TaxHacker

TaxHacker is a self-hosted accounting application aimed at freelancers, independent builders, and small businesses who want to automate the tedious parts of expense tracking. You take a photo of a receipt or upload an invoice PDF, and the app uses an AI model to read it and pull out the relevant details: the amount, date, vendor name, tax figures, and individual line items. All of that gets saved into a structured table you can filter and export.

The AI part is flexible. You can connect it to OpenAI, Google Gemini, or Mistral, or point it at a locally running model if you prefer to keep everything on your own hardware. You can also write your own prompts to extract specific fields that matter to your situation, such as project codes, addresses, or custom categories that match how your business is organized.

Currency handling is built in. The app detects the currency in any document and converts it to your chosen base currency using historical exchange rates from the actual transaction date. It supports over 170 currencies and 14 cryptocurrencies.

The whole application runs in Docker, so deployment comes down to downloading a single compose file and running one command. It ships with a PostgreSQL database included in that setup. Because you host it yourself, your financial documents stay on your own infrastructure and never pass through a third-party service.

Export works by generating CSV files with the transaction data and attached document files bundled together, which makes handing off records to an accountant or tax advisor straightforward. The project is open source under the MIT license and described as early in development, so some rough edges should be expected.