Human writing mixes short sentences with long ones. Machine drafts tend toward a uniform middle length. Higher variation reads more naturally.
Sentence lengths: 20, 8, 12, 14
Long average sentences make prose heavy regardless of who wrote it. Around 15-20 words reads comfortably; past 25 the reader starts working.
A small set of words appears far more often in generated text than in human writing. Their density is the single most visible tell.
delve · leveraging · robust · seamless · crucial · myriad · facilitate · landscape · underscores · embark · holistic · comprehensive
Phrases like "it is important to note that" add words without adding meaning. Cutting them shortens the text and sharpens it.
it is important to note · In today's fast-paced world
Models punctuate with em dashes several times more often than people typing on a keyboard, where the character is awkward to reach.
Sentences that begin Moreover, Furthermore, Additionally are connective throat-clearing. A few are fine; a pattern of them is a tell.
Moreover, leveraging robust frameworks is crucial for success. · Furthermore, this seamless approach will facilitate numerous outcomes … · Additionally, organisations should embark on a holistic transformation…

This tool measures the writing patterns that make prose read as machine-generated, and reports each one with the evidence behind it. It does not output a percentage chance that a model wrote your text, because that number cannot be produced honestly and the harm from producing it anyway is real.
Here is the problem with AI detectors. They are statistical classifiers guessing at authorship from surface features, and they are unreliable enough that OpenAI withdrew its own for low accuracy. Their false positives fall hardest on people writing in a second language, whose prose is measurably more regular and therefore looks more machine-like to a classifier. A confident-looking "87% AI-generated" gets used to accuse a student or reject a freelancer, and there is nothing underneath it that can carry that weight. Any tool that shows you such a number is inviting you to make a consequential decision on a coin flip.
What can be measured honestly is the text itself. Sentence-length variation, usually called burstiness, is the strongest single signal: human writing mixes a three-word sentence with a thirty-word one, while machine drafts settle into a uniform middle length. Vocabulary density counts how often the text reaches for the small set of words models overuse. Hedge density counts phrases like "it is important to note that" which add length without meaning. Em dash rate counts a character models produce several times more often than someone typing. Connective openers count sentences that begin with Moreover or Furthermore.
Each of those is a fact about the writing rather than a claim about the writer, and each is reported with the matches that produced it, so you can check the measurement rather than trust it. That makes the result useful for the thing it is actually good for: editing. Low variation and heavy hedging make prose worse regardless of who wrote it, and both are fixable.
Use it to diagnose your own drafts, to see which parts of a document read as formulaic, and to understand what the patterns are. Do not use it, and do not let anyone use it, to decide whether a person is telling the truth about their own work. Everything runs in your browser; nothing is uploaded, logged or stored.
Any prose - a draft, an article, a submission. It stays in your browser and is never uploaded.
Six measurements, each with its value, a plain-English explanation, and the matches that produced it.
Every signal lists what it actually matched. If a count looks wrong, the evidence tells you why.
Low variation and heavy hedging make writing worse whoever wrote it. Send the text to the humanizer, or edit by hand.
Reported as a coefficient of variation, so it does not depend on absolute length. The strongest single signal in the set.
Counts around 40 words models reach for constantly, normalised per 1000 words so short and long texts compare.
Phrases that add length without meaning, and sentences that begin with connective throat-clearing, both listed with matches.
Each signal shows the actual matches or the raw sentence lengths, so a measurement can be checked rather than believed.
Rewrite the habits that make prose read as machine-written - inflated vocabulary, hedge phrases, em dashes, and connective filler.
Detect and remove invisible Unicode characters from AI text - zero-width joiners, variation selectors, bidi controls, and tag characters.
Generate random numbers in any range, with unique-only and decimal options.
Count words, letters, sentences, paragraphs, and estimate reading time
0 comments
No comments yet. Be the first to share your thoughts!