On-the-spot accident analysis, real‑time accident information & AI‑powered insurance dispute resolution
AutoSureAI is an end‑to‑end intelligent motor‑insurance lifecycle platform.
It uses AI-assisted damage assessment, real-time communication, and role-based claim workflows to automate everything—from accident capture to settlement.
AutoSureAI simplifies the entire motor insurance workflow through:
- AI-powered accident severity prediction
- Real-time agent–driver communication
- Transparent claim dispute resolution
- Complete admin monitoring & analytics
- Secure multi-role access
- Exportable reports & audit logs
The platform ensures faster, fairer, and fraud-resistant claim resolutions.
Traditional insurance claims face:
❌ Manual and slow verification
❌ Human interpretation errors
❌ High risk of fraud & inflated estimates
❌ Lack of real-time communication
❌ No standardized damage evaluation
AutoSureAI solves this with:
✔ AI-driven damage severity estimation
✔ Automated & consistent cost predictions
✔ Real-time notifications
✔ Instant claim routing
✔ Role-specific dashboards
| Layer | Technologies |
|---|---|
| Frontend | React, Vite, TailwindCSS |
| Backend | Node.js, Express.js |
| Database | MongoDB Atlas |
| Realtime Engine | Socket.io |
| Storage | Supabase Buckets |
| Email/OTP | Brevo |
| Machine Learning | DenseNet121 (PyTorch / TensorFlow) |
| Maps & Geolocation | Google Maps / Mapbox |
| Authentication | JWT + 2FA |
[System Architecture Diagram]
## System Architecture
┌────────────┐ multipart/JSON ┌──────────────┐
│ Frontend │ <────────────────────> │ Express API │
│ React/Vite │ HTTPS │ Node.js │
└────┬───────┘ └────┬─────────┘
│ REST / Socket.io │
│ │
│ ┌──────────▼──────────┐
│ │ MongoDB (Atlas/local)│
│ └──────────┬──────────┘
│ │
│ AI inference (image upload) │
│────────────────────────────────────────>│
│ ┌──────────▼──────────┐
└────────────────────────────►│ ML Service (Flask) │
└─────────────────────┘
- Submit accident reports
- Upload vehicle images
- Get instant AI damage assessment
- Track claim progress
- Verify driver‑submitted data
- Communicate live with drivers
- Approve / reject / escalate claims
- Validate accident authenticity
- Approve police verification
- Global stats and dashboard
- Full system audit logs
- Manage disputes, exports, role access
- Instant on‑site accident data capture
- Auto-location detection (GPS + Map API)
- AI-based damage severity estimation
- Upload multiple images + videos
- Auto-generated incident report
- DenseNet121 predicts:
- Minor / Moderate / Severe damage
- Estimated repair cost
- Stores inference results in Supabase + MongoDB
- Live driver ↔ agent chat
- Notifications
- Typing indicators
- Agent reviews
- Fraud pattern checks
- Approval / rejection / escalation
- Global analytics
- Claim heatmaps
- ML accuracy stats
- CSV/PDF export
- Audit logs
- Pretrained model fine-tuned on accident dataset
- Outputs severity + cost estimation
- Runs on Flask inference API
- Deployed using Docker
- Geo-fencing fraud detection
- Multi-vehicle reconstruction
- Premium-adjustment engine
- GenAI-based explanation system
AutoSureAI/
├── Frontend/ # React frontend
├── Backend/ # Express backend
├── ml-service/ # DenseNet121 inference API
└── README.md
MIT License
Hardik Kumar (Lead Developer — MERN + ML)