Most AI writing prompts make your writing worse in a specific way: they sand off the voice that makes it yours. This pack is built on the opposite rule. Every prompt preserves your facts, your structure, and your voice, and tells the model so explicitly.
They are written for working writers who want AI as a collaborator, not a ghostwriter. Each prompt is a working tool. Copy it, fill the bracketed placeholders, and paste it into Claude or ChatGPT. Work through all eight and one piece goes from rough draft to finished copy without starting to sound like everyone else's.
One rule applies to every prompt in this pack. Your voice survives every step. Prompts 4 and 5 change the register or the reader; if you run either, run the voice match (prompt 2) again afterwards so the shift does not flatten the voice. The prompts edit and structure; the judgement about what the piece says stays with you. |
structure-from-mess voice-sample-match line-edit-light tone-calibrate audience-pivot cut-words opening-options closing-punch
| 01 / 08 The Outline Extractor structure-from-mess |
A messy draft usually contains a good piece in the wrong order. This shows you the structure you already have, and where it thins out, without touching a word.
You are a structural editor. Turn the rough draft below into a clean outline without rewriting it. |
Why it works: Rewriting before you can see the structure locks the mess in. An outline built only from what is on the page shows what to move, what to cut, and what is missing, while the words stay yours. |
| 02 / 08 The Voice Match voice-sample-match |
The centre of the pack. Paste 200 to 400 words of your own writing and the model rewrites the draft to sound like you, not like an AI.
You are rewriting the draft below in the voice of the writer whose sample is provided. |
Why it works: Voice lives in sentence rhythm and word choice, not in topic. Making the model name the two habits it copied forces it to study the sample instead of defaulting to its own house style. |
| 03 / 08 The Light Line Edit line-edit-light |
Most AI editing is rewriting in disguise. This one fixes clarity and leaves the voice alone.
You are a line editor. Edit the draft below for clarity. Do not rewrite it. |
Why it works: The before-and-after list keeps the edit honest. If a change cannot be quoted and defended, it was a rewrite, not a line edit, and you can reject it on sight. |
| 04 / 08 The Plain English Shift tone-calibrate |
For when the draft is accurate but only a specialist can read it. Same facts, plainer English, nothing simplified into being wrong.
You are translating a technical draft into plain English for a non-specialist reader. Accuracy comes first. |
Why it works: The replacement list shows exactly where precision was traded for clarity, so a subject expert can check the translation instead of re-reading the whole piece. |
| 05 / 08 The Audience Pivot audience-pivot |
A piece written for one reader rarely lands with another. This moves the framing to the new reader and keeps the argument where it was.
You are rewriting the draft below for a different reader. The argument stays; the framing moves. |
Why it works: Most audience rewrites fail by changing the argument along with the examples. Splitting the two, and flagging any example that would not convert, keeps the pivot honest. |
| 06 / 08 The 70% Cut cut-words |
Nearly every draft is carrying spare words. This cuts to 70% and shows its arithmetic.
You are cutting the draft below to 70% of its current length. The argument survives intact. |
Why it works: A named target beats 'make it shorter'. Stating the word counts makes the cut checkable, and the must-keep list stops the model deciding what matters on your behalf. |
| 07 / 08 The Three Openings opening-options |
First paragraphs fail more drafts than weak arguments do. This gives you three ways in: the claim, the detail, the reader's problem.
You are writing three different first paragraphs for the draft below. Same piece, three ways in. |
Why it works: Three openings on fixed routes force real alternatives rather than three versions of the same sentence. The 60-word cap keeps each one an opening, not a summary. |
| 08 / 08 The Closing Punch closing-punch |
Last lines are remembered longest and written most tired. This rewrites yours three ways, each paying off the piece's central claim.
You are rewriting the closing line of the draft below three ways. Each version lands harder than the current one. |
Why it works: A closing line works when it compresses the argument, not when it adds to it. Banning new claims keeps each option a payoff rather than a second ending. |
These eight prompts run as one workflow. The outline from the first prompt shapes the revision that goes into the voice match; the line edit tidies the result; the plain English shift and the audience pivot are branch steps for when the reader changes; the cut brings it to length; the openings and the closing line finish it. Use them in order for a single piece, or pull out the one you need today.
They all work on the same principle: the model edits and structures in the open, under constraints you set, and the voice it protects is yours. Used that way, AI sharpens your writing without flattening it.
One rule holds across all eight prompts: the facts, the structure, and the voice belong to the writer. If a prompt shifts tone or audience, run the voice match again before you call the piece finished.
| 8 PROMPTS | COPY, FILL, AND PASTE | ANY AI TOOL | PP10 |
|
Zymbos Intelligence Archive · Partner With Us · LinkedIn · Privacy Policy · Terms of Use You are receiving this because you subscribed at zymbos.ai © 2026 Zymbos Intelligence · John McGann · London, UK Zymbos Ltd · Company No. 16198848 · Teddington, England |
