ResearchPoC

Executable proof-of-concept for a geopolitical risk + supply-chain GenAI copilot. It uses synthetic WMS/TMS data to simulate a US trade block on China and recommends alternate sourcing.

What this demo includes

• Live Geopolitical News Scraping - Real-time news from UN, BBC, Reuters RSS feeds
• Severity Score Calculation - Mathematical formulation for shipment impact assessment
• Trade-block risk alerts for China-origin shipments with impact scores
• Alternate sourcing recommendations (Vietnam, Mexico, India, Malaysia)
• WMS/TMS Q&A for delays, supplier performance, and lead times (RAG over synthetic data)
• Enhanced Security - CSRF protection, input validation, security headers
• Dashboard UI with live news refresh capability

Quick start

python -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
python app.py

Then open http://localhost:8000.

New Features

1. Live Geopolitical News Scraping

Click the "Refresh Live News" button in the dashboard to fetch real-time geopolitical news from:

• UN News (Trade)
• BBC World News
• Reuters World News

News articles are automatically analyzed for geopolitical risk keywords and severity.

2. Shipment Impact Severity Score

Each risk alert now includes a calculated impact score (0-1) based on:

• Geopolitical Severity (40% weight)
• Delay Factor (30% weight)
• Supplier Performance (20% weight)
• Route Risk (10% weight)

See SEVERITY_SCORE.md for detailed mathematical formulation.

3. Security Improvements

• CSRF protection on all forms
• Input validation (max 500 chars)
• Security headers (X-Frame-Options, CSP, etc.)
• Debug mode disabled by default
• Secret key management via environment variables

See SECURITY_IMPROVEMENTS.md for complete security documentation.

Environment Variables

Create a .env file:

SECRET_KEY=<your-secret-key>
FLASK_DEBUG=False
LOCAL_LLM_BASE_URL=http://127.0.0.1:8085
LOCAL_LLM_MODEL=mistral
LOCAL_LLM_TIMEOUT=20

Generate a secret key:

python -c "import os; print(os.urandom(32).hex())"

Data sources

Synthetic data lives in data/ and is safe to modify for new scenarios.

Production Deployment

For production deployment:

1. Use a production WSGI server (gunicorn)
2. Set up HTTPS with nginx reverse proxy
3. Configure firewall and rate limiting
4. See SECURITY_IMPROVEMENTS.md for detailed checklist