Why your PDF never leaves your browser
Most PDF sites upload your file, process it somewhere, and send it back. We went the other way. Here is why, and what it took.
By the RatioAI team
Let's start with a slightly uncomfortable question. When you drop a bank statement into a free online PDF compressor, where does that file actually go? For most sites, the honest answer is: onto someone else's server, for some amount of time, under some retention policy you probably didn't read.
That's not necessarily malicious. It's just the easiest way to build a PDF tool. You put a powerful library on a server, accept uploads, run the job, and return the result. It works. But it means every single file passes through infrastructure you don't control.
When we started building RatioAI, we asked ourselves a simple thing: what if the server just… wasn't involved?
The browser is a lot more capable than people think
A modern browser is basically a small operating system. It can run compiled code at near-native speed through WebAssembly, spin up background threads with Web Workers, and store gigabytes of data locally with IndexedDB. Ten years ago you couldn't build a serious PDF editor in a tab. Today you can.
So that is what we did. When you compress, merge, split, rotate, watermark, sign or password-protect a PDF in RatioAI, the work happens on your own machine. The file is read from your disk into the tab, processed there, and saved back to your disk. There is no upload step, because there is nothing to upload to.
"Trust us" isn't a privacy policy
A lot of services promise to delete your files after an hour. Maybe they do. But you can't check. Our approach flips that around: instead of asking you to trust what happens on our servers, we just don't put your document on them in the first place.
You can actually verify this yourself. Open your browser's developer tools, switch to the Network tab, and run any browser tool. You'll see the app's code and fonts load, but you won't see your PDF going anywhere.
So why do we have a server at all?
Fair question. Some jobs genuinely need heavier machinery. Converting a Word document to PDF faithfully needs an office suite. High-quality OCR needs big language models. Very large files can push past what a phone browser comfortably handles.
For those, we're building optional server tools as part of our upcoming paid plans. The important word there is optional. Every tool in the workspace is labelled with where it runs, and nothing silently switches to the server behind your back.
- Browser tools: the default, free, no upload.
- Server tools: opt-in, clearly badged, outputs auto-deleted.
- No hidden fallback from one to the other.
The trade-offs we accepted
Building this way is harder. We can't just throw a bigger server at a slow job; we have to make it fast on your laptop or phone. We have to care a lot about memory, because a browser tab doesn't have unlimited RAM. And we can't fix a bug on the server and have it apply instantly to everyone's files.
We think that's a fair price. The result is a PDF workspace you can use for contracts, ID documents, medical records and financial statements without wondering where they ended up.