ream.

The four-second proof

Install once. Then the network is optional.

Offline mode is not a setting on this site. It is what happens when a PDF tool is a set of files rather than a service: once those files are on your device, there is nothing left to request. That also makes it the only privacy claim you can test by unplugging something.

Open a toolInstall steps

The test

  1. 1Open any tool page while online.
  2. 2Turn off Wi-Fi, or tick “Offline” in the network panel.
  3. 3Drop in a document and run it.
  4. 4It finishes. Watch the download appear.

Why this is not a gimmick

A page that processes documents on a server cannot pass this test, and there is no way to fake it: the request has to leave, and it cannot. Offline capability and the privacy guarantee are the same property observed from two directions.

Four platforms, four sets of steps

How to install it

Installing is optional. The tools work in a normal tab, offline included, once the page has been loaded. Installing adds an icon, a dedicated window and a cache that the browser is slower to evict.

Chrome or Edge, desktop

  1. 1Open any page on this site.
  2. 2Click the install icon at the right of the address bar, or open the browser menu and choose “Install page as app”.
  3. 3Launch it from your applications list like any other app.

Safari, macOS 14 or later

  1. 1Open the site in Safari.
  2. 2Choose File → Add to Dock.
  3. 3The site appears in the Dock and opens in its own window.

iPhone or iPad

  1. 1Open the site in Safari, not in an in-app browser.
  2. 2Tap the share button, then “Add to Home Screen”.
  3. 3Launch it from the home screen once so the cache fills.

Android

  1. 1Open the site in Chrome.
  2. 2Tap the three-dot menu and choose “Install app”, or “Add to Home screen”.
  3. 3Launch it from the app drawer or home screen.

Works with no network

Everything except the first load

  • Merge, split, extract, remove pages, organise, rotate and crop
  • Compress, repair and OCR — including the recognition itself
  • Word, PowerPoint, Excel and HTML into PDF
  • PDF to Word, PowerPoint, Excel, text, Markdown, PDF/A and images
  • Watermark, page numbers, forms, signatures, redaction and encryption
  • Compare two versions and summarise or translate with a local model endpoint

Needs a connection

Four honest exceptions

The first visit
The application has to arrive once, like any software. After that it is cached until an update replaces it.
An OCR language, the first time
The recognition model for each language is a separate download from the tessdata CDN. English is cached after the first use; a new language needs one fetch while you are online.
A hosted AI provider
If you point Summarise or Translate at a hosted endpoint, that call obviously needs a network. Point it at a model on your own machine and it does not.
Updates
A new version replaces the cached one on the next load while online. Offline, the version you have keeps working indefinitely.

What ends up on your disk

Cached code, never cached documents

The distinction matters. A cache holding application code is ordinary software installation. A cache holding documents would be a document store with a friendlier name. This site only ever does the first.

Application shell
HTML, JavaScript, CSS, fonts and the PDF.js worker
Cached once. This is what makes the tools open offline.
WebAssembly modules
Recognition and compression engines
Cached alongside the shell; a few megabytes, held only in the browser cache.
OCR language models
One per language you actually use
Fetched on demand from tessdata and cached. Unused languages are never downloaded.
Your documents
Nothing
No document is ever written to the cache, IndexedDB or localStorage. There is nothing to clear out afterwards.

Updates

A new version is picked up on the next load while online, in the background, and activated when you next open the app. Offline, the cached version keeps working for as long as it is cached. There is no forced reload mid-task.

Storage pressure

Browsers evict cached data when a device runs low on space. If the cache has been cleared, opening the app while online fills it again. Nothing of yours is lost in that process, because nothing of yours was in it.

To remove it: clear site data for this origin in your browser settings, or uninstall the app the same way you installed it.

On a phone

iOS Safari is the most aggressive platform about evicting cache, so install to the home screen if you rely on offline use. Memory is the tighter constraint: merge and split are comfortable, a 200 MB scan is not.

Where offline matters most

Four situations that are not hypothetical

A train, a tunnel, a plane

The work still has to happen. A cached toolset is the difference between finishing and waiting.

A hospital or a bank

Guest networks block large uploads, and some are monitored. Nothing to upload means nothing to monitor.

A client site

You cannot put a client document through a consumer web service on their network. A local tab is defensible.

A machine that is not yours

No install rights, no admin password, no software to request. A browser and a URL is enough.

A caveat worth stating

Offline proves the mechanism. It does not prove the code.

Running with the network off demonstrates that no document leaves the tab while it is processed. It does not audit the JavaScript that arrived on the first load. That is a separate question, answered by the content security policy, the lack of third-party scripts, and the short pinned dependency list — and it is why the security page spends most of its length on threats rather than strengths.

Security modelPrivacy policy
  • No document is cached, ever
  • No analytics to beacon while offline
  • No account check that could fail
  • No formal audit of the served bundle

Install it in ten seconds

Then turn the Wi-Fi off and use it.

Pick any tool, add the site to your dock or home screen, and process the file you have been meaning to deal with. The network was never part of the operation.

Compress a PDFChain two tools

no install requiredworks offline