Zettelrobbe

Zettelrobbe is an add-on for Paperless-ngx - not a document manager on its own. You keep running Paperless-ngx as your archive; Zettelrobbe sits alongside it and takes over the tedious part: sorting everything that comes in. (So you'll need a working Paperless-ngx instance first.)

Think of it as the friendly seal that dives through your paper chaos - the shoebox, the desk pile, the folder full of "Copy-of-scan_final_final2_reallyfinal.pdf". It's a community-driven fork of Paperless-AI, built to keep the project growing with regular updates and a real focus on what users ask for.

Once it's connected to your Paperless-ngx instance, you barely have to think about it. Every new document gets read, understood, and filed away with the right title, tags, document type, and correspondent - by the rules you set. Blurry scan or a photo of a crumpled receipt? It quietly runs those through an OCR queue first, so nothing slips through. OCR can be done by hosted AI (like Mistral) or by using local vision-capable LLMs like gemma4.

And you stay in charge: every change is logged, you can rescan a document whenever you like, and one click restores the original metadata if you don't agree. You pick the engine, too - from hosted models to a fully local setup with Ollama, plus Azure and any OpenAI-compatible endpoint.

Self-hosted, one Docker image, everything in a single local database. Your documents stay in your Paperless-ngx - and stay yours.

The name is German wordplay: Zettel (a note, a slip of paper) meets Robbe (a seal). A seal that hauls every lost document back to the surface - neatly labelled. 🦭🧾