icereed/paperless-gpt

Use LLMs and LLM Vision (OCR) to handle paperless-ngx - Document Digitalization powered by AI

paperless‑gpt – AI‑enhanced document handling for paperless‑ngx

What it does

  • Works as a companion to the open‑source document manager paperless‑ngx.
  • Sends scanned images or PDFs to a Large Language Model (LLM) – OpenAI, Ollama, Anthropic, etc. – which extracts the text (LLM‑based OCR) and then uses the same model to generate a sensible title, tags, creation date, correspondent and custom fields.
  • Looks for reference numbers inside a document (invoice number, contract ID, case number) and, if an exact match exists in the paperless‑ngx archive, fills the Document Link field so the two records point to each other.
  • Can also call external OCR services (Google Document AI, Azure Document Intelligence, Docling Server) if you prefer a non‑LLM provider.
  • Produces searchable PDFs with a transparent text layer and can upload them back to paperless‑ngx as a new version of the same document.

Why it matters

  • Traditional OCR gives you raw text but no context. By using an LLM the system can understand the meaning of a scan, produce accurate titles and tags, and connect related files automatically – saving hours of manual sorting.
  • All processing happens on‑premise (Docker) or on a local LLM via Ollama, so you can keep sensitive documents inside your own network.

How to get it running

  1. Prerequisites – Docker installed, a running paperless‑ngx instance, and an LLM API key (OpenAI, Ollama, etc.) or an OCR service account.
  2. Docker‑Compose – The README provides a ready‑made docker-compose.yml that starts both paperless‑ngx and paperless‑gpt side‑by‑side. You only need to fill in environment variables such as PAPERLESS_BASE_URL, PAPERLESS_API_TOKEN, and the LLM credentials.
  3. Configuration – Most behaviour is controlled by environment variables:
    • LLM_PROVIDER / LLM_MODEL choose the model (e.g. gpt‑4o or qwen3:8b).
    • OCR_PROVIDER selects the OCR backend (llm, google_docai, azure, docling).
    • Tag names (MANUAL_TAG, AUTO_TAG, FAIL_TAG) let you track which documents have been processed or failed.
    • Options for PDF handling (PDF_UPLOAD_MODE, CREATE_LOCAL_PDF, etc.) let you keep the original file, replace it with a searchable version, or store PDFs locally.
  4. Web UI – Once the container is up, a simple UI lets you review AI suggestions, tweak prompts, and run ad‑hoc analyses on selected documents.
  5. Security note – The service itself has no authentication; you must place it behind a reverse‑proxy or VPN to prevent unauthorised access.

Who might use it

  • Small businesses or freelancers who receive many invoices, contracts, and correspondence and want automatic filing.
  • Teams that already use paperless‑ngx and need smarter OCR without moving data to a third‑party SaaS.
  • Users who prefer self‑hosted AI (Ollama) for privacy or cost reasons.

Where to find more

  • Full docs live in the docs/ folder (workflows, linking logic, managed‑hosting details).
  • The project is MIT‑licensed, Docker images are on Docker Hub and GHCR, and a managed‑hosting option for Germany/Austria/Switzerland is offered via server.camp.

All information above is taken directly from the repository’s README.

Related

  • Project
    clusterzx/paperless-ai

    Paperless‑AI is an open‑source extension for the self‑hosted document manager Paperless‑ngx. It automatically classifies new documents, adds AI‑generated titles/tags, and builds a vector index so you can ask natural‑language questions across your whole archive (RAG chat). It works with any OpenAI‑compatible model—including local Ollama servers—and ships as a Docker image with a simple web UI for configuration and manual processing.

  • Project
    drmingler/smart-llm-loader

    A Python package that converts documents and scanned images into clean, structured markdown chunks using OCR and LLM-powered semantic chunking for RAG systems.

  • Project
    eikek/docspell

    Docspell is a personal document management system that uses machine learning and NLP to automate the tagging and organization of scanned documents and emails.

  • Project
    Dicklesworthstone/llm_aided_ocr

    An advanced OCR pipeline that uses LLMs to correct errors in raw Tesseract OCR output and transform it into high-quality, formatted Markdown documents.

  • Project
    SaiAkhil066/CORTEX-AI-SUPER-RAG

    Cortex AI SUPER RAG is a local, open‑source RAG app that lets you upload PDFs/Word docs, ask natural‑language questions, and receive answers with precise citations. It combines BM25 keyword search, FAISS vector search, optional reranking, entity‑graph expansion, and several LLM‑based tricks—all runnable via Ollama without cloud APIs. The repo includes a Streamlit UI, a full evaluation harness (e.g., FinanceBench benchmark), and Docker support.