📄 Invoice.AI (simXtract)

Serverless, AI-Powered Receipt & Invoice Extraction

Turn receipts into structured data instantly using Client-Side AI.

Features • Getting Started • Usage • Architecture

---

✨ Features

• 📱 Mobile-First Scanning — Optimized camera workflow with "Card View" for mobile devices.
• ⚡ Serverless Architecture — 100% Client-Side. No backend server required. Deploys to Azure Static Web Apps.
• 🤖 Azure OpenAI Integration — Leverages GPT-4o interaction for highly accurate data extraction.
• 🔒 Privacy Focused — Data stored locally in your browser (localStorage). API keys never leave your device (except to call Azure).
• 🎨 Material Design — Modern, clean Light Theme with intuitive interactions.
• 📊 Instant Analytics — Real-time tracking of expenses by category.
• 📤 Export Ready — One-click export to Excel (.xlsx) for accounting.

🚀 Getting Started

Since this is a static web application, you don't need to install Python or Docker.

Option 1: Run Locally

1. Clone the repository

   git clone https://github.com/taatim/simXtract.git
   cd simXtract

2. Open index.html

   • You can open index.html directly in your browser.
   • Recommended: Use a simple HTTP server (e.g., Live Server in VS Code) to avoid CORS issues with camera access.

   npx http-server .

Option 2: Deploy to Azure

1. Fork this repo to your GitHub.
2. Create a Static Web App in Azure Portal.
3. Link to GitHub and select this repo.
4. No Build Preset required (HTML/JS).

💡 Usage

1. Configure API

   • Click the Settings (⚙️) icon.
   • Enter your Azure OpenAI Endpoint and API Key.
   • These are saved locally in your browser.

2. Scan or Upload

   • Mobile: Tap "SCAN RECEIPT" to launch the optimized camera. Snap multiple photos. Tap "DONE".
   • Desktop: Drag & drop receipt images onto the dashboard.

3. Process Batch

   • Click "PROCESS BATCH". The AI will extract Vendor, Date, Total, and Category from all images.

4. Export

   • Use "EXPORT XLS" to download your data for expense reporting.

🏗️ Architecture

This project has evolved from a Python backend to a streamlined Static Web App:

simXtract/
├── index.html           # Main Application Entry
├── style.css            # Material Light Theme & Mobile Responsiveness
├── app.js               # Core Logic (State, Camera, UI Rendering)
├── azureOpenAI.js       # Client-side AI Extraction Logic
└── utils.js             # Helpers (Currency formatting, dates)

Note: Legacy Python/Docker files have been archived to legacy_v1/.

📄 License

This project is licensed under the MIT License - see the LICENSE file for details.

---

Built with ❤️ by taatim