AI Writing Pattern Check
Paste a draft and check it against a registry of formulaic, AI-associated writing patterns: delve-reliance, hedging padding, repeated sentence openers, and around forty more. This runs the real, live LanguageOps detector, not a mock. Nothing is rewritten and nothing is stored. The deterministic pass is free and keyless by design, which is the only part of LanguageOps exposed here.
Paste, then check. This calls the real LanguageOps detector, not a mock.
What this checks, and what it doesn’t
LanguageOps’ full registry has about 45 patterns. Roughly two-thirds of them are deterministic (exact-match, regex, or measurable structural signals like sentence length and repeated openers), and that’s what this page scores. The rest (sycophantic agreement, robotic empathy, conceptual vagueness, and similar) require actually judging meaning, not just matching text, so they need a model call. Those aren’t evaluated here; if any would apply, the result says so instead of guessing.
This isn’t an authorship detector. A clean score means nothing in the deterministic registry matched, not that a piece of writing is provably human.
$ POST languageops-web.vercel.app/api/lint
# deterministic pass only; refine is admin-gated