You wrote every sentence. Now you can prove it. VEYL runs six independent detection modules across your manuscript and gives you a detailed report that shows readers, publishers, and platforms your writing is authentically human.
15,000 words free every month. No credit card required.
How It Works
Upload your document, let our pipeline analyse it, and get a detailed annotated report in seconds.
Drag and drop any document — .docx, .pdf, .rtf, .txt, and more. Your files are processed securely and never stored permanently.
Our pipeline runs heuristic, statistical, hidden character, semantic, rewrite, and psycholinguistic analysis across your entire document.
Receive an overall AI style signal, a breakdown by module, and an annotated copy of your document with every flagged section explained.
Understanding Your Score
VEYL returns an AI style signal from 0–100%. Here's how to read it.
The prose shows no meaningful machine-writing signal in the style. This is what you want to see — your manuscript reads as authentically human.
Contemporary, voice-driven human writing and AI-assisted writing both land here — VEYL can’t reliably tell them apart on style alone. Weigh the flagged sections and the Provenance & Forensics panel, which reports hard evidence.
A strong machine-writing signal in the style. This points to AI involvement — text an AI helped write or edit — but is not proof of authorship. Confirm against the report’s flagged sections before relying on it.
This is a style signal, not proof. It measures how machine-like the writing reads — it is not a probability of AI authorship, and it can’t tell AI-written prose from prose an AI merely helped edit. Non-native English, highly technical writing, formulaic prose, and even punctuation conventions like heavy em-dash use can all affect it. In our own testing on human manuscripts held out from training, VEYL flagged roughly 4%. Always treat your VEYL result as one data point — never as proof.