AI Text Rewriter — Humanize AI Content
- Match mode to source AI. Use Standard or Strong for Gemini text. Use Simplify or Elevate for Claude or ChatGPT. Use Natural for a light-touch rewrite.
- Elevate – raises vocabulary and formality; strong against detectors trained on casual writing.
- Drift – varies sentence rhythm unpredictably; often the hardest style for detectors to flag.
- Aggressive – locally typical sampling; avoids both the most predictable words and the odd long-tail ones, giving a more human distribution.
- Simplify – use on dense, jargon-heavy input to help the model rephrase clearly.
- Strong / Natural – produce distinctly different styles; if one is flagged, switch to the other.
- Shorts – tuned for short fragments (1–3 sentences); pair with the ↻ selection button to fix flagged sentences.
- Ghost-Pro – Fast · balanced · short texts. Reliable for essays and reports.
- Ghost-Ultra – Slower · stronger · default. Produces a distinctly different rewrite of the same text; great when the first result is still flagged.
- Use the Model switch (in the settings bar above the editor) to switch. Toggle between passes of the same text to compare results.
- What it does: Runs two full rewrites back-to-back. The output of Pass 1 becomes the input for Pass 2, so the text is transformed twice with different modes, compounding the style shift far beyond what a single rewrite can achieve.
- Why it works: Each pass strips away different AI-writing patterns. A single rewrite may dodge one class of detector signals; two passes with contrasting modes hit both vocabulary-level and rhythm-level signals at once.
- How to use it: Click Run Ultra (the Ultra card in the settings row), pick a mode for each pass in the popup that appears, then click Run. The two output panels appear side by side, Pass 1 on the left, Pass 2 on the right.
- Best combo: Elevate → Drift. Elevate first raises formality and vocabulary, then Drift reshuffles sentence rhythm, producing the biggest detectable style gap.
- Tip: Check AI detection on the Pass 2 result. If it's still flagged, try swapping the order (Drift → Elevate) or switching the model between passes.
- Try Retry before switching modes. The model is probabilistic, so 2–3 retries on the same mode often produce a noticeably different result.
- Use Edit for fine-tuning. If one sentence is off, click Edit and fix it directly rather than re-running everything.
- Citations are preserved automatically. The model never alters citations, statistics, names, or technical terms.