Skip to content
All prompts
Artificial Intelligence

Teach AI Your Writing Style: Extract a Personal Style Card

Telling AI to "write like me" usually fails, because you can't fully describe what "like me" means, and AI can't guess it either. The result is always the same: generic text that sounds like nobody in particular. But your style is actually made of measurable parts: how long your sentences run, how often you ask questions, where you drop humor, which words you keep reaching for. These parts aren't a feeling, they're a pattern, and a pattern can be described. This prompt pulls that pattern out of your past writing and turns it into a "style card" you can paste at the top of every future AI request and say "write in this style." Not a one-time exercise, a permanent style signature.

Mehmet Kocabaş
Mehmet Kocabaşupdated: July 14, 2026
Prompt: copy it, fill in the fields, run it
You are a style analyst and writing coach. Your job is to extract the DNA of the user's personal writing style from the samples they give you: sentence rhythm, word choice, dose of humor, recurring patterns, opening and closing habits. Do this not with vague, generic adjectives, but with concrete patterns you can actually observe in the samples. Your output should not be a compliment, it should be a reusable style card. Remember: two different people can both say "I write short and clear," but one opens with questions and the other opens with confessions; your job is to catch that fine distinction.

Process the following writing samples:
- Sample 1: [paste here]
- Sample 2: [paste here]
- Sample 3: [paste here]
- Sample 4 (if available): [paste here]
- Sample 5 (if available): [paste here]
- Common platform for these samples (if any): [LinkedIn, blog, email, etc.]

DELIVER THE OUTPUT IN THIS EXACT ORDER, WITH THESE EXACT HEADERS:

=== SENTENCE RHYTHM ===
Observe the average length and variability of sentences (always short, does it swing long-short, does a long sentence get followed by a short punch line). Show it with quotes from the samples.

=== WORD AND PHRASE CHOICES ===
List specific words, connectors, or stock phrases that repeat across more than one sample. Give at least one quote from a sample for each. If you notice the kind of words the user avoids (say, never uses jargon, never drops in foreign terms), note that too.

=== OPENING AND CLOSING HABITS ===
Identify how the texts typically open (a question, a confession, an observation, a direct claim) and how they typically close (an invitation, a question, a summary, a trailing silence). State whether this pattern holds consistent across all samples or shifts by text.

=== HUMOR AND TONE DOSE ===
How often and how humor, irony, or warmth show up (rare-and-sharp, or frequent-and-light). State where the balance sits between seriousness and warmth.

=== STRUCTURE AND PARAGRAPH HABITS ===
Observe typical paragraph length (one-liners, a few sentences), whether bullet points get used, how transitions between ideas happen (direct, or bridged with connectors).

=== STYLE CARD (summary, reusable) ===
Turn the observations above into a short 6-8 point card; make it clear and copy-paste ready so it can be dropped at the top of a future AI request that says "write in this style."

=== MISSING DATA WARNING ===
If fewer than 3 samples were given, or if the samples are too similar to each other (all the same topic/format), state this clearly and say how many more samples are needed for a reliable card.

RULES:
- Don't invent linguistics terminology or unverifiable claims like "X% of writers use this style."
- Every observation must genuinely exist in at least one sample; don't make general claims without showing the example.
- Don't pad the style card with generic adjectives (impressive, professional); write concrete, observable patterns.
- Be short and practical; the card is not a long essay, it's a quick-read, copy-paste reference.

How to use it

  1. Gather at least 3, ideally 5-6 different writing samples (your LinkedIn posts, blog pieces, even a casual email). The more varied they are and the more they're genuinely written in your own hand, the sharper the resulting card.
  2. Run the prompt, read the resulting style card. Check whether each point actually fits you; if one feels off (say, it claims "you ask questions often" when you actually don't), note it and correct it on the next run with more samples.
  3. Save the style card you approve somewhere (a notes app, a doc). From now on, when you ask an AI to write on your behalf, paste this card at the top of the request and say "write according to this style card."

Example / tip

A consultant runs 5 of her last 6 months of LinkedIn posts through this prompt. The resulting style card shows: sentences run short (12-15 words on average), every post opens with an observation or a confession line, humor shows up rarely but lands sharp when it does, the words "actually" and "here's the thing" repeat often, closings are almost always a question or an invitation. She pastes this card at the top of her next request when having AI draft a blog post; the resulting draft no longer sounds like generic "content marketing" voice, it sounds a lot closer to her own posts.

Usage example

Copy the prompt as is, fill in the bracketed fields for your own business, and paste it into ChatGPT or Claude. The more concrete your input, the more useful the output. Take the first draft, ask for a one-sentence fix on whatever you don't like, then give it a final pass in your own voice.

When not to use it

Don't run this prompt on a single sample (say, just one email); extracting a pattern needs enough data, and a card built from one text mistakes that text's random quirks for your general style. You won't get a reliable style card from fewer than 3 distinct samples.

Output quality checklist

  • Is every point on the style card actually observable in at least one sample, or is it a generic guess?
  • Does the card describe you the way someone who actually knows you would, or is it padded with empty adjectives like "clear and warm"?
  • Does the next piece of text generated using this card actually sound more like you?

Newsletter

Once a week: the AI news that matters, and what it means for your business

No time to follow every headline? I pick what matters and sum it up with a “what it does for your business” note, in one email.

Permission-based list · no spam · one-click unsubscribe. We never share your email.