Audits a draft layer by layer (vocabulary, punctuation, sentence shape, openers, closers, grammar, narrative, per-model habits) against 13 studies, then rewrites what it flags in the author's voice.
