Have Gemini preserve the original argument and qualifiers, cut jargon and filler, keep claims auditable, and end with the three ideas the reader should remember.
The prompt
Rewrite [SOURCE] for [READER]. Preserve the original argument and factual qualifiers. Use plain language, concrete examples, and short paragraphs. Remove repetition and unsupported intensifiers. Flag any claim that needs a source instead of inventing one. End with the three ideas the reader should remember.
Replace the variables
[SOURCE]The text to rewrite.
[READER]Who it is for — sets the reading level.
Which model to use
- Gemini 3.6 Flash — Rewriting is fast on Flash; upgrade to Pro for technical nuance.
Example input
- [SOURCE] = a dense research abstract. [READER] = "a busy product manager".
Example output
We only publish example outputs from real model runs. This template has not been formally tested yet, so no output is shown. Run it yourself with the input above.
Why this structure works
- Preserving qualifiers stops the rewrite from overclaiming.
- "Flag claims that need a source" prevents invented facts.
Common failures & fixes
It changes the meaning.
Fix: Add: "Do not add claims not present in the source."
Known limitations
- Always re-read for meaning drift on technical content.
FAQ
Will this make text undetectable as AI?
No. It improves clarity and removes filler, but we make no claim about detector evasion. Focus on whether the writing is accurate and useful.