If your AI writing sounds generic, the problem is the source it reads from, not the prompt you wrote. A general chatbot is trained on the whole internet, so it returns the average of everything it has seen, and that average has a sound. To make AI write in your voice, you give it a source you authored: a Brand Brain that holds your identity, your through-line, your frameworks, and your voice rules. Every tool that reads from it writes like you instead of like the internet.

A few months back I sat down to write a week of content with AI, and I did everything you’re supposed to do.

I pasted in my bio. I pasted in three of my best posts. I told it to study my voice and match it.

What came back was clean, fast, and sounded like a stranger doing an impression of me.

It had the cadence of every LinkedIn post written that week. The little inspirational turn at the end. The tidy three-part rhythm. Smooth, competent, and completely not me.

I almost blamed the tool. Then I realized the tool was doing exactly what it was built to do.

Everyone’s pulling from the same well

A general chatbot learned to write by reading an enormous slice of the internet. Billions of pages, most of them average, because most writing is average.

So when you open a blank prompt and ask it to write, it gives you back the center of all that. The most likely next sentence. The safest phrasing. The median of everything it has ever seen.

That median has a texture. You’ve started to recognize it: the even rhythm, the gentle wisdom lines, the closer that wraps things in a bow. It’s the house style of a machine trained on the whole web.

Here’s the part most people miss. Every other person using that same tool, from that same blank prompt, is pulling from the same center. You’re all drawing water from one well. So you all come out sounding like cousins.

Your AI doesn’t sound generic because you wrote a bad prompt. It sounds generic because, with nothing else to go on, the only voice it has is the average one.

I pasted my bio. It didn’t take.

When you drop your bio and a few posts into the chat and ask it to match your voice, you’re handing it a thin, temporary hint.

It reads those few hundred words, makes a loose guess at your style, and the moment the conversation moves on, that guess starts to dissolve. The next session, you start over. You’re re-teaching the same lesson every single time, and it never really takes.

There’s a deeper issue under the obvious one. Your bio describes what you do. It doesn’t carry how you think, the moves your sentences make, the words you’d never use, the through-line that ties everything you make back to one idea. A few pasted samples can’t hold all of that, so the model fills the gaps with the average. Right back to the well.

Stop hinting, start authoring

The fix is to stop hinting and start authoring.

Voice codification is the practice of writing down how you actually sound, on purpose, as a source a tool can read: your sentence rhythm, your vocabulary, the patterns you lean on, the patterns you ban, the way you open and the way you close. Not a vibe in the chat window. A document.

That document is the front of something bigger. I call it a Brand Brain.

A Brand Brain is a single authored source that holds who you are: your identity, your through-line, your voice, your frameworks, the way you actually think and decide. You write it once, in your own words, and from then on every tool you run reads from it instead of from the internet.

The difference is the whole game. Feed a tool the web and you get the average of the web. Feed it your Brain and you get you. It can only write from what it knows, so when the only thing it knows is your authored source, the generic voice has nowhere to come from.

How do I get AI to write in my voice?

You give it a source you wrote, not an instruction you typed. Pasting "write in my voice" into a prompt gives the model a hint it forgets by the next session. Authoring a Brand Brain that holds your identity, voice rules, and frameworks gives it a source it reads from every time, so the output stays in your voice instead of drifting back to generic.

Will it still sound like me, or like a polished version of me?

It sounds like you, because it only knows what you wrote. If your Brain captures your real rhythm and your real vocabulary, including the rough edges, the output carries those too. The model isn’t inventing a smoother you. It’s reproducing the you that you authored.

Rented every time, or owned for good

There are two ways to chase a consistent voice with AI, and they pull in opposite directions.

The first is to prompt it fresh each time. You paste your samples, you write careful instructions, you nudge it back when it drifts. It can work for one piece. The cost is that you pay it again on the next piece, and the next, forever. The knowledge lives in your head and your effort, so nothing compounds.

The second is to author the source once. You write your Brand Brain, and every tool reads from it: the post, the email, the sales page, the caption. You did the thinking up front, and now it pays out every time you produce, without you re-explaining who you are.

The first is rented. The second is owned. One stays the same amount of work forever. The other front-loads the work and then quietly returns it for good.

One source, then a whole team

Once the source exists, you stop thinking in terms of a single assistant.

A general chatbot is one helper reading from everything. What you actually want is a team of specialists, each scoped to one job, all reading from the same Brain you authored: a copywriter, a content strategist, an email sequencer, a social operator, a sales advisor.

It runs in four layers, built from your identity outward. Your identity, named and codified. Your voice, captured so the work sounds like you. Your content, handled. Your business mechanics, built in. The voice layer sits near the foundation on purpose, because if the voice is wrong, everything stacked on top of it is wrong too.

That stack is what I built after twenty years of making other people’s brands coherent and watching my own wait in line. I called it the Brand Intelligence Engine.

The real number it replaces

Let me put a real number on it.

A few years back I paid a copywriter $3,500 to write one sales page. Good work, in my voice, and at the time that was the only way to get it right. One page, one role, one invoice.

The reason that page sounded like me is that a skilled human studied my voice and held it the whole way through. A Brand Brain does the same job, except it holds your voice across every page after the first, and it doesn’t start from zero each time.

Hiring that full team of specialists for real runs $30,000 to $70,000 a year, plus retainers. Most solo experts can’t carry that, so the writing falls back on them, and they end up pasting their bio into a chatbot at eleven at night, getting the average back, and wondering why their AI sounds like everyone else’s.

The way in

I’m opening the Brand Intelligence Engine at founder pricing for the first fifty people in, at $997. Less than I once paid for a single page. You can see how the whole engine works if you want the full breakdown.

It isn’t subscription software. You build it once, in your own Claude, and the team is yours for good. It comes with a 30-day guarantee, so the only real risk is the afternoon it takes to write your Brain.

If building the full engine is more than you want to take on right now, there’s a smaller way in. The $27 Expert Revenue Blueprint teaches the framework underneath all of this: the gap between what you know and what you actually earn from it, and how to start closing it.

Three things I’d leave you with

1. Your AI sounds generic because it reads from the average of the internet. Change what it reads from and you change how it sounds.
2. Pasting your bio is a hint the model forgets. Authoring a Brand Brain is a source it reads from every time.
3. Do the thinking once, write it down, and let every tool you run produce from it for good.

If you’ve ever read AI output of your own and felt the quiet disappointment of "that’s close, but it isn’t me," you already understand the problem. The source was wrong. You can change the source.

The door is open if you want to see how.