Clean Up AI Text

Frequently asked questions

Why does text pasted from ChatGPT look so strange?

AI tools write in Markdown: asterisks for bold, hashes for headings, dashes for lists. The interface renders it nicely — but when copied, the raw text comes along. Add to that characters you cannot see: non-breaking spaces, zero-width spaces, control characters for text direction and the markers ChatGPT uses to attach its sources. In e-mails, forms and CRM fields they stay put or trigger errors.

What is removed — and what stays?

Removed are formatting characters (bold, italic, headings, code, quote marks), table and HTML remnants, invisible characters, special spaces, double spaces and surplus blank lines; lines that only come from the line breaks of a PDF or e-mail become one paragraph again. The content stays: paragraphs, lists (as "- " and "1. "), numbers, accented letters and quotation marks. Links become "text (address)", emojis stay — both can be changed in the settings.

How do I recognise invisible characters?

You do not — that is the problem. A zero-width space is zero pixels wide, a non-breaking space looks like a normal one. It only becomes noticeable when a form rejects an e-mail address, the search in your CRM cannot find a name or a word in the subject line wraps strangely. The tool counts what it removed — so you can see whether your text was affected.

Does it also work with text from Word, PDF or websites?

Yes. The artefacts are the same: Word brings non-breaking spaces and soft hyphens, PDFs hard line breaks after every line and hyphenated words, websites HTML remnants and special characters. The tool is just not meant for program code — indentation would be gone afterwards. Code in Markdown code blocks (three backticks) is left unchanged, though.

What happens to my text?

Nothing. Cleaning happens entirely in your browser; the text is never sent to a server. So that an accidental reload deletes nothing, the browser remembers the text for this tab only — once you close it, it is gone.