B_TEA is an AI-powered financial analysis platform that transforms raw expense data into actionable insights. Upload your financial documents (CSV/PDF) and get professional-grade analysis, visualization, and forecasting—automatically.
Users struggle to understand their spending patterns without tedious manual analysis.
One-click upload → Comprehensive analytics dashboard with insights a financial analyst would provide.
- ✅ Better than Splitwise: Automatic parsing, real insights (vs manual logging)
- ✅ Better than Bank Apps: Privacy-first, smarter analysis, actionable recommendations
- ✅ Affordable: $5/month vs $500+ for real analyst
- Docker & Docker Compose
- Python 3.10+ (for local development)
- Node 18+ (for frontend development)
# Clone the repo
git clone https://github.com/cosmo95/B_TEA.git
cd B_TEA
# Start all services
docker-compose up --build
# Access:
# Frontend: http://localhost:3000
# Backend API: http://localhost:8000
# API Docs: http://localhost:8000/docsBackend Setup:
cd backend
python -m venv venv
source venv/bin/activate # On Windows: venv\Scripts\activate
pip install -r requirements.txt
python app/main.pyFrontend Setup:
cd frontend
npm install
npm start- 📤 File Upload: CSV & PDF parsing with auto-detection
- 🧹 Data Cleaning: Standardization, deduplication, currency handling
- 🤖 Auto-Categorization: ML-powered transaction categorization
- 📊 Analytics Dashboard: Beautiful charts, spending breakdown, trends
- 💡 Smart Insights: Subscriptions, anomalies, behavioral patterns
- 📈 Forecasting: Next month predictions with confidence intervals
- 💬 Recommendations: Actionable suggestions to optimize spending
- 📄 Export: PDF reports & cleaned CSV data
┌─────────────────┐
│ Frontend │
│ (React 18) │
└────────┬────────┘
│ HTTP/REST API
┌────────▼─────────────────────────────┐
│ FastAPI Backend │
│ ┌──────────────────────────────┐ │
│ │ Data Processing Pipeline: │ │
│ │ • Parser (CSV/PDF) │ │
│ │ • Cleaner │ │
│ │ • Categorizer (ML) │ │
│ │ • Analyzer (Insights) │ │
│ │ • Forecaster │ │
│ └──────────────────────────────┘ │
└────────┬──────────────────┬──────────┘
│ │
┌────▼───┐ ┌────▼──────┐
│ DB │ │ Redis │
│(PostgreSQL) │ (Caching) │
└────────┘ └───────────┘
Backend:
- FastAPI (API framework)
- Pandas & NumPy (data processing)
- Scikit-learn (ML categorization)
- pdfplumber (PDF parsing)
- SQLAlchemy (ORM)
- PostgreSQL (database)
- Redis (caching)
- Celery (async tasks)
Frontend:
- React 18 (UI framework)
- Recharts (data visualization)
- TailwindCSS (styling)
- Axios (API client)
- TypeScript (type safety)
B_TEA/
├── backend/
│ ├── app/
│ │ ├── main.py # FastAPI app initialization
│ │ ├── api/
│ │ │ ├── routes/ # API endpoints
│ │ │ └── models/ # Pydantic schemas
│ │ ├── database/ # SQLAlchemy models
│ │ └── utils/ # Config, logging, helpers
│ ├── data_pipeline/
│ │ ├── parser.py # CSV/PDF parsing
│ │ ├── cleaner.py # Data cleaning
│ │ ├── categorizer.py # Auto-categorization
│ │ └── analyzer.py # Analytics & insights
│ ├── requirements.txt
│ ├── Dockerfile
│ └── .env.example
├── frontend/
│ ├── public/
│ ├── src/
│ │ ├── components/ # React components
│ │ ├── pages/ # Page components
│ │ ├── services/ # API calls
│ │ ├── types/ # TypeScript types
│ │ └── App.tsx
│ ├── package.json
│ ├── Dockerfile
│ └── .env.example
├── docker-compose.yml
├── PRD.txt # Product Requirements Document
└── README.md
GET /health
→ Returns: { status: "healthy" }
POST /api/v1/analysis/upload
Body: { file: File }
→ Returns: { analysis_id: UUID, status: "processing" }
GET /api/v1/analysis/{analysis_id}
→ Returns: { transactions, metrics, status }
GET /api/v1/insights/{analysis_id}
→ Returns: { subscriptions, anomalies, trends, recommendations }
GET /api/v1/forecast/{analysis_id}
→ Returns: { predictions by category, confidence }
GET /api/v1/report/{analysis_id}
→ Returns: PDF file download
1. File Upload
↓
2. Validation & Parsing (CSV/PDF)
↓
3. Data Cleaning (standardization, deduplication)
↓
4. Auto-Categorization (ML model)
↓
5. Analytics (metrics, patterns, anomalies)
↓
6. Forecasting (next month predictions)
↓
7. Report Generation (PDF + JSON)
ML model trained on common transaction patterns automatically assigns categories:
- Food & Dining
- Transportation
- Bills & Utilities
- Entertainment
- Shopping
- Healthcare
- Education
- Personal Care
- Subscriptions
- Other
- Subscriptions: Detect recurring monthly charges (Netflix, Spotify, gym)
- Anomalies: Flag unusual spending spikes
- Trends: Show category growth/decline over time
- Behavioral: Weekend vs weekday patterns
- Recommendations: Actionable suggestions to save money
- Next month spending prediction
- Category-wise forecasts
- Confidence intervals
- Seasonal adjustment
- ✅ Encrypted file upload (AES-256 at rest)
- ✅ HTTPS for all communication
- ✅ Automatic file deletion (30 days)
- ✅ Rate limiting on API
- ✅ Optional user authentication
- ✅ GDPR-compliant data retention
- ✅ File parsing & cleaning
- ✅ Auto-categorization
- ✅ Analytics dashboard
- ✅ Insights generation
- ✅ PDF reports
- User accounts
- Save/manage analyses
- Mobile responsive
- Performance optimization
- Freemium launch
- Stripe integration
- API for Business tier
- Bank API integrations
- Budget planning
- Mobile apps
- 10k+ users
Contributions are welcome! Please:
- Fork the repository
- Create a feature branch (
git checkout -b feature/amazing-feature) - Commit changes (
git commit -m 'Add amazing feature') - Push to branch (
git push origin feature/amazing-feature) - Open a Pull Request
This project is licensed under the MIT License - see LICENSE file for details.
Cosmo - GitHub
- 📖 Read the PRD.txt for detailed product specifications
- 🐛 Report bugs in Issues
- 💬 Discuss ideas in Discussions
- 📧 Contact via email (if available)
Made with ❤️ for open source