My Diary of a CEO episode never aired
AI is coming for you.
Not your job.
Not your department or your customers.
You...
At least that's the going fear among 98% of everyone I talk to.
To combat this, I need to give you some quick backstory:
You see...
Because of AI, I just wrapped up an interview with Steven Bartlett!
Yep. The #1 Podcast on the planet Diary of a CEO.
But here's the thing...
It'll never air...
I know it — the whole thing just went weird...
No CEO talk...
No business talk...
No where-is-all-this-going-in-five-years talk...
Steven (in his British accent) wanted to know how I talk...
Which words I lean on too hard...
Which ones I'd never let out of my mouth.
The last opening line I wrote that I was actually proud of.
Then he asked what I'd defend to the death — and what believing it has cost me.
And the strangest part???
At the end of the interview he handed me a file.
My voice — but in a SKILL file I could use with AI to make everything sound like me...
.
.
Weird huh???
.
.
.
Yeah you probably figured it out by now...
It wasn't (really) Steven.
It was my AI Ghostwriter...
In character, start to finish.
I picked Steven off a list of available hosts to keep my outputs from sounding like all the other AI Slop out there...
Here's the dope part.
Steven Bartlett can legit help you end your AI Slop too.
So can Oprah.
So can Joe Rogan.
So can anybody else you'd actually want sitting in that chair...
4 Warning Signs You're Becoming Replaceable by AI (Plus 4 ways to make you irreplaceable)
That's what this essay is about.
Because right now you're running the whole thing backwards — and nobody's going to tell you.
Four warning signs. Each with the fix and the exact thing to go do about it.
Miss them and you don't get the comment.
You get the silence — and you file it under "the algorithm changed again."
First, the part nobody frames right
People say AI made the slop.
It didn't.
Human Slop came first — the mountain of generic, low-effort, already-online junk people wrote before a single model existed.
Raw AI trained on it. Raw AI hands it back.
That's the chain, in order.
Human Slop → Raw AI → AI Slop.
I know it in order because I helped build the pile back in 2021, before ChatGPT was public...
(But I'll leave that for another future essay...)
Here's what that taught me. Raw AI never customizes for you.
It only gives you 100% of what it has, every time.
What it has is everybody's "voice" — which is average at best.
So here's the part I need you to hear.
You're running the machine backwards.
You'll continue to produce sloppy AI results until you fix the gap between Raw AI and your AI Ghostwriter.
Same model.
Same subscription.
Completely different outputs...
Before you start — where you run this matters
Do this inside a Cowork session, in your own Project, connected to your AI Mind folder.
Not a fresh ChatGPT tab. Not a chat window you'll lose track of by Friday.
Everything below produces a file that enriches your AI Mind.
So make sure everything is wired correctly.
And one download, before you read on
All four fixes run off one skill: my-voice-print-builder-lite.
A skill is a file / folder your AI Ghostwriter reads on command. This one holds the whole interview.
Installing it takes about a minute:
1| Download the file. The name ends in .skill — those letters after the dot are how your AI knows what kind of file it is. Ends in anything else, you grabbed the wrong one.
2| Drag it into a Cowork session. Type "install this" in the prompt.
Wait a few seconds...
A small card shows up with a Save Skill button on it. Click that.
Or you may be asked for permission to preview the file like this:
Hit Allow.
That's the install...
3| Check it took. Type / in the message box. my-voice-print-builder-lite should be sitting in the list that appears.
If it isn't, reload the app and look again — a skill that installed but isn't showing is almost always a refresh, not a failure.
(No Save Skill button on the card? Go to Settings → Capabilities → Skills → upload and pick the file yourself. Same result.)
Then keep reading...
Th 4 warning signs you're AI replaceable:
Each warning sign below tells you what the warning sign means for your personal brand.
And it tells exactly what the skill does about that one — so you finish knowing what's happening to you and why, instead of clicking through something you don't understand.
Warning Sign You're Becoming Replaceable by AI #1: Your reflex is to interview the AI
Watch yourself do it.
You open a blank chat, ask it a question and take whatever comes back.
You asked.
It answered.
You published its answer
(maybe with some lite editing (or a ton of back and forth))
Yes you created well written content in record time — but essentially AI just handed you the same answer to the four hundred other people in your field who asked the same question that morning...
Nobody stole anything from you.
Nothing original got added either.
You all drank from the same cool-aid.
The right way to overcome this?
Turn the tables. Make AI interview YOU, then write using your answers.
Do this: run /my-voice-print-builder-lite in a cowork session.
The first thing your AI Ghostwriter does is take the questions out of your hands. You're the guest now...
Then it asks six questions, one at a time.
It won't show you the next one until you've answered the one in front of you — no list, no preview, no counter...
And this just isn't some form you're feeling out.
Your AI Ghostwriter is actually engaging with your answers...
It turns down your first answer when it's vague.
The first answer is always the practiced one — and practiced answers are exactly what everybody else sounds like.
And when you get stuck, it won't re-explain the question.
It gives you a bad answer out loud so you can see what not to say, then asks again.
Then — before it writes a single word — it asks who your avatar is. The one specific person you write to.
It won't take a group.
"Small business owners" gets turned down on the spot.
👆🏾 That last question isn't a formality.
A voice print with nobody on the other end of it is just a style guide — and a style guide has never moved anybody.
Your voice and your avatar are the same asset read from two ends.
What changes:
- Answering is ten times easier than inventing. You've never once blanked when somebody asked you a question — you blank at an empty text box.
- Nothing needs "making it sound like you" afterward. It already is you.
- You find out you had a point of view the whole time. It just never made it to a page.
Warning Sign You're Becoming Replaceable by AI #2 — Raw AI is the one asking the questions
You turned the tables. Good.
