An integration fork of Paperless-AI – picking up where the original left off.
If you find this tool helpful, consider supporting its development:
Also consider supporting the original author:
Paperless-AI next is an AI-powered extension for Paperless-ngx that brings automatic document classification, smart tagging, and semantic search using OpenAI-compatible APIs and Ollama.
This fork collects pending upstream PRs, applies security patches, and tests improvements. Since the upstream project is no longer actively maintained, it has grown into the most up-to-date version available.
ℹ️ Upstream Credit – All original work belongs to clusterzx
This fork may not be fully compatible with the original Paperless-AI nor with previous Paperless-AI-patched versions.
⚠️ Important: Upgrading or migrating from the original project can lead to data loss (including metadata, history, or local app data) - please create a full backup before any upgrade or migration. A fresh install and document reprocessing may still be required for the best experience.
I use this fork for my own setup, but every deployment is at your own risk. Keep regular Paperless-ngx backups, especially before any upgrade or migration, as data loss can still occur.
I am actively improving the upgrade path and will provide detailed migration instructions in the documentation. 🙏
What makes this fork stand out:
-
🚀 Performance upgrades
- Server-side history pagination
- Centralized tag caching with configurable TTL
- Faster dashboard loading with async/lazy stats
-
🛡️ Security hardening
- Regular dependency updates with security patches
- Improved input validation and error handling
- Container image optimizations for smaller attack surface
-
🧠 Smarter OCR + AI fallback flow
- Mistral OCR queue for weak/failed text extraction
- AI-only re-analysis from stored OCR text (no re-run OCR required)
- OCR output preview/info per document
-
🧰 Advanced processing controls
- Include + ignore tags for selective automation
- Better dashboard status visibility (processed / OCR-needed / failed)
- Dedicated permanently-failed queue with manual reset workflow
-
🧪 Maintenance-focused development
- Integrated upstream PRs and community patches
- Expanded regression tests for critical workflows
- Active docs/changelog tracking for every integrated fix
See the complete fix list in the documentation:
📚 Included Fixes & Changelog →
Please check the docker variables here for all configuration options.
Lite version – AI tagging & OCR only (~500–700 MB):
services:
paperless-ai-next:
image: admonstrator/paperless-ai-next:latest-lite
container_name: paperless-ai-next
restart: unless-stopped
ports:
- "3000:3000"
volumes:
- data:/app/data
environment:
- PAPERLESS_API_URL=http://host.docker.internal:8000
- PAPERLESS_API_TOKEN=your_paperless_api_token
volumes:
data:Full version – AI tagging + RAG semantic search (~1.5–2 GB):
services:
paperless-ai-next:
image: admonstrator/paperless-ai-next:latest-full
container_name: paperless-ai-next
restart: unless-stopped
ports:
- "3000:3000"
volumes:
- data:/app/data
environment:
- PAPERLESS_API_URL=http://host.docker.internal:8000
- PAPERLESS_API_TOKEN=your_paperless_api_token
volumes:
data:Then open http://localhost:3000 to complete setup.
⚠️ First-time install: Restart the container after completing setup to build the RAG index (Full version only).
| Image Tag | Size | RAG |
|---|---|---|
admonstrator/paperless-ai-next:latest-lite |
~500–700 MB | ❌ |
admonstrator/paperless-ai-next:latest-full |
~1.5–2 GB | ✅ |
Docker Hub: admonstrator/paperless-ai-next
| 📖 Full documentation | paperless-ai-next.admon.me |
| 🐛 Report issues | GitHub Issues |
| 📜 License | MIT – original work by clusterzx |
Made with ❤️ by the community, for the community
⭐ If you find this useful, please star the repository!
Last updated: 2026-03-03