Audit-AI

AuditAI

AuditAI extracts structured compliance fields from unstructured clinical notes using an LLM, runs them through configurable rules, and returns a detailed audit report flagging billing and clinical compliance risks. Built for healthcare platforms that need automated documentation review.

Quick start

# Clone and run
git clone https://github.com/puneethkotha/Audit-AI.git
cd Audit-AI

# Set your Anthropic API key
export ANTHROPIC_API_KEY=your_key_here

# Start all services
docker-compose up --build

# Seed default compliance rules (first run)
docker-compose exec backend python seed_rules.py

• Frontend: http://localhost:3000
• Backend API: http://localhost:8000
• Metrics: http://localhost:8000/metrics

GitHub Pages

The frontend deploys to GitHub Pages on each push to main. Enable it in your repository: Settings > Pages > Source: GitHub Actions. The static site runs at https://puneethkotha.github.io/Audit-AI/. For full audit functionality, run the backend locally or deploy it separately and set VITE_API_URL when building.

Architecture

┌─────────────┐     ┌─────────────┐     ┌─────────────┐
│   React     │────▶│   FastAPI    │────▶│   Claude    │
│  Frontend   │     │   Backend    │     │    API      │
└─────────────┘     └──────┬───────┘     └─────────────┘
                          │
           ┌──────────────┼──────────────┐
           ▼              ▼              ▼
    ┌────────────┐  ┌──────────┐  ┌──────────┐
    │ PostgreSQL │  │  Redis   │  │Prometheus │
    └────────────┘  └──────────┘  └──────────┘

API endpoints

Method	Path	Description
POST	/audit	Run compliance audit on a clinical note
GET	/audit/{id}	Retrieve audit result by ID
GET	/rules	List all compliance rules
POST	/rules	Create a new rule
GET	/health	Health check
GET	/metrics	Prometheus metrics

Sample request

curl -X POST http://localhost:8000/audit \
  -H "Content-Type: application/json" \
  -d '{"note_text": "Patient is a 67-year-old male presenting with chest pain. History of hypertension. Medications: metformin, lisinopril, oxycodone PRN. Vitals: BP 158/92. Admitted for observation."}'

Tech stack

• Backend: Python, FastAPI, SQLAlchemy async, Anthropic Claude
• Frontend: React, TypeScript, Tailwind CSS
• Data: PostgreSQL, Redis (deduplication)
• Observability: Prometheus metrics

Project structure

auditai/
├── backend/           # FastAPI service
│   ├── main.py
│   ├── routers/       # audit, rules
│   ├── services/      # extractor, rule_engine, audit_service
│   ├── models/        # db, schemas
│   └── core/          # config, metrics, database
├── frontend/          # React app
└── docker-compose.yml

License

MIT