What This Tool Checks
This tool scans your writing for patterns that are statistically more common in AI-generated text than in human writing. It does not use machine learning or claim to definitively identify AI content. Instead, it highlights specific linguistic patterns so you can review and revise your writing to sound more natural and personal.
I built this after noticing that a lot of the content I was editing, including some of my own drafts, kept falling into the same patterns. Sentences that all started the same way. Paragraphs that followed an identical structure. Hedge words scattered everywhere. These patterns are not proof of AI use, but they are signs that writing could benefit from a human editing pass.
Patterns the Checker Identifies
The analyzer looks at several categories. Repetitive sentence openers flag when too many sentences begin with the same word or phrase, which is common when AI generates text that follows predictable templates. Filler and hedge phrases catch overused words like "it is important to note", "in today's world", and "it is worth mentioning" that add word count without adding meaning. These phrases appear in human writing too, but AI tends to rely on them more heavily and more consistently.
Passive voice frequency measures how often your text uses passive constructions. AI models default to passive voice more often than experienced human writers because passive phrasing sounds formal without requiring a specific perspective. The tool also checks vocabulary diversity by comparing the ratio of unique words to total words. AI-generated text often reuses the same vocabulary in ways that feel repetitive across paragraphs even when discussing different subtopics.
Sentence length uniformity measures whether your sentences vary in length or follow a monotonous rhythm. Natural human writing mixes short punchy sentences with longer complex ones. AI tends to produce sentences that cluster around a similar length, creating a metronomic quality that readers notice subconsciously even if they cannot articulate what feels off.
How to Use the Results
The overall score is not a verdict. A high score means the text exhibits several patterns common in AI writing and could benefit from revision. A low score means the text already reads naturally. Human writers sometimes score moderately on individual metrics, especially in formal or academic contexts where passive voice and consistent structure are appropriate.
The highlighted text view shows exactly which phrases and patterns triggered each flag. Use this to target specific revisions rather than rewriting everything. Often, changing sentence openers and removing a handful of filler phrases transforms a piece from sounding generic to sounding like it was written by someone with something to say.
Limitations
No heuristic tool can reliably determine whether text was written by AI or a human. This tool cannot and does not make that determination. Formal writing, academic papers, legal documents, and technical documentation naturally use patterns that overlap with AI writing patterns. The tool is designed to help you improve your writing, not to accuse anyone of using AI.
The analysis works best on English prose of 100 words or more. Very short texts do not provide enough data for meaningful pattern analysis. Code, poetry, dialogue, and non-English text will produce unreliable results.
Privacy
All analysis happens in your browser. Your text is never sent to a server, never stored, and never logged. Close the tab and the text is gone. There are no accounts, no usage limits, and no data collection.