✍️ Writing & storytellingIntermediate

Humanize stiff AI text

Strip the 'AI tells' from any draft and rewrite in a real, breathing voice.

Use when: Editing AI-drafted blog posts, emails, or social copy.

Fill these placeholders

[PASTE_TEXT][AUDIENCE][VOICE_DESCRIPTION]

Replace each with your specifics — the more concrete, the better the model performs.

Want the whole thing?

Grab the full assembled prompt with section headers — paste it straight into ChatGPT or Claude.

Or copy block by block

🎭 Role

You are a senior editor who has trained 30+ writers to sound like themselves online. You can spot 'AI tells' from one paragraph.

📍 Context

Audience: [AUDIENCE]. The voice should feel like: [VOICE_DESCRIPTION] (e.g., 'a smart friend explaining over coffee — direct, dry humor, never condescending').

🎯 Task

Rewrite the text below to remove 8 AI tells: 1) symmetric tricolons ('not just X, but Y, and even Z') 2) 'In today's fast-paced world' style openers 3) hollow superlatives (powerful, robust, seamless, transformative) 4) hedge stacking ('it's important to note that, generally speaking') 5) generic 'remember' / 'imagine' framings 6) mirrored conclusions that restate the intro 7) em-dashes used as breath markers in every paragraph 8) the word 'delve' Then tighten by 20–30%.

🛡️ Constraints

Preserve every concrete fact and number. Do not invent quotes, stats, or sources. Keep the voice consistent — no slipping into corporate-speak halfway through.

📐 Output format

1. **Rewritten text** (clean, no markup). 2. **Tells removed** — bullet list of which of the 8 you actually found and fixed. 3. **Word count:** before → after.

Example

Text: [PASTE_TEXT]

Why this works

Pattern interrupt — fluency vs. authenticity tradeoff

AI text is *too* fluent. Reading it triggers low-effort scanning, which feels good but doesn't create memory. Slightly rougher human cadence creates micro-friction that the brain registers as 'real person speaking,' raising trust and recall.