Stop getting generic AI content that sounds like everyone else. Here's how to make ChatGPT write like you — and only you.
Here's the frustration nobody talks about honestly: you've been using ChatGPT to help with your writing, and the content it produces is technically fine. Grammatically correct. Logically structured. Completely soulless.
It doesn't sound like you. It sounds like every other piece of AI content flooding the internet right now — the same slightly formal tone, the same hedging phrases, the same predictable paragraph rhythm. You could have written it in your sleep, which kind of defeats the point.
The problem isn't ChatGPT. The problem is that you haven't told it who you are.
ChatGPT doesn't know that you write in short punchy sentences that hit like jabs. It doesn't know that you use humor to soften hard truths. It doesn't know that you never use words like "utilize" or "leverage," that you hate bullet points, that your intros always start with a situation rather than a statement, or that your audience is exhausted founders who want straight talk with no fluff.
Until you tell it all of that — precisely, in writing, in a structured way it can actually use — it will keep producing the default. The beige. The "sounds like AI" output that your readers see through immediately.
The solution is a mega-prompt. And this article is going to show you exactly how to build one from scratch.
What Is a Mega-Prompt?
A mega-prompt is a single, comprehensive prompt that you give ChatGPT before any writing task. It doesn't ask for anything to be written yet. Its only job is to install your voice, your rules, your style, and your preferences into the conversation — so that everything ChatGPT produces afterward sounds like it came from you.
Think of it as a writer's brief — the kind of document a brand gives to a ghostwriter before they start working. The ghostwriter doesn't guess what you want. They read the brief, absorb it, and write from inside your perspective rather than their own.
A well-built mega-prompt can contain your sentence rhythm preferences, your vocabulary rules, your structural habits, your tonal range, your audience definition, your forbidden phrases, and examples of your best writing. When ChatGPT has all of this at the start of a session, the output doesn't just improve — it transforms.
Here's how to build it.
Section One: The Voice and Tone Definition
This is the foundation. Before anything else, you need to give ChatGPT a clear, specific description of how you sound when you're writing at your best.
Vague instructions like "write in a friendly tone" or "keep it conversational" produce marginally better output than nothing. Specific instructions produce a dramatically different result.
Here's a template for the voice section of your mega-prompt:
My writing voice has the following characteristics:
— Tone: [e.g., direct and confident without being aggressive. Warm but not soft. I say what I mean without dressing it up.]
— Energy level: [e.g., medium-high. The writing moves. I don't let ideas sit too long before pushing forward.]
— Humor: [e.g., dry, occasional, never forced. I use a wry observation to break tension, not to entertain.]
— Register: [e.g., professional but not corporate. The way I'd talk to a smart colleague over coffee, not how I'd address a board meeting.]
— Personality markers: [e.g., I'm slightly impatient with fluff. I reward people who read carefully with a payoff line. I occasionally break the fourth wall.]
Fill in each of these with honest, specific descriptions of how you actually write — not how you wish you wrote, or how you think you should write. The more specific and true these descriptions are, the more useful they become as instructions.
-Section Two: Sentence and Paragraph Structure Rules
One of the clearest signals of a writer's voice is the rhythm of their sentences. ChatGPT defaults to a pattern that most people find recognizably AI: medium-length sentences with similar structure, organized into neat paragraphs of roughly equal length. It's readable. It's not distinctive.
Your sentence rhythm is your fingerprint. Here's how to describe it:
My sentence structure preferences:
— Sentence length: [e.g., I mix short and long deliberately. Short sentences land a point. Then I'll use a longer sentence to build context, add nuance, or create rhythm before dropping another short one.]
— Paragraph length: [e.g., I use single-sentence paragraphs when I want something to hit hard. Most paragraphs are 2-4 sentences. I never write a paragraph so long it becomes a wall of text.]
— Sentence openings: [e.g., I vary them constantly. I don't start three sentences in a row with "The" or "I." I use fragments intentionally. Starting with "And" or "But" is fine — it creates rhythm.]
— Transitions: [e.g., I don't use formal transitions like "Furthermore" or "In conclusion." I move between ideas with short bridges or just move directly.]
If you're not sure how to describe your sentence patterns, here's a useful exercise: take three paragraphs from something you've written that you're proud of, paste them into a document, and count the words in each sentence. Notice the pattern. That's your rhythm — now describe it.
Section Three: Vocabulary and Forbidden Words
Every good writer has words they love and words they never use. ChatGPT has its own default vocabulary that bleeds into everything it writes — words like "crucial," "delve," "leverage," "utilize," "it's worth noting," "in today's world," and "game-changing." These phrases are the AI fingerprints that readers now recognize immediately.
Your mega-prompt should include explicit vocabulary rules on both sides:
My vocabulary rules:
Words and phrases I use regularly and you should incorporate:
[List your actual words. If you say "real talk" or "here's the thing" or "full stop" or "the honest answer is" — write them down. If you use specific industry terms or slang that your audience uses — include them.]
Words and phrases I never use — do not include these under any circumstances:
— Utilize (I always say "use")
— Leverage (as a verb)
— Delve
— Crucial / paramount / pivotal
— It's worth noting that
— In today's fast-paced world
— Game-changing / groundbreaking / revolutionary
— Seamlessly
— Empower / empowering
[Add your own — everyone has a list]
Punctuation habits:
[e.g., I use em dashes — like this — when I want to add emphasis or a parenthetical thought. I use ellipses occasionally for effect... but sparingly. I don't overuse exclamation points.]
The forbidden words list is one of the highest-impact sections of any mega-prompt. It's worth spending real time on it, because those words will creep back in every single time you don't include them.
