Your mail stays with you.

Privacy

Your documents are never uploaded

When you open a PDF, your browser reads the pages, recognizes the text, suggests where each letter starts, and builds the files you download. All of that happens on your own computer. Your PDFs, page images, and recognized text are not sent to this website or to anyone else.

What your browser stores

To let you close the tab and pick up where you left off, the tool saves your review progress and page-reading results in your browser’s own storage (IndexedDB). To continue, choose the same PDF again. They are stored only in this browser, on this device. Use Forget saved work on this device on the tool’s start screen to remove them; removing a PDF from the list on the left also forgets that PDF. Clearing your browser’s site data for this website also removes them.

Files you download (the separate PDFs and the CSV index) are yours; clearing saved data does not touch them. Removing data from browser storage is not a secure-erasure guarantee.

Sending feedback

The Send feedback button in the tool sends only what you choose, and only to this website:

• the message you type;
• your email address, only if you enter one, so we can reply;
• the Accuracy report for the PDF on screen, only if you tick the box. It holds counts (such as pages, documents, and corrections), never document text, names, dates, file names, or images. The form lets you read it before you send;
• your browser and system in general terms (for example “Chrome 141 · Windows”), to help reproduce problems.

Your PDFs are never sent. Feedback is stored in a database run by our hosting provider, Cloudflare, and is read only by the site’s administrator. It is kept until the administrator deletes it; to have yours removed, send another message asking for that.

The administrator sometimes uses an AI assistant to help sort feedback. When that happens, the feedback the assistant reads is sent to the company that runs it (for example Anthropic, for Claude) to be processed under that company’s terms. Each assistant gets its own access key, which the administrator can withdraw at any time. A key can read feedback and, only if the administrator allows it, mark messages as read or resolved and add a private note. No key given to an assistant can delete feedback.

To limit spam, the site counts messages from each network per day. It keeps a scrambled code instead of your IP address, and deletes those counts the next day.

What your browser downloads

On your first visit the browser downloads the app and its page-reading engine from this website: about 10 MB for clean scans, and up to about 40 MB when pages are hard to read and a second, more careful reader is needed. Your browser keeps a copy, so later visits start quickly. These files are the same for everyone; downloading them tells this site nothing about your documents.

Anonymous usage counts

To learn whether the tool is useful, the site counts visits and a few actions. It does this without cookies and without building a profile of you.

What the page sends to this website:

• which page you opened (home, tool, or this page) and, the first time you click, type, or touch it, that you did;
• how many PDFs you chose, whether each one finished reading and how many pages it had, and how many files you exported;
• where the visit came from, as a short label: the other website’s name (such as “reddit.com”) or a link tag (such as “newsletter”), never the full address;
• whether your browser says it is being controlled by automation software, so robots are not counted as people.

Never sent: file names, document text, page images, anything you type, or full web addresses.

To count each visitor once a day, the server mixes your IP address and browser name with a random key that changes every day, keeps only the scrambled result, and deletes that day’s key and scrambled codes shortly after the day ends (days are counted in UTC). The result cannot be turned back into your IP address, and visits on different days cannot be linked. What remains is totals only, such as “120 visitors, 30 from reddit.com, 25 finished reading a PDF”, including the approximate country (from Cloudflare) and the type of device (computer, phone, or tablet).

If your browser sends Global Privacy Control (a privacy setting in some browsers and extensions), nothing is counted.

What the site does not do

• No cookies.
• No advertising or third-party analytics, and no tracking across websites.
• No accounts, sign-in, or payments.
• No third-party fonts, scripts, or embedded content. The page is only allowed to load files from this website.

Like any website, the hosting provider may keep standard request logs (such as IP address and which files were requested). Those logs never include your documents.

Limitations

Letter Splitter gives you suggestions to review, not guarantees. Always check the splits, names, and dates before you export.

• English only. Text recognition is tuned for English.
• Typed text. Printed and typed letters work best. Handwriting is read poorly.
• Scan quality matters. Faded, low-contrast, or skewed scans need more review.
• Page orientation. Upside-down pages are detected and turned the right way up, but check them.
• Blank pages (such as blank backs of letters) are detected and left out by default. The tool shows how many it found, and you can choose to include them.
• Very large files. How much the tool can handle depends on your device’s memory. Split packets of more than about 500 pages into smaller files.
• Missing dates can default to the processing date. Check them before you export.

Supported browsers

Use a current version of Chrome, Edge, Firefox, or Safari on a computer.

• In Chrome and Edge you can save the finished files straight into a folder.
• In Firefox and Safari the finished files download as a single ZIP.