Skip to content

README.md

Latest commit

 

History

History
177 lines (121 loc) · 6.08 KB

README.md

File metadata and controls

177 lines (121 loc) · 6.08 KB

DEEVO Decision Intelligence

Insurance Decision Operating System

Real-time decision infrastructure for insurance claims.

Version License Deploy

by Bader Alabddan

Live Demo

Claims Decision Control Room · Monitor · Command Center

What This Is

We built a real-time decision system for insurance claims.

It doesn't just show data.

It tells you:

  • What's happening — fraud signals, anomaly detection, risk scoring
  • Why it's happening — entity network analysis, decision drivers, evidence trail
  • What to do — AI-driven recommendations with confidence scores
  • What the financial impact is — approval cost, rejection saving, litigation risk

All decisions are auditable and explainable.

This reduces fraud loss, speeds up claims, and improves compliance.

This is not a dashboard. This is a Decision Operating System for insurance.

Claims Decision Control Room

The flagship interface. Route: /claims

What it does

  • Detects fraud signals in real-time
  • Provides AI-driven claim decisions
  • Shows explainable reasoning
  • Tracks full audit trail
  • Enables human override

Key Features

Feature Description
Decision Engine Approve / Reject / Escalate / Request Docs
Risk Scoring 8-factor weighted assessment with category breakdown
Anomaly Detection Statistical deviation alerts with severity classification
Entity Relationship Graph Network visualization — flagged entities, suspicious links
Decision Drivers Signal-to-decision bridge showing weighted impact factors
Financial Impact Approval cost, rejection saving, litigation risk, reinsurance recovery
Forensic Audit Timeline SHA-256 hashed, actor-typed event log (AI / SYSTEM / ANALYST)
Evidence Tracking Document verification status with completeness scoring

Layout

TOP:     Claim ID · Claimant · Policy · Amount · Risk Score · Status · SLA
LEFT:    Entity Graph + Claim Metadata + Evidence
CENTER:  Risk Summary + Decision Drivers + Time Series + Anomalies
RIGHT:   AI Recommendation + Confidence + Reasoning + Impact + Actions
BOTTOM:  Forensic Audit Timeline

Signal Intelligence Monitor

The macro intelligence layer. Route: /monitor

Five visualization engines for GCC-wide market intelligence:

Engine Function
GCC Map Interactive country risk map with trade corridors
Wave Simulation Causal wave propagation with animated particles
Cognitive Engine Multi-path reasoning — competing futures per scenario
Pre-Causal Engine Pressure fields detecting risk before signals form
Video Renderer Cinematic scenario video generation

Architecture

Signals → Graph → Rules → Simulation → Decision → Action → Audit

┌─────────────────────────────────────────────────────┐
│ 7. Governance        Audit trail, SHA-256, PDPL     │
├─────────────────────────────────────────────────────┤
│ 6. UI & Actions      Control Room, Decision Panels  │
├─────────────────────────────────────────────────────┤
│ 5. API               REST endpoints, WebSocket      │
├─────────────────────────────────────────────────────┤
│ 4. Agent Layer       AI recommendations, reasoning  │
├─────────────────────────────────────────────────────┤
│ 3. Decision Logic    Risk rules, anomaly detection   │
├─────────────────────────────────────────────────────┤
│ 2. Feature Engine    Signals, entity graphs, scoring │
├─────────────────────────────────────────────────────┤
│ 1. Data Ingestion    Claims, policies, fraud feeds   │
└─────────────────────────────────────────────────────┘

Stack

Layer Technology
Frontend React 18 + Vite + TypeScript + Tailwind CSS
Design System Graphite-gray 4-layer surface system (d-* tokens)
Visualization SVG entity graphs, sparkline charts, time series
Backend FastAPI + PostgreSQL + LangGraph
AI Ollama (local GPU), GPT-4 (cloud)
Deployment Vercel (frontend), Docker Compose (backend)

Run Locally

cd frontend
npm install
npm run dev

Opens at http://localhost:5173 — claims demo works without backend.

For full stack:

pip install -r requirements.txt
python run.py          # Backend on :8000
cd frontend && npm run dev  # Frontend on :5173

Deploy

Deploy with Vercel

Set VITE_API_BASE_URL to connect a backend. Leave empty for demo mode.

License

MIT — see LICENSE

Built by Bader Alabddan

Star this repo to follow development.