How I built 15 PDF tools that never upload your files (100% in the browser)

Mitesh — Tue, 04 Aug 2026 19:43:55 +0000

A few months ago I needed to merge two bank statements and password-protect the result. Every "free online PDF tool" I found wanted me to upload those files to their servers first. For a document full of account numbers, that felt completely wrong — I have no idea what happens to a file after it's uploaded, how long it's kept, or who can see it.

So I built the thing I actually wanted: PDF WithMe, a set of PDF tools where your files never leave your device. Everything runs in your browser — no upload, no server, no account. Here's how it works and what I learned.

The core idea: no backend at all

The design flips the usual model. Instead of "send your file to my server, I process it, I send it back," the browser does 100% of the work. Your PDF is read, edited, and saved entirely on your own machine. Nothing is transmitted anywhere — and that single decision (no backend) shaped everything else in a good way.

The stack

Modern browsers are far more capable than people give them credit for. Almost everything I needed already existed as a JavaScript or WebAssembly library:

pdf-lib handles the structural work — merging, splitting, rotating, reordering and removing pages, page numbers, watermarks and signatures.
pdf.js (Mozilla's PDF engine) renders pages, which powers PDF-to-image conversion.
qpdf compiled to WebAssembly does real AES-256 encryption and decryption — that's what lets you password-protect a PDF, or remove a password from one you own, right on your device.
Tesseract.js (also WebAssembly) runs OCR to pull text out of scanned documents.
JSZip bundles multiple outputs (like split pages) into one download.

WebAssembly is the quiet hero. qpdf and Tesseract are mature C/C++ projects; compiling them to WASM means desktop-grade capabilities in a browser tab, with zero install and zero server.

A nice side effect: it works offline

Because there's no server to call, the app doesn't actually need the internet after it loads. A service worker precaches the app and its libraries, so it's a PWA — turn off your Wi-Fi, merge a PDF, and it just… works.

The trade-offs (it's not all free)

Memory is the limit. A server can stream a huge file; a browser tab holds it in RAM, so very large PDFs are constrained by the device.
Bundle size. WASM binaries aren't tiny, so I lazy-load each tool's dependencies only when you open that tool.
Some problems are genuinely hard client-side. PDF-to-Word is the best example — reconstructing editable layout is messy, so I ship two modes: a best-effort "editable" rebuild, and an "exact look" fallback that drops each page into Word as an image.

Why it was worth it

Beyond privacy — which is why I started — no backend has lovely properties: nothing to run so it costs almost nothing to keep online, no database of user files to breach or subpoena, and "your files never leave your device" isn't a slogan I have to defend, it's just how the architecture works. You can verify it by watching your network tab.

If you want to try it, it's free with no sign-up: pdfwithme.com. It's early, so I'd genuinely love feedback — on the tools, the client-side approach, or what to build next.