yoniapp/Deresegn-ai

bill manager in ethiopian context # ScanLogic: AI-Powered Receipt & Document Assistant ScanLogic is a production-ready, full-stack AI assistant designed to simplify expense tracking and document management. It leverages Google Gemini for intelligent OCR and data extraction, Supabase for secure data persistence, and Telegram as a lightweight, accessible interface. --- ## 🚀 How to Use ### 1. Telegram Bot Interface The primary way to interact with ScanLogic is through the Telegram bot. - **Register Receipts**: Simply send a photo of a receipt. The AI will automatically extract the merchant, total, currency, date, and category. - **Query Expenses**: Ask natural language questions like: - *"How much did I spend this week?"* - *"Show me the receipt from Kaldis Coffee."* - **Export Reports**: Use the `/export` command or the "Export Report" button to receive an Excel summary of your data. - **Fill Forms**: Upload a PDF form, and the bot will offer to fill it using your saved profile data. ### 2. Web Dashboard Access the web interface to view a visual summary of your spending, manage your profile, and see a history of all uploaded documents. --- ## 🛠 Tech Stack - **Frontend**: React 19, Vite, Tailwind CSS 4, Motion (Framer Motion). - **Backend**: Node.js (Express), TypeScript. - **Database**: Supabase (PostgreSQL) for structured data and Auth. - **AI Engine**: Google Gemini 1.5 Flash (`@google/genai`) for high-speed OCR and intent analysis. - **File Processing**: `pdf-lib` for PDF manipulation and `xlsx` for Excel report generation. - **Communication**: Telegram Bot API. --- ## 🏗 Project Structure ```text ├── server.ts # Main entry point: Express server & Telegram Webhook ├── src/ # React Frontend application │ ├── components/ # UI components (Tailwind + Lucide) │ ├── App.tsx # Main dashboard logic │ └── main.tsx # React entry point ├── .env.example # Template for environment variables ├── package.json # Dependencies and scrip …