Humanize

Overview

Humanize surgically removes AI fingerprints from text while keeping the author's message, structure, and intent intact. It works with both English and Russian texts and produces output that reads as if a real person wrote it from scratch — not a sanitized AI output.

Three-Pass Process

1. Pass 1: Vocabulary and formatting — replace all banned AI words (delve, tapestry, leverage, seamless, etc. in English; "следует отметить," "представляет собой," "на сегодняшний день" in Russian), fix em dash abuse, reduce excessive bold, remove chatbot artifacts.
2. Pass 2: Structure and rhythm — break uniform paragraph lengths, convert excessive bullet lists to flowing prose, vary sentence structure, fix synonym cycling, remove significance inflation and hollow intensifiers.
3. Pass 3: Anti-AI audit — re-read with fresh eyes. Ask for each paragraph: "Would a human editor flag this as AI-generated?" Fix residual patterns: text too clean, too balanced, transitions too smooth.

Pattern Categories

• Banned Vocabulary — 30+ English and 18+ Russian phrases replaced on sight
• Structural Patterns — bullet overuse, uniform paragraphs, formulaic structure, signposting, generic conclusions
• Formatting Artifacts — em dash abuse, curly quotes, excessive bold, emoji in non-casual text
• Communication Tells — chatbot artifacts ("I hope this helps!"), collaborative "we," sycophantic tone
• Linguistic Patterns — copula avoidance, synonym cycling, significance inflation, false concession, rule of three
• Russian-Specific — textbook tone, restating the same idea, "room temperature" text with no position, absurd metaphors

Key Rules

• Preserve meaning — every factual claim must survive intact
• Preserve voice — amplify the author's style, don't replace AI voice with your own
• No over-correction — not every em dash or bullet list is wrong
• Context-aware — technical docs can be more formal than blog posts