How it works

UnderByte makes one claim and means it literally: your file never leaves your device while the browser tool is doing the compression.

The privacy claim, made precise

The tool lives on /app, not on the marketing pages. That route uses a strict Content-Security-Policy that blocks third-party scripts and outbound requests. If you want proof, open DevTools, use the Network tab, drop a file, and watch what happens: nothing leaves the page.

How the compression works

For images, UnderByte repeatedly tries different quality levels until it finds the highest-quality result that still fits under your target. There is no AI, no guesswork, and no fuzzy percentage slider. It is deterministic math aimed at a hard byte ceiling.

The architecture

The site is static. Compression happens in your browser with same-origin scripts, WebAssembly codecs, and a worker thread so the page stays responsive. Files live as in-memory buffers while the job runs and are not uploaded to a server for processing.

PDF trade-offs

The browser PDF path is best for scanned documents and upload portals. If a PDF already contains real text, raster compression can make it larger or turn text into images. UnderByte warns before that happens and stops when the PDF is already efficient.

What the paid tier changes

Today Pass and Lifetime unlock more files and batch processing in the browser. They do not weaken the privacy model. The browser app still runs locally with the same CSP and the same same-origin assets. The pricing change is about workflow, not surveillance.