AI-Powered Parametric Income Protection for India's Gig Workers
When rain halts deliveries, when heat waves shut down your zone, when a bandh steals your workday — Lattice Protocol detects it automatically and streams money into your wallet. No claims. No forms. No waiting.
India has 7.7 million gig workers (Swiggy, Zomato, Uber, etc.) who lose ₹800–₹2,500/week to external disruptions — heavy rainfall, extreme heat, flooding, pollution, bandhs, and platform outages. Traditional insurance doesn't work for them: annual premiums are too expensive, claim processes take weeks, and fraud is rampant with GPS spoofing.
Lattice Protocol is a parametric micro-insurance platform that:
- Detects disruptions automatically using real-time weather APIs, AQI data, and news feeds
- Fires parametric triggers when scientifically measurable thresholds are breached (e.g., rainfall > 30mm/hr for 2+ hours)
- Validates claims with AI using an Isolation Forest model that cross-checks GPS, cell tower, accelerometer, Wi-Fi, and peer cluster signals
- Pays workers instantly via UPI — zero forms, zero claims, zero waiting
Unlike traditional insurance where you file a claim and wait, parametric insurance pays based on the event itself, not the damage. If it rained >30mm in your zone for 3 hours, you get paid. Period. No adjuster, no form, no argument.
src/
├── app/
│ ├── layout.js # Root layout + SEO + PWA
│ ├── page.js # Landing page (marketing)
│ ├── globals.css # Complete design system (400+ lines)
│ ├── onboard/page.js # 5-step worker onboarding
│ ├── plans/page.js # Pricing tiers + premium formula
│ ├── api/
│ │ ├── weather/route.js # OpenWeatherMap integration + fallback
│ │ ├── premium/route.js # XGBoost premium calc + SHAP output
│ │ └── simulate/route.js # Full pipeline simulation endpoint
│ ├── dashboard/
│ │ ├── layout.js # Sidebar wrapper
│ │ ├── page.js # Worker overview dashboard
│ │ ├── policy/page.js # Active policy detail
│ │ ├── payouts/page.js # Payout history
│ │ └── weather/page.js # Live conditions + trigger status
│ └── admin/
│ ├── layout.js # Sidebar wrapper
│ ├── page.js # Platform analytics
│ ├── map/page.js # Interactive Leaflet risk heatmap
│ ├── triggers/page.js # Live trigger monitor
│ ├── fraud/page.js # Isolation Forest fraud detection
│ ├── simulation/page.js # Live pipeline simulation demo
│ ├── explainability/page.js # SHAP waterfall chart
│ └── financials/page.js # Revenue & unit economics
├── components/
│ ├── Navbar.js # Landing navigation
│ ├── Sidebar.js # Dashboard/admin sidebar
│ ├── RiskMap.js # Leaflet map with zone overlays
│ ├── WhatsAppPreview.js # Payout notification preview
│ └── CountUp.js # Scroll-triggered animations
└── core/
└── data.js # Business logic + premium engine
- Model: XGBoost regression trained on historical disruption data
- Features: Zone risk factor, seasonal multiplier, flood history, AQI severity, rider density
- Explainability: SHAP waterfall chart shows exactly why your premium is ₹59 vs ₹75
- Formula:
Premium = Base Rate × Zone Risk × Season Multiplier + Plan Surcharge
- 7 signals validated per claim: GPS location, cell tower, Wi-Fi BSSID, last order timestamp, accelerometer pattern, claim frequency, peer cluster detection
- Results: 91.8% auto-approved, 5.5% auto-rejected, 2.7% manual review
- False positive rate: 1.8%
- Why Isolation Forest: Short path length = anomaly. Works without labeled fraud data (unsupervised).
- 6 disruption types: Heavy rainfall, extreme heat, severe AQI, urban flooding, curfew/bandh, platform outage
- 7-step pipeline: Data Ingestion → Normalization → Threshold Engine → Geo-Fence Match → Fraud Detection → Payout Calculation → UPI Disbursement
- Latency: Full pipeline executes in < 5 seconds
| Feature | Description |
|---|---|
| 15 Pages | Landing, onboarding, plans, worker dashboard (4 pages), admin panel (7 pages) |
| 3 API Routes | /api/weather, /api/premium, /api/simulate |
| Interactive Risk Map | Leaflet.js with dark tiles, color-coded zone overlays across 5 Indian cities |
| Live Trigger Simulation | Watch the full 7-step pipeline animate in real-time with cascading payout notifications |
| SHAP Explainability | Interactive waterfall chart — change city/zone/plan and see premium recalculate with reasons |
| Fraud Signal Analysis | Case-by-case breakdown showing GPS, cell tower, accelerometer, Wi-Fi, peer cluster validation |
| WhatsApp Preview | Shows exactly what the worker receives — payout message + shareable receipt card |
| Financial Projections | CSS-only bar charts, unit economics table, revenue stream breakdown |
| Real Weather API | OpenWeatherMap integration with climate-aware simulation fallback |
| PWA Support | Manifest + mobile-first design for worker phones |
# Install dependencies
npm install
# Run development server
npm run dev
# Open in browser
open http://localhost:3000# For real weather data (free tier: https://openweathermap.org/api)
OPENWEATHER_API_KEY=your_api_key_here/— Landing page: the pitch, the problem, the solution/onboard— Walk through 5-step signup (Phone → Platform → Zone → Plan → Pay)/dashboard— "Here's what the worker sees every morning"/dashboard/weather— Live conditions with trigger status for all 5 cities/admin— Platform-wide analytics (12,800+ workers, ₹48L+ payouts)/admin/map— Interactive risk heatmap of India/admin/simulation— ⚡ Inject a flood in Mumbai and watch money flow to workers live/admin/explainability— SHAP waterfall: "Here's how our AI isn't a black box"/admin/fraud— Click each case: "Here's how we catch GPS spoofers"/admin/financials— Unit economics proving business viability
| Other hackathon teams | Lattice Protocol |
|---|---|
| Call weather API, update a database row | 7-step evaluation pipeline with geo-fencing, fraud scoring, and UPI payout |
| Form-based claims | Zero-claim parametric triggers — worker never files anything |
| "AI" = calling an API | Signal-by-signal Isolation Forest analysis with 7 cross-validation checks |
| Annual premium model | Weekly micro-premiums (₹29-99) aligned with gig worker payout cycles |
| One-size-fits-all pricing | XGBoost-driven zone risk factors with SHAP explainability |
| Generic Bootstrap UI | Hand-crafted 400-line CSS design system with premium dark-mode aesthetic |
| Static demo | Interactive map, live simulation, animated pipeline, cascading payout feed |
- Framework: Next.js 14 (App Router)
- Styling: Vanilla CSS (400+ line bespoke design system)
- Maps: Leaflet.js + react-leaflet
- Fonts: Inter + JetBrains Mono (Google Fonts)
- APIs: OpenWeatherMap (weather data)
- AI/ML: XGBoost (pricing), Isolation Forest (fraud detection), SHAP (explainability)
- PWA: Web App Manifest + mobile-first responsive design
DevTrails — Built for the Guidewire Hackathon 2026
"Don't let rain steal ₹1,200 from your pocket this week."