🚀 Kizuna ⚡

AI-Powered Income Protection for Gig Delivery Partners

Guidewire DEVTrails 2026 | Team Kizuna

---

🧩 The Problem

Gig delivery partners face unpredictable income loss due to:

• Extreme weather (rain, heat, floods)
• Pollution (AQI spikes)
• Platform disruptions (app/store downtime)

There is no real-time income protection, and workers bear 100% of the financial risk.

---

💡 The Solution

Kizuna is a parametric, event-driven income protection system that:

• Detects disruptions via APIs
• Estimates income loss using AI
• Pays 20–30% of lost income instantly

---

🧠 Core Principle

Kizuna provides a partial income buffer, not full replacement

• Keeps premiums affordable
• Prevents misuse
• Ensures long-term sustainability

---

⚙️ Unified System Design

Kizuna is a trust-driven system, powered by:

👉 Trust Score

Controls:

• Premium
• Payout %
• Claim limits
• Settlement speed
• Fraud penalties

---

🔢 Trust Score Engine

TS = clip(50 + 0.20I + 0.30R + 0.30A - 30F, 0, 100)

Definitions

• I (Identity): 100 if e-Shram verified, else 0
• R (Rating): 20 × avg rating (1–5)
• A (Activity): Recency-weighted (0–100)
• F (Fraud): Number of fraud events

Each fraud event reduces score by 30 points and removes benefits temporarily.

---

🧭 Onboarding Flow

Step 1 — Platform Verification

• OTP + Worker ID
• Checks:
  • Active account
  • Weekly orders > 20
  • Rating ≥ 3.5

---

Step 2 — e-Shram Integration (Optional but Recommended)

🔐 What is e-Shram?

• Government of India database of unorganised workers
• Acts as a verified identity layer (not self-declared data)
• Ensures the user is a real delivery partner / gig worker
• e-Shram verification gives a UAN (Universal Account Number) — a unique ID used to verify identity and prevent duplicate or fake worker accounts

---

⭐ Benefits (Trust Score Boost)

e-Shram verified users receive:

• +15 Trust Score
• Lower premiums
• Higher claim limits
• Faster payouts

---

🛡 Fraud Prevention Impact

Prevents:

• Duplicate identities
• Fake workers
• Multi-account abuse

Adds a government-backed validation layer, improving system reliability

---

Step 3 — Zone & Shift

• Micro-zone selection
• Risk classification

---

Step 4 — Coverage Selection

Dynamic Premium Preview

Premium is dynamically calculated based on:

• Zone risk
• Trust score
• Historical disruption frequency

---

💰 Weekly Premium Model

Zone	Premium Range
Low	₹29–₹39
Medium	₹45–₹65
High	₹75–₹99
Extreme	Dynamic

Final premium dynamically adjusted using risk multipliers.

---

🛡 Coverage Scope

✔ Income loss only
❌ No health, vehicle, pandemic, or systemic coverage

---

⚡ Triggers (Phase 2 — Implemented)

• Rain / Weather: Triggered when rainfall exceeds threshold disrupting deliveries
• Pollution (AQI): Triggered when AQI crosses unsafe level restricting outdoor work
• App Downtime: Triggered when platform app is unavailable beyond threshold duration
• Dark Store Downtime: Triggered when assigned store is non-operational
• Incentive Chain Break: Triggered when rider loses incentive due to disruption

---

🔮 Phase 3 Triggers

• Account Freeze (Selective): Covered only if not caused by platform fraud detection
• Curfew / Social Disruption: Triggered when access to delivery zones is restricted
• Infrastructure Failure: Triggered during network/electricity outages affecting work

---

💸 Coverage Tiers

🟢 Basic

• 25% payout
• 2 claims/week

🔵 Plus

• 30% payout
• 2 claims/week

🔴 Pro

• 40% payout
• 3 claims/week

---

⚡ Payout System

Trust Level	Payout %
High	30%
Medium	25%
Low	20%

---

🧠 Earnings Fingerprint

Each worker has a personalized earning rate:

Earning Rate = Total Earnings / Active Hours (recency-weighted)

Estimated Loss = Earning Rate × Time Lost

Payout = Tier % × Estimated Loss

👉 Same disruption → different payouts per worker

---

🛡 Fraud Detection

• GPS validation
• Device fingerprinting
• Activity verification
• Duplicate claim prevention

---

🌍 Zone Risk Model

Zone	Risk
Low	Stable
Medium	Moderate
High	Frequent disruptions
Extreme	Severe

---

📊 Actuarial Model & Stress Testing

Key Formula

1. Adjusted Payout

P' = N × D × (Income × Cap%) × Claim Factor

2. Insurer Cost

Cost = P' × (1 - Platform Share) × Retention

3. Loss Ratio

Loss Ratio = Cost / Premium Pool

---

🔤 Definitions

• P′ (Adjusted Payout)
  → Total payout after applying caps and claim limits

• N (Number of Riders Affected)
  → Total number of insured workers impacted in the event

• D (Disruption Duration)
  → Number of days (or equivalent time) the disruption lasts

• Income
  → Average daily income per rider (₹1200 baseline or personalized via earnings fingerprint)

• Cap% (Payout Cap Percentage)
  → Maximum % of income that can be paid (e.g., 20%)

• Claim Factor
  → Adjustment for claim limits
  (e.g., if only 5 out of 12 days are claimable → 5/12)

---

Scenario 1 — Multi-City Monsoon

• Riders (N) = 5000
• Days (D) = 7
• Income = ₹1200
• Cap = 20% → ₹240
• Claim factor = 4/7
• Platform share = 20%
• Retention = 20%

Calculation

P' = 5000 × 7 × 240 × (4/7) = ₹48,00,000
Cost = 48L × 0.8 × 0.2 = ₹7.68L
Premium Pool = ₹12L

Loss Ratio = 7.68 / 12 = 64%

✅ Stable

---

Scenario 2 — Extreme Pollution

• Riders = 2000
• Days = 12
• Cap = ₹240
• Claim factor = 5/12
• Platform = 20%
• Retention = 20%

Calculation

P' = 2000 × 12 × 240 × (5/12) = ₹24,00,000
Cost = 24L × 0.8 × 0.2 = ₹3.84L
Premium Pool = ₹4.8L

Loss Ratio = 3.84 / 4.8 = 80%

✅ Controlled

---

Scenario 3 — App Failure

• Riders = 1000
• Duration = 1.5 days
• Payout/day = ₹240

Calculation

P' = 1000 × 1.5 × 240 = ₹3.6L
Cost = 3.6L × 0.8 × 0.2 = ₹0.576L
Premium Pool = ₹2.4L

Loss Ratio = 0.576 / 2.4 = 24%

✅ Highly Safe

---

🧠 Key Insight

Even under extreme correlated events, loss ratio remains controlled using caps, reinsurance, and dynamic pricing.

---

🛠 System Safeguards

• 70% → Premium adjustment

• 85% → Enrollment restriction

• 100% → Emergency controls

---

⚡ Catastrophic Handling

• Auto-trigger payouts
• No manual claims
• Strict caps applied

---

🤖 Why Parametric?

• Instant payouts
• No paperwork
• API-triggered events
• Low operational cost

---

💼 Business Model

• Insurer → Underwriting
• Platform → Distribution
• Kizuna → AI + system

---

🏆 Moat

• Trust score engine
• Earnings fingerprint
• Fraud detection
• Zone pricing

---

🚀 One-Line Pitch

“Kizuna protects gig workers by instantly paying a portion of lost income when real-world disruptions occur.”