Your file does not leave your device. Here is the whole path.

What happens when you open a file

1. 01

   You pick a file

   The browser hands the page a File object. That is a handle to bytes on your disk, not a copy and not an upload. Nothing has moved anywhere yet.

2. 02

   It is read into memory

   The page reads the bytes into an ArrayBuffer, which lives in the tab's memory. Close the tab and it is gone. There is no database here and no disk we control.

3. 03

   An engine runs on it

   JavaScript for most tools, WebAssembly for the two that need it. It runs on your processor, in the same sandbox as any other web page.

4. 04

   The result is built in memory

   A new Blob, still in the tab. For most tools the output never touches disk until you ask for it.

5. 05

   You save it

   The page creates a temporary blob: URL and clicks it for you, which is what your browser's download prompt responds to. The URL points at memory in your own tab and means nothing to anyone else.

There is no step in that list where your file crosses a network, because there is nowhere for it to go. filetity is a static site: HTML, CSS, JavaScript and a few WebAssembly files on a CDN. It has no server that can receive a file, no database, and no API.

Which network requests actually happen

Being precise matters more than being reassuring, so: requests do happen. They happen when the page loads, and they never carry your file.

While the page loads

• The page itself, its stylesheet and its JavaScript, from this domain.
• For a few tools, an engine compiled to WebAssembly, also from this domain. Measured over the wire: 347 KB for HEIC decoding, 381 KB for rendering PDF pages, and 9.2 MB for OCR. Each downloads once, on the page that needs it, and is then cached. Nobody who never opens the OCR page downloads a byte of the OCR model.
• Two analytics scripts, which count that a tool finished. They record the tool name, how long it took and the input size in bytes. They never see file contents or file names.

While your file is being processed

None. Not one. The work happens on your processor with the engine already in memory.

Check it yourself, in about thirty seconds

Do not take our word for it. The most convincing test needs no tools at all:

1. 1. Open any tool, for example Merge PDF, and let the page finish loading.
2. 2. Turn off your wifi, or switch on aeroplane mode.
3. 3. Now use the tool. It still works, and you still get your file.

A tool that uploaded your file could not do that. If you would rather watch the traffic directly, open your browser’s developer tools, go to the Network tab, and run a conversion: you will see no request carrying your file, because none is made.

What this does not mean

Local processing is a real privacy property and it is not a magic word, so here is the honest edge of it.

• The page is still code you are running. You are trusting that this site does what it says, the same as any web page. The difference is that it is checkable, and the aeroplane-mode test above is how you check it in a way that does not require trusting us.
• Your device is not automatically secure. If the machine is compromised, nothing a web page does changes that.
• We do count usage. Tool name, duration and input size, so we know which tools people finish. Not contents, not filenames, and no account, because there are no accounts.
• The day any tool here needs a server, that tool’s page will say so in the same breath, in the interface, not buried in a policy. None of them do today.