Section Four: Structural Habits
Beyond sentence rhythm, your writing probably has structural habits that define its shape — the way you open pieces, how you build to a point, how you close. These are learnable patterns that ChatGPT can replicate when you describe them explicitly.
My structural habits:
How I open:
[e.g., I almost never start with a thesis statement or a definition. I open with a scene, a situation, a specific moment, or a problem the reader is living right now. The first paragraph makes them feel seen before it tries to teach them anything.]
How I build:
[e.g., I usually set up a problem or tension in the first third, explore it honestly in the middle, and then resolve it — or at least reframe it — in the final section. I don't reveal everything at once.]
How I close:
[e.g., I don't summarize. I hate "In conclusion" energy. I usually close with one line that reframes everything that came before it — a reversal, a quiet landing, or a challenge to the reader.]
Formatting preferences:
[e.g., I use subheadings for long pieces but they should feel like chapter titles, not corporate headers. I rarely use bullet points unless I'm listing genuinely list-shaped information. No bold text in every paragraph — only for true emphasis.]
Section Five: Audience Definition
ChatGPT writes differently depending on who it thinks it's writing for. If you don't define your audience, it will default to a vague "general reader" — and that's exactly the tone that makes content feel like it was written for nobody.
My audience:
Who they are:
[e.g., Early-stage founders and solopreneurs, mostly in the 28-42 age range, who are smart, time-poor, and deeply allergic to content that wastes their time. They've read a lot of business content and have a finely tuned BS detector.]
What they already know:
[e.g., They don't need basic concepts explained. They understand how business works. They need frameworks, honest takes, and specific advice — not definitions and backgrounders.]
What they're trying to do:
[e.g., They're trying to build something that works without burning themselves out. They want clarity, not more information.]
What they hate:
[e.g., They hate being talked down to. They hate generic advice that applies to everyone and therefore helps no one. They hate content that promises a lot and delivers vague platitudes.]
When ChatGPT knows exactly who it's writing for, it makes different word choices, different structural choices, and different tonal choices. Audience definition is the part of the mega-prompt that most people skip — and it shows in the output.
Section Six: Voice Examples
This is the most powerful section and the one that takes the most effort. Give ChatGPT two or three examples of your writing at its best — paragraphs or short sections that you feel truly represent your voice.
Here are examples of my writing style. Study the voice, rhythm, structure, and word choices in these examples and apply them to everything you write for me:
[Example 1 — paste 2-3 paragraphs from a piece you're proud of]
[Example 2 — paste 2-3 paragraphs from a different type of piece — maybe something more conversational, or more technical, to show your range]
[Optional Example 3 — if you want to show how you write in a specific format — like a social caption, an email, or an opening hook — include one here]
Humans learn to mimic writing styles by reading examples. So does ChatGPT. The combination of explicit rules (sections one through five) and concrete examples (section six) is what makes the voice truly land.
Putting the Full Mega-Prompt Together
Here's how all six sections look assembled into a complete mega-prompt you paste at the start of a ChatGPT session before any writing task:
Before we write anything, I need you to internalize my writing voice completely. Everything you produce in this session should sound like it came from me — not from a generic AI assistant.
VOICE AND TONE
[your voice description]
SENTENCE AND PARAGRAPH STRUCTURE
[your rhythm rules]
VOCABULARY RULES
[your word preferences and forbidden phrases]
STRUCTURAL HABITS
[how you open, build, and close]
MY AUDIENCE
[who they are, what they know, what they want, what they hate]
EXAMPLES OF MY WRITING
[paste your examples]
Confirm that you have absorbed this by briefly describing my writing voice back to me in your own words before we start. Do not begin any writing task until I have confirmed your description is accurate.
That last instruction — asking ChatGPT to reflect back your voice before starting — is a quality control step that most people miss. It forces the model to demonstrate comprehension before it starts producing content. If it gets the description wrong, correct it before you ask for anything to be written. That correction becomes part of the instruction set for the rest of the session.
How to Use Your Mega-Prompt Going Forward
The mega-prompt isn't a one-and-done document. It's a living brief that you refine over time.
The first version you build will be a starting point. After a few sessions using it, you'll notice where it's working and where ChatGPT is still slipping into habits that aren't yours. Add corrections to the prompt. Tighten the forbidden words list when new offenders appear. Swap in better examples as you write pieces you're prouder of.
Save your mega-prompt in a place you can access quickly — a notes app, a pinned document, a dedicated file. At the start of any new ChatGPT session where you want it to write in your voice, paste the whole thing before your first content request. Each new conversation starts fresh, so the mega-prompt needs to be re-introduced each time.
Some writers keep several versions: one for long-form articles, one for social media captions, one for email newsletters. The core voice sections stay the same; the structural and format rules vary for each medium.
One More Thing: What a Mega-Prompt Can't Fix
The mega-prompt is a powerful tool. It is not a replacement for your thinking.
ChatGPT can replicate your voice when it has good instructions. What it can't replicate is the experience behind your opinions, the specific stories that prove your points, the hard-won perspective that makes your most resonant writing actually resonate. Those have to come from you.
The best use of a mega-prompt isn't "write this piece for me." It's "I have this idea, here's my thinking, write a draft in my voice that I can sharpen." The AI handles the production; you handle the insight. That division of labor is where the results actually get good.
Your voice is the one thing AI can approximate but never genuinely own. The mega-prompt makes the approximation close enough to be useful — and close enough to save you hours of editing generic output into something that finally sounds like you.
Have you built a voice prompt for ChatGPT or any other AI tool? Share what's working for you in the comments — or drop a before-and-after example if you've seen a dramatic difference in output quality.
0 Comments