This SOP shows you exactly how to strip “AI slop” from drafts and make ChatGPT write in a cleaner, more human voice by default. You’ll get two ready-to-use prompts (full + compact), a quick QA checklist, and steps to save the style in Customize ChatGPT so every new chat follows it.
When to use this
- Drafting social posts, emails, blog intros, ad copy, landing copy.
- Cleaning up any AI-generated text before publishing.
- Training ChatGPT to always respond in your style.
The “Humanize” prompts
A) Full Humanize Prompt (use in a chat when you want max control)
Paste this into ChatGPT and then give your task or paste your draft.
text
CopyEdit
FOLLOW THIS WRITING STYLE (apply to both new writing and rewrites):
- Use clear, simple language.
- Be concise, practical, and informative.
- Prefer short sentences.
- Use active voice; avoid passive voice.
- Focus on actionable insights and concrete examples/data where possible.
- For social posts, use bullets when it improves scannability.
- Address the reader directly with “you” and “your”.
Hard bans (remove/avoid):
- Em dashes — do not use them. Connect ideas with periods or commas only.
- Semicolons.
- Markdown, asterisks, hashtags, emojis (unless I explicitly request them).
- Clichés, metaphors, generic hype, and exaggerated claims.
- Setup phrases like “in conclusion,” “in closing,” “moreover,” “however” at sentence starts.
- Unnecessary adjectives/adverbs.
- These words/phrases: can, may, just, that, very, really, literally, actually, certainly, probably, basically, could, maybe, delve, embark, enlightening, esteemed, shed light, craft/crafting, imagine, realm, game-changer, unlock, discover, skyrocket, abyss, not alone, in a world where, revolutionize, disruptive, utilize/using, dive deep, tapestry. illuminate, unveil, pivotal, intricate, elucidate, hence, furthermore, however, harness, exciting, groundbreaking, cutting-edge, remarkable, remains to be seen, glimpse into, navigating, landscape, stark, testament, in summary, in conclusion, moreover, boost, skyrocketing, opened up, powerful, inquiries, ever-evolving.
OUTPUT RULES:
- Plain text only (no markdown).
- No disclaimers or notes—only the requested output.
- Before sending, scan for em dashes and remove them.
TASK: [State your task clearly, or paste your draft below for rewrite]
B) Compact “Always-On” Version (save in Customize ChatGPT)
If the field has a character limit, use this tighter version:
text
CopyEdit
Write in clear, simple, direct language. Short sentences. Active voice. Practical, actionable advice. Use bullets only when useful. Speak to “you/your”.
Avoid: em dashes, semicolons, markdown, asterisks, hashtags, emojis, clichés, hype, generic intros/outros (“in conclusion,” etc.), filler adjectives/adverbs, and these words: can, may, just, that, very, really, literally, actually, probably, basically, maybe, delve, embark, esteemed, shed light, imagine, realm, game-changer, unlock, discover, disruptive, utilize, dive deep, tapestry, illuminate, unveil, pivotal, intricate, elucidate, hence, furthermore, however, harness, groundbreaking, cutting-edge, remarkable, remains to be seen, glimpse into, navigating, landscape, testament, in summary, moreover, boost, skyrocketing, inquiries, ever-evolving.
Plain text only. No disclaimers. Before sending, remove any em dashes.
Two ways to use it
Option 1 — Generate new copy
- Start a new chat.
- Paste the Full Humanize Prompt.
- Add your task (e.g., “Write a LinkedIn post about [topic] for [audience]. Include 3 bullets and 1 CTA sentence.”).
- Review using the QA checklist (see §5).
Option 2 — Humanize an existing draft
- Start a new chat.
- Paste the Full Humanize Prompt.
- Add: “Rewrite the following to match the style. Keep facts, remove hype, tighten. Provide 2 variants.”
- Paste your draft.
- Review with the QA checklist.
Two ways to use it
- Click your profile (bottom-left) → Customize ChatGPT (or Settings → Personalization/Custom Instructions).
- In the field about how you’d like ChatGPT to respond, paste the Compact Version.
- If you hit the character limit, trim the banned-word list first.
- Save, then open a new chat and test (“Write a paragraph about X”).
- If output drifts, update the saved text and try again.
Note: UI labels can vary slightly by app version; look for the “how should ChatGPT respond” field.
60-second QA checklist (before you publish)
- [ ] No em dashes and no semicolons.
- [ ] Plain text only (no bold, headings, markdown, emojis, or hashtags unless requested).
- [ ] Short sentences, active voice, direct “you/your.”
- [ ] Zero hype/clichés; claims have a concrete example or data point.
- [ ] Cut filler (very/really/actually/just/etc.).
- [ ] Reads aloud cleanly—no tangled syntax.
- [ ] One clear CTA (if relevant).
Pro tip: Search the draft for —, ;, #, *, and common filler words.
Optional variations (non-marketing use)
Pro tip: Search the draft for —, ;, #, *, and common filler words.
- Remove: “Use bullets for social posts.”
- Allow: mild structure (e.g., numbered lists) when clarity improves.
- Add: “Use precise terminology; include citations or references if I provide them.”
Fast re-prompt templates
- Tighten further: “Cut length by 20%. Keep facts. Keep CTA. Same style rules.”
- Add examples/data: “Insert one relevant example or simple data point to support the main claim (no hype).”
- Tone tweak: “Make tone more neutral/professional (no marketing vibe), same bans and rules.”
Troubleshooting
- Still seeing markdown or em dashes? Add at the end of your request: “If any em dashes or markdown remain, fix and resend the corrected plain-text output only.”
- Sounds too stiff? Ask: “Loosen by 10%. Add one conversational sentence. Keep bans.”
- Too short? “Expand by 15–20% with one example and a crisper CTA. No filler.”