AIVIS (AI Vehicle Insurance Investigation System) is a world-class, enterprise-grade platform engineered for Vehicle Digital Forensics, Multi-Agent AI Fraud Investigation, Knowledge Graph Analytics, Commercial OEM Telematics Gateways, and Compliance Auditing.
Designed for Insurance Carriers, Vehicle Forensic Analysts, OEM Engineers, Fleet Operators, and Government Agencies, AIVIS combines high-density digital telemetry, cryptographic evidence vaulting, multi-agent AI debates, and real-time knowledge graph mining to detect, investigate, and prevent complex vehicle insurance fraud.
graph TD
Client[React 19 + TypeScript + White Claymorphism UI] --> API[FastAPI v6.0 REST & WebSockets]
API --> Auth[JWT + RBAC Security Governance]
API --> Forensics[Digital Forensics Engine]
API --> GraphEngine[Knowledge Graph & PageRank]
API --> AskAIVIS[Ask AIVIS & 11 AI Experts]
API --> Operations[Workflow & ISO Compliance]
API --> Commercial[OEM Gateways & Predictive AI]
API --> DB[(PostgreSQL 15 Relational DB)]
API --> Redis[(Redis 7 Cache & Celery Queue)]
API --> Neo4j[(Neo4j 5 Graph DB)]
| Feature Module | Path | Key Capabilities |
|---|---|---|
| SOC Command Center | /dashboard |
Executive KPIs, fraud alerts, vehicle health scores, claims DataGrid |
| Ask AIVIS Assistant | /forensics/copilot |
Conversational AI chat, 11 Specialized AI Experts, Multi-Agent Conflict Resolution |
| Explainable AI (XAI) | /forensics/explainable-ai |
SHAP Waterfall & LIME feature risk attribution plots |
| OEM Telematics Gateway | /commercial/oem |
Tesla API, GM OnStar, FordPass, BMW ConnectedDrive, Geotab, Guidewire Bus |
| Predictive AI Engine | /commercial/predictive |
Fraud probability (94.2%), component failure risk, repair cost inflation |
| Developer Portal | /commercial/developer |
API key generator (aivis_live_sk_...), rate limits, webhook manager |
| SaaS Licensing | /commercial/licensing |
Multi-tenant tier quota meters (Enterprise Carrier, Fleet, Branch) |
| Executive Portfolio | /commercial/portfolio |
Risk heatmap, AI F1-Score (97.3%), Cloud-Native Kubernetes pod latency |
| Workflow Task Engine | /operations/workflow |
KanBan task board, forensic checklist enforcement, investigator matrix |
| SLA Breach Engine | /operations/sla |
Real-time countdown tickers, priority matrix, Chief Risk Officer escalations |
| Digital Signatures PKI | /operations/approvals |
RSA-2048 & ECDSA-P256 PKI digital signatures for verdict sign-offs |
| Compliance Audit | /operations/compliance |
ISO 27001, ISO 21434, NIST CSF, ISO 27037 Chain of Custody Markdown Exporter |
| Team Collaboration | /operations/collaboration |
Live comments feed with @mentions, case sharing, version audit timeline |
| Knowledge Graph | /intelligence/graph |
Interactive Neo4j SVG canvas with 15 entity node types & link inspector |
| Graph Algorithms | /intelligence/algorithms |
PageRank centrality rankings, Louvain community modularity clusters |
| VIN Cloning & Ghost Policies | /intelligence/vin-policy |
Cross-carrier registration audit, duplicate policy pings across state databases |
| Workshops & Surveyors | /intelligence/workshops |
Body shop estimate inflation ratios & compromised field surveyor override rates |
| Organized Fraud Syndicates | /intelligence/entities |
Shared identity fingerprinting (phone, IP, bank account), syndicate ring dossiers |
| Money Flow Analysis | /intelligence/money-flow |
Payout flow tracking from insurance carriers to body shops and kickbacks |
| Geospatial Fraud Heatmap | /intelligence/heatmap |
Spatial density analysis of coordinated collisions and body shop clusters |
| OBD-II Acquisition | /forensics/obd |
USB/Bluetooth/Wi-Fi adapter emulator, DTC codes, live sensor gauges |
| Sensor Intelligence | /forensics/sensors |
Multi-category sensor health scores, deviation %, anomaly scoring |
| CAN Bus Forensics | /forensics/canbus |
ASC/BLF/PCAP log parser, payload decoder, cyber attack vector detector |
| ECU Reflash Forensics | /forensics/ecu |
Firmware version, SHA-256 calibration hash, VIN matching, reflash history |
| EDR Black Box Crash | /forensics/edr |
CDR crash reconstruction, 5-sec pre-crash stream, peak G-force |
| GPS & Telematics | /forensics/gps |
GPX/JSON log parser, interactive route replay, harsh braking heatmap |
| Digital Evidence Locker | /forensics/evidence |
Cryptographic evidence locker with SHA-256, SHA-512, MD5 signatures |
| Document OCR | /forensics/ocr |
Policy/invoice OCR extraction, font splicing anomaly detection |
| AI Damage Vision | /forensics/damage |
Bounding box damage severity rating & repair cost estimator |
# Clone repository
git clone https://github.com/aivis/aivis.git
cd aivis
# Launch full-stack environment
docker compose up -d --buildAccess points:
- Frontend App: http://localhost:3000
- FastAPI REST API: http://localhost:8000/docs
- Neo4j Graph Browser: http://localhost:7474
# Frontend Setup
cd frontend
npm install
npm run dev
# Backend Setup (In a new terminal)
cd backend
python -m venv venv
source venv/bin/activate # On Windows: .\venv\Scripts\activate
pip install -r requirements.txt
uvicorn app.main:app --reloadComplete enterprise technical documentation is available in the docs/ directory:
- 📕 Complete PDF Documentation Manual
- Backend System Architecture & Technical Novelty
- Project Structure & Directory Tree
- System Architecture Blueprint
- Local Installation & Setup Guide
- Production Deployment Guide
- OpenAPI REST Reference
- Database Schema (Relational, Redis, Neo4j)
- Security & Compliance Architecture
- Vehicle Digital Forensics Engine
- Knowledge Graph Analytics Engine
- Ask AIVIS & Multi-Agent Engine
- Commercial Platform & OEM Integrations
Proprietary Enterprise Software. All Rights Reserved. ISO 27001 & ISO 21434 Verified.