Train AI to Write Like You

You’ve got a voice.

It shows up in how you talk to your audience, the way you phrase things in emails, and the personality that comes through when you’re not overthinking it. But the second you ask AI to write something for you, that voice vanishes.

What comes back reads like it was written by a committee of people who’ve never had a casual conversation in their lives. Stiff, generic, and nothing like the way you actually communicate. The problem isn’t AI. The problem is that you haven’t taught it how you sound.

Most people treat AI like a vending machine.

They put in a topic, hit the button, and accept whatever comes out. But AI is more like a session musician, it can play in any style, but only if you hand it the sheet music first.

Your voice is that sheet music. When you take the time to define your writing patterns, your word preferences, your rhythm, and your personality quirks, AI can produce content that sounds like you wrote it on a good day. Not a copy. Not a parody. Just you, but faster.

The prompts ahead are designed to help you capture your voice in a way AI can actually use.

You’re not going to describe your voice in vague terms like “friendly and professional.”

You’re going to break it down into specific, concrete patterns that AI can replicate consistently. Once you dial this in, every piece of content AI generates will feel like it came from your keyboard.

The fastest way to teach AI your voice is to show it examples of your best writing.

You already have content that sounds like you, emails that got great responses, blog posts that felt natural, social media captions that people engaged with.

That existing content is a goldmine for voice training. Instead of trying to describe your style from scratch, you let AI reverse-engineer it from the real thing. Grab three to five pieces of content you’ve written that you feel represent how you actually want to sound.

These should be pieces where you weren’t trying too hard, where the words just flowed. Paste them into AI with this prompt and let it do the analysis:

I’m going to share several samples of my writing. Analyse these samples and create a detailed voice profile for me. Identify: my typical sentence length and structure patterns, the types of words I gravitate toward (casual, technical, emotional, etc.), how I open paragraphs and transition between ideas, any recurring phrases, expressions, or speech patterns, my rhythm,  do I use short punchy sentences, longer flowing ones, or a mix, how I address my reader and what relationship I establish, what I tend to avoid (formality, jargon, certain phrases), and my overall personality as it comes through in the writing. Be very specific. I want to be able to hand this profile to someone and have them write convincingly in my voice. Here are my samples: [paste your writing samples]

What you get back is essentially a user manual for your own writing style. Some of it will be obvious, things you already knew about how you write. But AI will also pick up on patterns you never consciously noticed, like the fact that you always start with a problem statement, or that you tend to use three-part lists, or that your paragraphs follow a specific emotional arc. Those unconscious patterns are what make your voice yours, and now they’re documented.

A voice profile is great for understanding your style, but what you really need is a compact prompt you can paste at the top of any AI request to ensure the output matches your voice.

Think of it as a style preset. Every time you ask AI to write something, you include this prompt first, and the output comes back sounding like you instead of sounding like default AI.

Take the voice profile from the previous step and use this prompt to compress it into a reusable instruction set. The key is making it specific enough to be useful but short enough that you’ll actually use it every time:

Based on the voice profile you created from my writing samples, create a condensed voice instruction prompt that I can paste at the beginning of any AI writing request. It should be no longer than 150 words and should cover: tone and personality, sentence structure preferences, words or phrases to use and avoid, how to address the reader, paragraph style and flow, and any unique quirks that make my writing recognisable. Write it as direct instructions to AI, starting with “Write in the following voice:” Make it specific enough that the output will be noticeably different from default AI writing.

Save this voice prompt somewhere you can grab it fast, a pinned note, a text expander snippet, or the top of your favourite AI template. From this point forward, you start every content request by pasting this voice prompt first, then adding your specific topic or task.

The difference in output quality will be immediate. Instead of spending twenty minutes editing AI content to sound like you, you’ll spend five minutes making minor tweaks because the foundation is already in your voice.

Even with a solid voice prompt, AI will sometimes drift.

It might nail your casual tone in one paragraph and then slip into corporate speak in the next. Or it might overdo one of your patterns, like using your favourite transition phrase in every single paragraph. This is normal, and the fix is iterative feedback. You’re not starting over, you’re adjusting the dials.

When you get output that’s almost right but has specific issues, use this follow-up prompt to correct the drift without losing what’s working. Be as specific as possible about what needs to change:

The content you just wrote is close to my voice but has some issues. Here’s what needs adjustment: [describe exactly what feels off,  for example: “The opening paragraph sounds too formal,” “You’re using the word ‘leverage’ which I never use,” “The paragraphs are too short and choppy,” or “The tone shifts from casual to stiff in the second half”]. Rewrite the content with these corrections applied while keeping everything else that was working. Then update my voice instructions to prevent these issues in future requests.

That last part, asking AI to update your voice instructions, is how your voice prompt evolves over time. Each correction makes the prompt more accurate. After five or six rounds of refinement, your voice prompt will be dialed in to the point where AI consistently produces content that sounds like you with minimal editing. The upfront investment pays off massively because you’re building a permanent asset that makes every future piece of content faster and more on-brand.

Your voice should sound like you whether you’re writing a blog post, a sales email, a social media caption, or a product description.

But each of those formats has its own constraints.

A social post needs to be punchier. A sales email needs more emotional pull. A blog post has room to breathe. The test of a good voice prompt is whether it adapts to different formats while still sounding unmistakably like you.

Run this prompt to test your voice across multiple formats and see if it holds up. If certain formats sound off, you’ll know exactly where your voice prompt needs strengthening:

Using my voice instructions [paste your voice prompt], write the following five pieces about [pick a single topic in your niche]: a 3-paragraph blog post introduction, a promotional email for a product related to this topic, a casual social media post, a lead magnet description for an opt-in page, and a short bio blurb that references this topic. All five should sound unmistakably like the same person wrote them. Adjust the energy and format for each context but keep the core voice consistent throughout.

When you compare the five pieces side by side, you should be able to feel the same personality running through all of them. If one feels like a different person wrote it, that tells you where your voice prompt has gaps.

Maybe it handles long-form well but loses your personality in short-form. Maybe it nails casual content but stiffens up when writing anything sales-related. These are fixable problems, and each one you identify makes your voice prompt stronger.

In a world where everyone has access to the same AI tools, your voice is the thing that can’t be replicated by someone else typing the same topic into a prompt bar.

It’s your competitive advantage, your brand signature, and the reason people follow you instead of the thousands of other creators in your space.

Taking the time to train AI on your voice isn’t a nice-to-have. It’s what separates content that builds a loyal audience from content that gets scrolled past. Grab your best writing samples, run the first prompt, build your reusable voice instruction, and start testing it across everything you create.

Refine it every time something feels off.

Within a week, you’ll have a voice prompt that turns AI into a writing partner that actually sounds like you. And from that point on, every piece of content you produce will carry your personality, your perspective, and your unique way of connecting with the people you serve. That’s something no template or shortcut can replace.