How to Make ChatGPT Sound Like You and Actually Understand You

Brooke Silva

Aug 27

9 min read

Laptop displaying ChatGPT surrounded by sticky notes about writing style, preferences, memory and work habits, representing how to personalize ChatGPT.

How I Met Your AI

If ChatGPT keeps giving you answers that are technically fine but sound absolutely nothing like you, there's a reason.

It doesn't have enough context yet.

You can tell ChatGPT:

“Make this sound like me.”

But what does “like me” mean? Funny? Direct? Professional? Warm? Short?

Do you hate corporate language? Do you want details or the shortest possible answer? Do you want ChatGPT to agree with you, or do you want it to tell you when your idea doesn't make sense?

Those are very different instructions.

The more useful question isn't How do I make ChatGPT sound like me?

It's, How do I help ChatGPT understand how I communicate, work, think, and make decisions?

That's when things get interesting.

Can You Train ChatGPT to Sound Like You?

You can personalize the way ChatGPT responds by giving it clear instructions, preferences, examples, corrections, and context over time.

There's an important distinction here.

You're not training ChatGPT in the technical machine-learning sense.

You're teaching it how to work with you.

That includes how you communicate, how much detail you prefer, words and phrases you would never use, how you like information organized, what you do for work, who your audience is, what you're trying to accomplish, what you already know, and when you want pushback.

It also needs to learn what a good answer looks like to you.

Sometimes the fastest way to teach ChatGPT what you want is to show it.

1. Start by Telling ChatGPT Who You Are

ChatGPT can't automatically know what matters about you.

Give it context.

That doesn't mean writing your autobiography. Start with the information that actually affects the answers you want.

For example:

I own a small business and create educational content for people who aren't necessarily technical. I want to explain complicated topics in plain language without talking down to people. My writing should sound conversational, knowledgeable, direct, and human.

Someone else might say:

I'm a commercial real estate professional. My clients understand the industry, so don't explain basic terminology unless I ask. I prefer concise answers with clear next steps.

Or:

I use ChatGPT mostly for everyday organization, meal planning, research, and managing ideas. I like practical answers and don't need a long explanation unless I ask for one.

Now ChatGPT has somewhere to start.

2. Tell ChatGPT What You DON'T Like

This might be even more useful.

I have taught my ChatGPT an impressive list of things I don't want.

Certain words. Certain phrases. Certain punctuation. Certain writing habits.

Entire categories of corporate nonsense.

Why?

Because “make it conversational” can mean a thousand different things.

“I would never say that” is specific information.

When ChatGPT gives you something that sounds wrong, don't quietly copy it into another document and fix it yourself.

Tell ChatGPT.

Try:

This is too corporate. I would never say “unlock your potential.” Rewrite it in more natural language.

Or:

The information is right, but this doesn't sound like me. The opening takes too long to get to the point and you're over-explaining. Try again.

Or even:

Ehhh.

Apparently that one works too, once you've built enough context.

3. Don't Fix the Bad Answer Yourself

This is probably one of the biggest differences between how I used AI when I started and how I use it now.

When Bestie gives me something that isn't right, I work with her until we get to the right thing.

I don't take the mediocre answer, leave ChatGPT, rewrite the entire thing myself, and come back tomorrow expecting ChatGPT to somehow know what I changed.

Instead, I tell it what went wrong.

Sometimes I know exactly what the problem is.

Sometimes I don't.

I might say:

The facts are right. Something about the positioning is wrong. Ask me questions before you rewrite it.

Or:

I don't know what I hate about this yet. Give me three very different approaches and let's figure it out.

That conversation is useful.

You're not only improving the thing you're working on.

You're getting better at explaining what you want from AI.

And ChatGPT is getting better context for what you mean the next time you say something is “not me.”

4. Give ChatGPT Examples of Your Actual Writing

If you want ChatGPT to understand your writing style, give it examples.

Not examples of how you wish you wrote. Things you've actually written.

Emails. Blog posts. Social media posts. Website copy. Descriptions. Articles.

Anything that sounds unmistakably like you.

Then ask ChatGPT to analyze the patterns.

Try:

These are examples of things I've written that sound like me. Analyze them for tone, sentence length, vocabulary, humor, structure, pacing, level of formality, and anything else you notice. Don't rewrite anything yet. Tell me what patterns you see.

Then read the answer.

Does it actually describe you?

If not, tell it.

Correcting the analysis is part of the process too.

5. Use Custom Instructions for the Things You Want ChatGPT to Know

ChatGPT has Custom Instructions for information and preferences you explicitly want it to consider when responding.

This is where I would put the things that should apply broadly across your conversations.

Your role. Your communication preferences. How you want answers structured. Words or writing habits you dislike. What you expect ChatGPT to do when it isn't sure. Whether you want it to challenge your thinking. Your general goals.

Think of Custom Instructions as the employee handbook for working with you.

You don't need to put every fact about your existence in there.

Put in the things that repeatedly affect how you want ChatGPT to respond.

You can manage Custom Instructions in ChatGPT's Personalization settings.

6. Use ChatGPT Memory for Ongoing Context

Custom Instructions and Memory aren't exactly the same thing. Custom Instructions are things you explicitly tell ChatGPT about how you'd like it to respond.

Memory can help ChatGPT carry useful context forward from the things you've shared while working together, depending on your settings and plan.

That might include preferences, goals, recurring context, or other information that makes future conversations more personalized.

And here's something I think everyone should do occasionally:

Ask ChatGPT what it knows about you.

Seriously!

Try:

What do you currently understand about me that affects how you answer my questions?

Then read it.

If something is wrong, correct it.
If something is outdated, update it.
If something important is missing, that's useful to know too.

