Check your essay for AI before you hand it in.
You get the sentences that look machine-written and the reason each one was flagged, not just a number at the top of the page. That way you can check the result yourself instead of taking it on trust.
1,000 free credits when you sign up. Works with ChatGPT, Gemini, Claude and any GPT model, on essays, papers, blogs, social posts and PDFs.
The industrial revolution reshaped British society in ways contemporaries struggled to name. It is important to note that this transformation was multifaceted and far-reaching in its implications. Wages in the northern mill towns rose unevenly, and the parish records from Oldham suggest the gains arrived a decade later than Manchester’s. Furthermore, it is worth considering that these developments fundamentally altered the social fabric of the era. Gaskell saw the mills up close and wrote them as neither triumph nor ruin.
1Formulaic hedging. “It is important to note that” appears in machine-written text far more than in student prose.
2Low burstiness. Both flagged sentences run 18–20 words with near-identical rhythm. Human paragraphs vary more.
3No specifics. The unflagged sentences name Oldham, Manchester and Gaskell. The flagged ones name nothing.
Every flag comes with a reason.
Most detectors give you a number and leave you to argue with it. If you’re going to rewrite a paragraph or talk to a student about theirs, you need to see what the result is based on.
A score you can place
One percentage for the document and a separate reading per paragraph, so one quoted block doesn’t drag the whole file down.
The actual sentences
Flagged sentences stay where they are in your text, so you read them in context instead of hunting through a summary.
Why each span was flagged
Each flag says what triggered it, in plain language: sentence rhythm, stock hedging phrases, wording that names nothing specific.
Anything you can open
Paste text, or upload .docx, .pdf, .txt and .html. Detection works the same way whatever the file is.
Built by four people who saw both sides of this.
Half of us tutored, half of us ran SEO. Both sides wanted something better than a guess.
Students
Check your own draft before you hand it in. If a paragraph reads as machine-written, you see it first and you decide what to do about it.
Teachers and professors
Go into the conversation with something specific to point at. The per-sentence reasons give you and the student the same thing to look at.
Marketers
Google’s spam policies target low-value, mass-produced content however it was made. Check a post before it goes live and gets indexed.
Three steps, about a minute.
-
Pick the detector that matches your document
Essays, papers, blog and SEO copy, resumes and PDFs each get their own model. An essay and a product page do not read alike, and one detector for both is how false flags happen.
-
Paste your text or upload the file
Drop in .docx, .pdf, .txt or .html, or paste straight into the box.
-
Read the marked-up report
Score, highlighted sentences and the reason for each one, laid out like the example at the top of this page.
Pay once. Your credits never expire.
Start with 1,000 free credits. After that you buy one-time packs. No subscription, no auto-renewal, nothing to cancel later.
You get these when you create an account. That’s 1,000 words through any detector, with the full marked-up report.
Need more than that:
One time only. Never auto-renews.
One time only. Never auto-renews.
1 credit = 1 word. Unused credits stay in your account. Refunds are covered by our Cancellation and Refund Policy.
Common questions.
How does AI detection actually work?
Language models pick each next word from a probability distribution, and they tend to pick likely words. That leaves statistical fingerprints in the finished text: low perplexity (few surprising word choices) and low burstiness (sentences that run to a similar length and rhythm, where human paragraphs vary more).
On top of those signals we look at phrasing patterns that show up far more often in generated text than in human drafts, like formulaic hedging and abstract summary sentences that name nothing concrete. Every flag on your report says which of these triggered it.
How accurate is it, honestly?
No AI detector is perfect, and any tool that tells you otherwise is selling you something. Detection is a probability judgement, not a fact. Short passages are hardest, because there is less signal to read. Heavily edited AI text and text written by someone with a very regular style are the two ways results go wrong in each direction.
This is the reason we show the reasoning behind every flag rather than just a percentage. You can look at what the detector keyed on and decide whether you agree with it. A number on its own gives you nothing to check.
Can I use a result to accuse a student?
No, and we would ask you not to. A detection score is a reason to open a conversation, not evidence to close one. Use the flagged sentences and the stated reasoning as something to talk through with the student, alongside what you already know about their work.
False positives happen to real students who wrote every word themselves. That risk is exactly why the per-sentence reasoning matters more than the headline percentage.
Which AI models does it detect?
ChatGPT, Gemini, Claude and any GPT-family model. The signals we read are properties of generated text in general rather than fingerprints of one vendor, so output from newer models is still picked up.
What can I upload?
Paste text straight in, or upload .docx, .pdf, .txt and .html. There are separate detectors for essays, papers, blog and SEO copy, resumes and PDFs. An essay and a product page don’t read alike, and running both through one model is how false flags happen.
How do credits work?
One credit is one word. You get 1,000 credits free when you create an account, which is a 1,000-word document through any detector with the full marked-up report.
After that, credits come in one-time packs: $14 for 20,000 or $24 for 40,000. They don’t auto-renew and unused credits never expire, so there’s nothing to cancel later.
What happens to the writing I upload?
How we handle and retain documents is set out in our Privacy Policy. If you want your data deleted, contact us using the form below and we will action it.
Try it on something you’ve written.
Sign up and run your first document through it. You get 1,000 credits free and it takes about half a minute.
Something not working? Tell us.
Feature requests too. We usually reply within the hour.