When a document-level rewriting tool is the wrong tool

Do not use a document-level rewriting tool if your text is not in English, if what you have is a single stray paragraph, if the file has no formatting worth protecting, if you need to steer the tone, if you want a large free trial before committing, or if what you actually need is for something to be written for you. Each of those is a limitation published on our own site rather than a hypothetical.

These are our constraints, quoted from our own pages. That is not modesty as a marketing device; it is the only version of this article that is checkable. Every quotation below is on our site right now, and you can go and read it in the surrounding context, which is generally a page trying to persuade you of something.

A tool that publishes only what it is good at leaves you to discover the rest at the worst possible moment. So:

The bluntest one, and there is no clever workaround for it.

"Our rewrite engine is English-only today, and there is no way around it."

It appears twice, in two different places on the site, in almost the same words: "The rewrite engine is English-only."

If your document is in another language and you need its prose rewritten, this is the end of the road here.

"The cost is that we are slower and heavier than a text box and poor value for a stray paragraph."

There is also a minimum charge per job, which means a very short piece of text submitted on its own is poor value in the literal sense. Our own guidance is to put several short pieces into one file rather than run them individually.

The general form: if what you are handing in at the end is a passage rather than a file, the fastest route is the right one, and the fastest route is not a document upload. Our own summary of the choice is "If it is a passage, take the fastest route."

The whole argument for working on a file rather than a text box is that the file has something in it worth preserving — heading styles, citation fields, table structures, figure numbering. We have written about why those objects are fragile in what a plain-text round trip does to Word fields.

Turn that around and it becomes a reason not to bother. A plain essay with no styles, no fields, no reference list and no submission formatting requirements loses nothing by going through a text box. The overhead buys you protection you do not need.

Some tools let you pick a register. We do not, and say so:

"We lock the rewrite to academic register automatically; you cannot steer it toward casual or creative. If you need that flexibility, they serve you and we do not."

That sentence is on a page comparing us with something else, which is the context in which companies are least likely to say it. If your text is a blog post, a personal essay in a distinctive voice, or anything where the voice is the point, an automatic academic register is the wrong instrument.

"You want a large free trial before paying. We give 100 credits and that is it."

And in the same breath as several other limits: "We only serve the first case, and we are slower, heavier, English-only, and without a large free tier."

If your process is to test six tools on your whole thesis before choosing, we are the wrong shape for that process.

"There is no 'write this for me' entry point. That is not a gap in the roadmap; it is not something we intend to build."

Worth stating plainly because it is the request behind a good share of the traffic in this category. Rewriting works on expression; it has nothing to say about a document that does not exist yet.

"What we do not offer: a guarantee, a bundled detector, an extension, or any language but English."

The first item is the one that matters. There is no detected-or-refunded promise here, and the published reason is that "Detection systems update, and we will not promise a result we do not control."

Read that as a constraint on what you can expect from anyone in this category, not as a virtue of ours. Whatever a tool tells you about outcomes, the system doing the measuring is changing on a schedule nobody outside it knows.

Which leaves you with something better to do than shopping for a stronger promise:

• Go and ask what happens if the number does not move. Your department, your supervisor or the journal's editorial office can tell you what the actual procedure is when a submission comes back flagged again. That is a knowable fact with a person attached to it, and it is the one thing in this situation that no tool can be substituted for. If you only do one thing before buying anything, do this one.
• Then test on your own material, not on a demo. In our own words on the comparison pages: "any tool should be judged on your own material, ours included." A paragraph of your actual writing tells you more in five minutes than any published claim.

The remainder is narrow enough to state in one line: an English document, with formatting or citations worth keeping, that has come back from a detector with specific passages marked.

For everything else on the list above, the alternatives are real:

• A stray paragraph — a text box, or your own hands. Which edits actually change the statistics is a separate question, and we went through the manual version in humanizing AI text without a tool.
• A whole document flagged in only a few places — narrow the scope rather than reprocess everything. The argument is in rewriting only the flagged paragraphs.
• A tone problem — a human editor, or a tool built for register.
• Another language — not us, today.

Which is the whole list. If none of the six above applies to you, the situation this tool is built for is the one you are in — and even then, the test that decides it is a paragraph of your own writing rather than anything on this page.

Can it rewrite text in a language other than English? No. Our published position is that the rewrite engine is English-only and there is no way around it; other functions such as citation formatting and translation are separate and have different language coverage.

Is it worth uploading a single short paragraph? Generally not. There is a per-job minimum, and our own guidance is to combine several short pieces into one file rather than run them separately.

Can I choose a casual or creative tone? No. The rewrite is locked to an academic register, which is a limitation rather than a setting, and it is the wrong tool where voice is the point.

Is there a guarantee that a score will drop? No, and the published reason is that detection systems update and we will not promise a result we do not control. Treat outcome guarantees anywhere in this category with the same question in mind.