Detection Pipeline
Each module examines a different dimension of the writing. When they agree, you can trust the result.
Checks for AI-typical vocabulary, contraction avoidance, sentence uniformity, and eight other linguistic patterns that distinguish human and machine writing.
Measures perplexity, burstiness, word frequency distributions, and bigram predictability to spot the telltale smoothness of AI-generated text.
Detects invisible Unicode characters, zero-width spaces, homoglyphs, and other steganographic tricks used to disguise AI-generated content.
Uses Claude to evaluate whether the text exhibits the semantic patterns, reasoning style, and structural conventions characteristic of large language models.
Asks an AI to rewrite the text and measures how much it changes. Human writing changes substantially; AI writing stays almost the same.
Analyses cognitive complexity, emotional variance, and authorial voice consistency — the subtle fingerprints of a real human mind behind the words.
Who It's For
Whether you're proving your work is genuine or verifying someone else's, VEYL gives you the evidence.
Prove to publishers, agents, and readers that your manuscript is authentically human-written. Export an annotated report with evidence — a verification that your voice is truly yours.
Check student submissions before grading. VEYL gives you a detailed breakdown — not just a percentage — so you can have an informed conversation about academic integrity.
Screen incoming manuscripts and freelance submissions before they reach your desk. Catch AI-generated content early and protect your editorial standards.
Verify that freelance deliverables, blog posts, and marketing copy meet your standards for human-authored content. Integrate into your editorial workflow.