Support
Straight answers about how the tools handle your files and where browser-only processing runs out.
No. Each tool reads your file in the page and writes the result back to your device. There's no upload endpoint behind this site. To confirm it, open your browser's network panel while you use a tool: you'll see the page load its own scripts, and no request carrying your file.
Tool inputs are kept in your own browser's local storage so a refresh doesn't lose your work. Nothing is transmitted to us, and we can't read it. On a shared machine, use the Clear saved data button on the privacy page before you walk away.
There's no account system at all, so there's nothing to sign up for, nothing to log in to and no email address to hand over.
Advertising, and only if you accept advertising cookies. Ads are kept off the homepage, the blog and every policy page.
The tool extracts the text and its line structure and writes a valid .docx. It doesn't reconstruct fonts, images, tables, columns or positioning, because doing that faithfully needs a full document rendering engine. You get editable words, not a visual copy.
No. A scan is a picture of text with no character data behind it, and there's no OCR here. If the tool finds nothing to extract, it tells you instead of handing back an empty file.
Compression re-renders each page as an image, which is what makes the file smaller. Links, form fields and searchable text don't survive that. On a text PDF that's already optimised it can even make the file bigger. When that happens, the tool tells you and offers the original back.
It was withdrawn. It never worked, and there's no honest way to render slide layouts in a browser without a rendering engine we can't ship. Exporting to PDF from PowerPoint itself gives a better result than anything we could have built.
PNG, JPEG and WebP everywhere, plus BMP through a small encoder written for the site. AVIF depends on your browser. If yours can't encode it, the tool tells you it has fallen back instead of silently handing you a different format.
It matches a colour you pick and removes pixels close to it. That works on a plain, evenly lit backdrop and fails against a cluttered or shadowed one. It's a colour key, not subject detection, and the page says so up front.
No. It decodes the header and payload so you can read what a token claims. Only a service holding the signing key can tell you whether the token is genuine.
Version 4 draws from the browser's cryptographic random source, so yes for cases where unpredictability matters. Version 7 encodes a timestamp by design, so it leaks creation time.
Each one names the published method it implements (Mifflin-St Jeor, Naegele's rule, reducing-balance amortisation), so you can check the result against the source. The assumptions behind every figure are listed under the result.
Tell us, and include the values you entered. Corrections jump the queue ahead of everything else. If a formula really is wrong, we fix the code; we don't quietly adjust the number.
No, and the site doesn't claim otherwise. The formulas come from named published sources, but nobody with medical, legal or financial credentials has reviewed the content.