AI Humanizer
The Humanizer is a controlled transformation workflow. It asks the model to rewrite the passage, then checks the result against invariants intended to protect meaning, numbers, entities and language.
Humanization workflow
- Read the source
The pipeline detects language and extracts protected facts, numbers and structural cues.
- Rewrite
The selected provider produces a natural alternative with the requested tone.
- Validate
Invariant checks compare language, numbers, entities, links and length behavior.
- Return or stop
A rewrite that risks changing the source meaning can be rejected instead of silently replacing the text.
Modes and tones
| Control | Options | Effect |
|---|---|---|
| Mode | Advanced or Maximum | Selects the eligible workflow and credit multiplier |
| Tone | Preserve, conversational, professional, concise | Guides style without changing the requested meaning |
| Minimum input | 12 words | Provides enough context for a controlled rewrite |
Usage and privacy
Advanced uses one word credit per input word. Maximum uses three. Server-side humanization sends the submitted passage to the configured processing provider only to perform the requested rewrite.
Signed-in users can enable private text history. Without that opt-in, submitted content is retained only for the processing needed to return the result, subject to operational and security records described in the privacy policy.