Your files stay here: PDF tools without uploads or cloud AI
A small manifesto about where your documents go: nowhere. Rebuild's free tools process files on your device — no uploads, no server copies, no cloud AI reading your pages.
The round trip most online PDF tools make
Most online PDF tools work the same way behind the scenes: your document is uploaded to a remote server, a program or an AI model processes it there, and the result is sent back down to you. The page you watch is a progress bar for a journey your file is making to a computer you will never see.
That round trip raises questions you should not have to think about when all you wanted was to rotate a page: who can read the file while it sits on that server, how long copies are kept, which country the server is in, and what happens to the extracted text afterwards. Reputable services publish deletion policies — but a policy is a promise about a copy that already exists.
- Upload: your file travels to a remote server.
- Process: software or a cloud AI model works on it there.
- Download: the result returns — and you trust the copies are deleted.
Rebuild skips the trip entirely
Rebuild's PDF, image and text tools run directly inside your browser, using your own device's processor and memory. When you choose a file, it is opened in the tab you are looking at — not sent to us. The work of merging, splitting, compressing, converting, editing or recognising text happens on your machine, and the result is saved from your machine.
This is not a special mode or a premium switch. It is how the free tools are built. The moment you close the tab, there is nothing left anywhere else to clean up — because nothing was sent anywhere in the first place.
- Open: the file loads in this tab.
- Process: your device does the work.
- Save: the result comes straight from your device.
Saying no to cloud AI — on purpose
It has become normal for AI features to be pushed into tools whether you asked for them or not — and for documents to be sent to cloud AI models as a side effect of ordinary work. We think a PDF tool should not make that decision for you. So Rebuild's free tools take a different position: your document is never sent to a cloud AI model. Not for OCR, not for compression, not for editing, and not quietly in the background.
Each task here uses software made for that task. That is not a limitation — it is why the results are predictable. A purpose-built converter does the same thing every time; it does not guess, improvise or invent. If we ever offer an optional feature that needs more than your device can do, it will be clearly labelled, off by default, and always your choice — never a default you have to find and switch off.
What never leaves your device
The precise promise is worth spelling out. When you use the tools on this site, we never receive your document, its filename, its contents, the text extracted from it or the results produced from it. There is no account requirement for the tools, no server-side queue, and no stored copy to request deletion of later.
Being precise also means saying what this promise is not. The website itself is still a website: your browser fetches pages and, if you consent, analytics and advertising make their own ordinary web requests. None of that traffic contains your documents. The promise is about your files — and it has no exceptions.
The honest trade-offs
Local processing means your device does the work, so speed depends on your hardware: a large scanned book takes longer on an old phone than on a new laptop. Very large files are limited by your browser's memory rather than by a data centre's.
We think that trade is worth it. There is no upload time, no queue behind other people's files, no failed transfer at 97 per cent, and no per-file cloud bill quietly shaping the product. Your device already bought the processing power — the tools simply use it. That is also part of why the core tools can stay free.
Credits and licences
Rebuild Free is built by one person — and it exists only because so many people far smarter than me developed this knowledge and shared it freely. Years of open-source work by developers and researchers around the world, on document formats, page rendering, text recognition and image processing, are what made a tool like this possible at all.
Those licences are respected and travel with the software. If you want the plain-language version of how the site handles data, the privacy policy covers it; if you spot something that could be clearer, the contact page reaches a person who reads it directly.