Skip to content

AI-powered invoice parser with editable review flow and CSV export. Built with Avalonia (Desktop) and Blazor (Web), powered by OpenAI + OCR.

License

Notifications You must be signed in to change notification settings

Nick4978/InvoiceParserAI-Solution

Repository files navigation

InvoiceParserAI

InvoiceParserAI is an AI-powered tool for extracting invoice data such as vendor name, date, and total amount from PDF documents. It supports both desktop (Avalonia) and web (Blazor) interfaces and uses OpenAI for parsing, with fallback to Tesseract OCR.


Features

  • Create folders like "income" or "expenses"
  • Bulk upload invoice PDFs
  • AI-driven extraction of:
    • Vendor Name
    • Invoice Date
    • Total Amount
  • Visual PDF viewer with field highlighting
  • Manual review and editing of extracted fields
  • CSV export of all reviewed data
  • Runs locally or via web app
  • Configurable OpenAI model and API key

Technologies

  • .NET 7/8
  • Avalonia (Desktop UI)
  • Blazor (Web UI)
  • OpenAI API (via OpenAI NuGet package)
  • PdfPig for PDF parsing
  • Tesseract OCR for fallback text extraction

Getting Started

Prerequisites

Running the Desktop App

cd InvoiceParserAI.Desktop
dotnet run

Running the Web App

cd InvoiceParserAI.Web
dotnet run

Configuration

Create or update appsettings.json in your project:

{
  "OpenAI": {
    "ApiKey": "your-api-key-here",
    "Model": "text-davinci-003"
  }
}

Folder Workflow

  1. Create a folder (e.g. "income") and upload invoice PDFs
  2. The system will mark each invoice as Processing, Review, or Complete
  3. You can edit extracted values manually
  4. After completing all, export the results to CSV

License

This project is licensed under the MIT License.


Author

Developed by Nick Posey and contributors.

About

AI-powered invoice parser with editable review flow and CSV export. Built with Avalonia (Desktop) and Blazor (Web), powered by OpenAI + OCR.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published