Paperless-ngx is a self-hosted document management system that turns scanned paper into a searchable online archive, so a filing cabinet full of receipts, contracts and letters becomes something you can query by keyword. It is aimed at people who accumulate paper they later need to find again — households, freelancers, small offices — and who would rather keep that archive on their own server than upload it to someone else's. The project is the official successor to the original Paperless and Paperless-ng projects, and is run by a team of maintainers rather than a single author.
What it does
The core promise is simple: you feed documents in, and you get a library you can search. Paperless-ngx runs optical character recognition on every page it ingests, so the text inside a scan becomes indexed text, not just pixels. A blurry photo of a receipt stops being an image you have to remember the filename of and becomes a document you can find by typing a word that appears on it.
On top of the text index, the system helps with the filing work most people never do:
- It suggests tags, dates and document types for incoming documents instead of making you enter them by hand.
- It stores everything in a single archive with a web interface built for browsing and filtering large collections.
- It keeps the search fast enough to be useful when the archive runs to thousands of documents — the video shows exactly that: one keyword, one document.
- It is translated by the community through Crowdin, so the interface is not English-only.
The end state the project is selling is in its own name: less paper. You scan, you shred or file the original, and the copy that matters is the searchable one.
How it works
Paperless-ngx is a Python project built on Django, with an Angular front end — the GitHub topics list both, along with ocr, pdf, archiving and machine-learning. Documents arrive, OCR extracts their text, and the extracted text plus the metadata the system infers go into an index that the web UI searches against. The machine-learning side is what powers the suggestions for tags, correspondents and document types: the more you correct, the more the suggestions reflect how you actually file things.
Everything runs on hardware you control. There is no hosted tier described in the project's materials — you install it on your own server, and the documents stay there. The code is GPL-3.0, which means forks and modifications stay open too.
Getting started
The quickest way to judge it is the public demo at demo.paperless-ngx.com, which the maintainers run with a pre-filled archive so you can try searching before installing anything. That demo is the same interface the video shows.
Beyond that, the project maintains full documentation at docs.paperless-ngx.com covering installation and configuration, and a community chat on Matrix for questions. The repository itself carries continuous integration, coverage reporting and an active push history, so the install instructions in the documentation are the ones to follow rather than anything reproduced second-hand.
When to use it / when not
Use it if you have a scanner or a phone camera, a recurring pile of paper, and either a home server, a NAS or a small VPS you are comfortable maintaining. It fits the person who has been meaning to digitise years of tax paperwork and wants that archive to outlive any particular vendor. It also fits privacy-sensitive material — medical records, legal documents — precisely because nothing leaves your machine.
Skip it if you do not want to run and back up a server. A self-hosted archive is only as durable as your backup discipline, and an OCR archive of every document you own is a bad thing to lose or to expose. If your document volume is a dozen files a year, a folder with careful filenames will do the job with none of the upkeep.
Alternatives
Within the same family, the honest comparison is historical: the original Paperless and Paperless-ng came first, and Paperless-ngx exists because both needed a maintenance model that did not depend on one person. If you land on either of the older repositories, this is the project they point to. Outside the family, the alternatives are commercial cloud document services, which trade control for zero administration, and doing nothing — which is what most filing cabinets represent.
Take Paperless-ngx seriously if you own paper you will need to find later and you are willing to run one more service to make that possible. Forty-five thousand stars and a maintainer team make it about as safe a bet as self-hosted software gets, and the payoff is concrete rather than speculative: the difference between a drawer you dread and a search box that answers in a second.