What's hidden in your document?
Let Harry, our tool, review your texts.
A Word file or a PDF carries more than just the text: who created it, with which program, how many minutes were spent on it, text that someone thought they had deleted, comments in the margins — and with a PDF also all previous versions, complete. And in the text itself there can be characters that you don't see.
Paste your text or choose your file. You'll see what's in it, what it means, and where it's all been looked at. You decide what comes out.
What Harry reviews
In the text. Characters you don't see but that are there: spaces with no width, control characters that reverse the reading direction, special spaces and line endings, and letters from Cyrillic, Greek or Armenian that look like ordinary letters. These come along when you copy your text from a chat window, a PDF or a website, and they stay in what you then send on.
In the file. A Word file carries the name of who created it and who last saved it, the company that was filled in in the program, how many minutes were spent on it, revision codes with which two documents can be linked to each other, text that someone thought they had deleted, and comments in the margins. A PDF carries that too, and more — see below.
What happens to your document
Nothing goes outside. Harry uses no AI and calls no other service. It is a program that reads and compares. Your file is read on our server in memory and goes nowhere else.
We do not retain your document. No file, no text, no copy. What is needed to give you the clean file back sits for a quarter hour in working memory and disappears afterwards by itself. With a PDF that is the file itself — a PDF is processed on the spot and not rebuilt from text, so without the original there is nothing to clean up. Nothing goes to disk and nothing into a database.
What we do count. Pasting text is unlimited and is not counted. For files, one per day without an account, and for that we must be able to count how many times you have had a file reviewed today. That happens with a random number in a cookie and with your internet address in encrypted form — enough to count with, not enough to trace anyone with. After seven days that too is gone.
You decide. Harry removes nothing without you checking it, and shows per finding why it doesn't belong there.
What's in a PDF
A PDF carries the same kind of data as a Word file, and usually more: besides the author and the timestamps also a document number that links every export of the same source file to each other, an edit history with timestamps, and sometimes a complete embedded file or the EXIF data of a photo — camera model, serial number, and with a phone photo the place where it was taken.
And the biggest point: a PDF does not overwrite, it pastes on top. A document that has been edited three times contains all three versions complete — deleted pages, earlier formulations, text that someone thought they had deleted. Harry rebuilds the file in one go and with that the whole history disappears. That is not bycatch but the most important cleanup there is.
What Harry in a PDF finds and does not remove: text under a black bar, text that is not drawn, text in the color of the paper, and content outside the crop. These are not traces in the wrapper but content of the page; removing them changes what the reader sees, and that is a decision of the writer and not of a cleanup program. You'll see what's there and on which page.
Two things you need to know
One cleaned Word file is rebuilt: only the text, the headings, the lists and the tables come back. That is precisely why there is no author name and no edit time in it — those were in the wrapper, and that wrapper is new. The formatting you had applied does not carry over.
With a pdf that works differently, and it must: fixed formatting is the entire reason it is a pdf. It is not rebuilt from text but edited in place — the pages remain exactly as they were, and what carries identity is removed. A pdf with a password Harry cannot check; remove the protection first.
And Harry promises no undiscoverability. He removes the author, the timestamps, the history and the attachments. But the way a file is structured still reveals which program it was made with — after cleaning that is ours. That is usually precisely what you want, but it is not the same as no trace at all.
Harry does not remove statistical watermarks, and does not promise to either. What is not there as a character in the text, he cannot remove. Anyone who promises otherwise is selling you something that does not exist.