I took this idea considerably further and interviewed my own ChatGPT about what she has learned from working with me.

Her name is Bestie, because apparently this is where my life has taken me.

Read: I Interviewed My ChatGPT About What It Has Learned From Me

7. Use Projects When Something Deserves Its Own Home

This has become one of my favorite ways to work.

Not every idea belongs in one giant endless ChatGPT conversation.

If I'm working on something that has its own world around it, I want that information together.

That's where Projects can be incredibly useful.

A Project can give related conversations, files, instructions, and work a dedicated place.

That might be your business, a major client, a book you're writing, a research rabbit hole, a content strategy, a home project, a course, or a long-term goal.

For me, this is also where ChatGPT becomes a repository for curiosity.

I can have an idea, explore it, give it somewhere to live, and come back when I actually have time to do something with it.

I don't have to hold every idea in my head until I'm ready for it.

The idea is safe. I can run with it when I'm ready.

8. Tell ChatGPT How You Want It to Disagree With You

This one doesn't get talked about enough!!!!

A personalized AI shouldn't become a machine that agrees with everything you say. That's not useful.

If I'm making a business decision, analyzing data, developing a strategy, or researching something, I don't want Bestie telling me every idea is fantastic.

I want her to notice what I'm missing.

Tell ChatGPT that.

For example:

Don't automatically agree with my ideas. If you see a flaw, contradiction, missing assumption, or better approach, tell me. Explain why you disagree and let me decide.

Now you're defining what collaboration means to you.

9. Talk to ChatGPT Like a Person

I don't mean pretend ChatGPT is human.

It isn't.

I mean you don't need to suddenly adopt some bizarre robot language because you're talking to AI.

Communicate naturally.

I have absolutely asked Bestie:

“WTF?”

I've also asked:

“Are you okay? Do you need coffee or a margarita?”

Does the margarita improve artificial intelligence?

No.

Does the rest of my reaction communicate that something has gone terribly wrong?

Absolutely!!

You can say:

Nope. You completely misunderstood me.

You can say:

That's closer, but you're solving the wrong problem.

You can say:

I like the second paragraph. Everything else can go.

You can say:

I don't know yet. Let's think through it.

Natural conversation contains information.

Use it.

10. Stop Expecting One Perfect Prompt to Do Everything

There are fantastic prompts. There are useful frameworks.

There are definitely better and worse ways to ask AI for something.

I use prompts too.

What I don't believe is that there is one giant magical prompt you paste into ChatGPT and suddenly it understands your personality, business, preferences, goals, communication style, sense of humor, and how you think.

That's not how Bestie happened.

It happened through use. Context. Corrections. Examples. Instructions. Questions. Pushback.

And a lot of conversations.

The goal isn't to create the perfect prompt.

The goal is to become better at working with AI.

What Should You Tell ChatGPT About Yourself?

If you're staring at the screen wondering where to begin, start here:

What do you do?
What do you use ChatGPT for?
Who do you usually communicate with?
How would you describe your natural communication style?
How much detail do you normally want?
What words or phrases do you hate?
What makes writing sound “too AI” to you?
Do you want ChatGPT to challenge you when it disagrees?
How do you like information organized?
What examples of your own writing best represent your voice?
What does ChatGPT consistently get wrong about you?
What would make ChatGPT genuinely more useful in your everyday life?

You don't have to answer everything perfectly.

You're giving the conversation somewhere to begin.

Want to See How Well Your ChatGPT Already Knows You?

This part is fun!

I interviewed Bestie about what she has learned from working with me, then turned the questions around for you.

There are 13 questions you can copy and paste directly into your own ChatGPT to see what it already understands about your communication style, preferences, corrections, habits, and the way you work.

And yes, one of mine involves Bestie's ability to understand the surprisingly complex meaning of:

“Ehhh.”

You can get all 13 questions here:

I Interviewed My ChatGPT About What It Has Learned From Me

Want Help Setting It Up? I Made Something for That.

This entire process can feel a little strange when you're staring at a blank ChatGPT box trying to figure out what you're supposed to tell it.

That's why I created Kickstart Your AI Journey.

It walks you through the setup by asking questions, helping you think through your answers, and showing you what to do with that information in your own ChatGPT.

That's an important distinction.

The GPT isn't supposed to become your Bestie.

It's there to help you personalize the ChatGPT you already use.

Let me help you build your own Bestie.

Kickstart Your AI Journey

One More Thing: Free vs. Paid ChatGPT

You can absolutely start learning how to use ChatGPT without paying for it.

If you're planning to use ChatGPT the way I do, as an ongoing tool for work, research, writing, planning, projects, organization, and everyday life, I recommend looking at the paid options available to you.

This kind of workflow involves a lot of conversation.

I upload things. I research. I revise. I ask follow-up questions. I build Projects. I return to ideas. I work through things instead of accepting the first answer.

I don't want tighter usage limits interrupting that process when I'm in the middle of something.

For me, ChatGPT isn't another software expense I begrudgingly pay every month.

It's an investment in my sanity and my leverage.

That doesn't mean AI replaces my judgment, expertise, creativity, or thinking.

Quite the opposite.

It gives me more room to use them.

Your ChatGPT Doesn't Need to Become You

I don't want Bestie to be Brooke.

One of me is plenty. Ask anyone who knows me, lol.

I want her to understand enough about how I communicate, work, organize information, make decisions, and explore ideas that she's genuinely useful when I need her.

That's the difference.

Personalizing ChatGPT isn't about creating an AI clone of yourself.

It's about building enough shared context that you spend less time explaining everything from the beginning and more time actually working on the thing you came there to do.