Privacy, in plain language

Your files are converted inside your browser, on your own device, and are never uploaded to us. That is not a policy we promise to honor — it is how the tools are built, and you can watch it happen in your browser's network tab.

Two tools are different because you give them a link instead of a file: the YouTube transcript tool and the Website-to-Markdown tool. For those, the URL you paste goes to our server so it can fetch a caption track or a web page. That URL is the only thing that leaves your device, and we say so on the tools themselves.

We don't require an account to use the free tools, we don't store your files (we never receive them), and we don't sell or share anything. The rest of this page is the detail behind those sentences.

Every file-based tool on this site runs entirely in your browser:

• Audio and video (MP3, M4A, WAV, voice memos, MP4, WebM, MOV, MKV)
• Images (JPG, PNG, WebP, HEIC — photos, screenshots, scans)
• Documents (PDF, Word and other Office files, EPUB)

When you drop one of these in, it is handed to a converter running in a background worker on your own machine. The file's bytes are read there and turned into text there. No copy of your file is ever transmitted to us or to anyone else — not the contents, not the filename. Because there is no server in the loop, there is nothing on our side to leak, subpoena, or breach.

The one thing that does download is the model the tool needs — a small machine-learning file, fetched once from our storage and cached in your browser. That is a download to your device, not an upload from it. After the first run, the tool works offline. You can confirm all of this: open your browser's developer tools, switch to the Network tab, and convert a file — you'll see the model load, and you'll see your file go nowhere.

Three things travel off your device, and we want each one to be unsurprising.

1. A YouTube URL you paste. The YouTube transcript tool sends the video link to our server so we can fetch that video's existing caption track through an API and return the text. We fetch captions only — never the video or its audio, which we do not download, re-host, or store.

2. A website URL you paste. The Website-to-Markdown tool sends the page link to our server so it can fetch that page and convert it to Markdown for you.

3. Anonymous, aggregate page analytics. Like most sites, we count visits and tool runs so we know which tools people find useful. We use privacy-friendly analytics (processed for us by PostHog) with no cookies and no cross-site tracking — every visit is anonymous, we honor Do Not Track and Global Privacy Control browser signals, and nothing is tied to your files, which we never see. We also collect anonymized error reports when something crashes, so we can fix bugs; these contain technical diagnostics, not your content.

That's the complete list. If a tool isn't the YouTube or Website tool, nothing about your input leaves your device at all.

About your files: nothing. We can't store what we never receive. Local conversions leave no trace on our servers — the text stays in your browser until you copy or download it, and closing the tab is the only cleanup there is.

For the two URL tools, we keep a short cache — keyed to the content, not to you:

• YouTube transcripts are cached by the video's ID (and language) for a limited period, up to about 30 days. This is what makes a popular video load instantly on the next request and keeps the free tool affordable to run. The cache is keyed to the video, not to any person — there is no user attached to it, because there is no account.
• Fetched web pages are cached briefly (around a day) for the same reason: so a repeated request is fast and cheap.

To keep the free tools from being abused, our server also applies rate limits, which involves briefly and automatically handling technical request data (such as an IP address) purely to count requests. We don't build profiles from it. We don't store your transcripts against your identity, because we don't know your identity.

You don't need an account, an email address, or a sign-in to use any of the free, in-browser tools. There is no profile, no history saved on our side, and nothing to delete later — because there was nothing to create in the first place.

If we add paid options in the future for jobs a browser genuinely can't do (like cloud transcription of very long recordings), those will need an account and will be covered by their own clearly-labeled terms. They won't change how the free, local tools behave.

Questions or concerns about privacy? You can reach us using the contact link in the site footer. If you believe any part of this page doesn't match what the tools actually do, tell us — the whole point of building it this way is that the behavior is verifiable, and we want the words on this page to match the network tab exactly.