Paperless-GPT

A self-hosted AI add-on for paperless-ngx that organizes documents, links related records and performs OCR with local Ollama models or external services.

Paperless-GPT adds AI document processing to an existing paperless-ngx archive. It's for people who want to reduce manual sorting and extract text from scanned paperwork.

Document organization generates titles, tags, creation dates, correspondents and custom field values, with suggestions you can review and refine.
Related document links connects invoices, reminders, contract amendments and case letters through references found in their text; ambiguous matches remain unlinked.
Text extraction uses OpenAI or local Ollama models for image OCR, or dedicated services including self-hosted Docling Server, Google Document AI, Azure Document Intelligence and Mistral OCR.
Searchable PDFs adds selectable text while preserving scanned pages' appearance, then saves the enhanced files locally or returns them to paperless-ngx; this capability requires Google Document AI.

The project is free, open source under MIT and fully self-hostable, with Docker deployment. Ollama supports local processing, including reasoning models such as qwen3:8b. Choosing a cloud LLM or OCR service sends the document content needed for that task to the provider. Managed hosting through server.camp is available in Germany, Austria and Switzerland.

The web interface and API have no built-in authentication. Anyone who can reach them can modify connected documents, trigger processing jobs and change settings, so access protection must come from the hosting environment.