, feature pause → root cause → published learning. Transparent log builds long-term trust.

12 — GROWTH OPERATIONS

The customer engine, demystified.

Personas · onboarding · support · churn
P1

Persona 1 — The Sharma family (Coimbatore)

Father 48, IT services manager, ₹12 L/year. Mother 44, homemaker, manages family health. Two children (16, 12). Father has borderline diabetes; mother has thyroid; daughter has PCOS concerns. Decision-maker: mother. Trigger: father's recent diagnosis. Smartphone-first, WhatsApp-native, prefers Tamil voice and English text.

P2

Persona 2 — The Verma joint family (Nagpur)

Three generations under one roof. Adult son (35) is the smartphone-savvy member. Grandparents have chronic conditions. Two kids needing vaccinations. Decision-maker: adult son. Trigger: grandmother's prescription confusion. Hindi-first UX critical.

P3

Persona 3 — The young couple (Tier 2 city)

Late 20s, newly married, planning a child. Both working professionals. Trigger: pre-pregnancy planning. Already use fitness apps. High intent, high willingness to pay, low data — they grow into the family vault as their family grows.

ONBOARDING

App download → first vault entry in <10 minutes

3-screen signup (mobile OTP, ABHA optional, language). Add primary user + 1 family member in 2 minutes. Upload first prescription via camera with OCR — AI extracts in 30 seconds. First AI insight delivered within first session. Hook: family health score visible immediately.

SUPPORT MODEL

WhatsApp-first vernacular support

WhatsApp Business API for chat support in Hindi, Tamil, Telugu, Kannada, English. Voice-call escalation for high-value users. Hours: 8am–10pm IST, 7 days. Median first response <15 minutes. AI-assisted agents handle 70% of queries.

CHURN FRAMEWORK

Why users leave — and the win-back

Top churn reasons: (1) didn't see value in week 1. (2) family member didn't adopt. (3) payment failure. Win-back: behavioral nudges day 3 + day 7 + day 14, free month for adding family member, dunning sequences for payment failures. Target churn <5%/month at scale.

ENGAGEMENT LOOPS

Daily & weekly habit hooks

Daily: medication reminders, hydration nudges, family health tip in vernacular. Weekly: family health score update, AI insight summary, preventive alerts. Monthly: family wellness report, doctor consult prompts. Target: 4+ sessions/week per active user.