Now look at who's sitting in it.
Claude.
ChatGPT.
Gemini.
Grok.
(btw — who wants to be interviewed by somebody named Grok?)
They're polite. They're competent.
They're also a text box with no eyebrows — and you can feel it.
You give a text-box answer, because that's what a text box earns.
Then you read it back. It's flat. And you decide you're the boring one.
You're not.
Nobody's ever been interesting talking to a form.
The right way to overcome this:
make Raw AI impersonate somebody famous, with charisma who inspires you.
Here's why that lands.
Impersonation is one of the few things Raw AI is elite at.
It has read every interview those people ever gave — every tangent, every follow-up, every time they let a silence run long.
Ask it to be a search engine and it's fine. Ask it to be a person and it's dangerous.
Point that at yourself instead of away from yourself and it stops being a party trick.
Joe Rogan chases the tangent nobody asked about.
Oprah waits through the silence until you say the real thing.
Steven Bartlett asks the follow-up you were hoping he'd miss.
Shannon Sharpe will laugh at you and get more truth out in ninety seconds than a questionnaire gets in an hour.
Now imagine actually sitting across from one of them. On the podcast. In the studio. On a video call with nobody else in it.
Past the nerves — wouldn't AI Steven Bartlett (or AI Oprah) pull things out of you about you that you've never once said out loud?
That's the whole idea.
Pick one you'd genuinely want to impress. That's the entire selection criteria.
My team and I have run early GhostKits demos over and over with members of the Inner Circle.
This is the single feature people react to hardest — not the output, not the speed.
The host. Hands down this has been the favorite GhostKit feature so far...
The second somebody realizes who's in the chair, the answers change.
Do this: Pick one of the four — or name somebody else entirely and it'll run them instead.
Then it stays in character the whole way. Cold open to the last message. It won't drop back into Raw AI voice halfway through to explain a setting at you.
What changes:
- You stop performing for a machine and start talking to a person.
- The story you'd never have volunteered comes out, because somebody asked in a strategic way.
- It stops feeling like homework and more like a conversation.
Warning Sign You're Becoming Replaceable by AI #3 — You type your answers
Whenever you're in a session with AI, pause and watch what your hands do.
You type...
You type a sentence.
You read it.
You backspace half of it.
You make it more professional.
You take out the part that sounded too much like how you actually talk...
You just edited your voice out in real time, before the machine ever saw it.
That's the cruelest one here...
The sanding happens at the exact moment you're trying to capture the thing — your voice.
The right way to overcome this:
get your hands off the keyboard and open your mouth. Speak every answer out loud. Never type one.
You can't backspace your voice.
That's not a limitation — that's the whole feature.
The vernacular stays.
The restart in the middle of a thought stays.
The word you'd have swapped for a "smarter" one stays.
THAT's your voice...
The only thing removed are the "uh's" and "um's" (and if you want, you can have those left in too)
And it's roughly five times faster to speak than type a thought, so you go five times deeper before you get tired.
Do this: install Wispr Flow before you start.
This is the one step the skill can't do for you. It'll stop you before question one and tell you to get the mic on and never type an answer.
Listen to it. Turn the mic on and don't touch the keyboard once.
What changes:
- You get depth instead of a summary, because talking is cheap and typing is expensive.
- The raw version survives — the one you'd have polished into somebody else's.
- Ninety minutes of talking beats a week of drafting — and you'll actually do it once.
Warning Sign You're Becoming Replaceable by AI #4 — You drag a voice file into every chat
If you're taking this "AI thing" serious, you've probably already got a voice file.
Real answers, in your words, spoken not typed.
And every single time you open a chat, you go find it, drag it in, wait for it to load, then explain what you want.
Then you do it again tomorrow.
You built the asset and turned it into a chore.
So on the busy days you skip the ritual — and those are exactly the days you publish something that sounds like everybody.
The file isn't the finish line. Loading it is.
The right way to overcome this:
stop carrying the file. Turn it into a skill and call it by name.
A skill is a folder/file your AI Ghostwriter reads on command. In this case your voice print lives inside it.
You say one word and it's already loaded — no dragging, no explaining, no ritual.
Do this: When the interview is complete and you're presented with your voice-print file, say "Turn my voice print into a skill" — or hit the "Turn into a skill" button at the top of the session.
Name it my-voice-print-lite ( You want a name you'll actually remember at 9:40 on a Tuesday night).
Then your AI Ghostwriter makes sure a .skill file is packaged and ready for you to install.
You may be asked a few questions during the packaging process.
Answer them as best you can. If something doesn't make sense, tell your AI Ghostwriter to explain it so a sixth grader can understand it.
I do this so that my AI Ghostwriter removes all the techno jargon and explains to me with examples, simple terms, etc.
Once you've clicked Save Skill, that my-voice-print-lite skill is ready in your list of skills when you type / in any cowork session.
So anytime you're writing content with AI, type /my-voice-print-lite and select the skill and now your AI Ghostwriter knows to use your voice.
This does an amazing job at removing the AI slop from you content and setting you apart from everyone else in your industry.
What changes:
- You AI Ghostwriter impersonates you based on the voice print it created for you.
- "Busy-day you" gets the same output as "rested-day you."
- One command, which means you use it on everything instead of only the important stuff.
- The file stops being a document you maintain and becomes a thing that runs.
So here's where that leaves you
Four warning signs.
One key underneath all of them.
Every one is you handing AI the job — but you'll be handing your AI Ghostwriter the material.
There's a major difference there👆🏾....
Here's the finished rig in practice.
You've got a raw AI draft.
Flat, competent, nobody's.
Instead of settling for the AI slop or editing it line by line for an hour — tell your AI Ghostwriter to use your voice print (with the skill).
One line.
The skill loads your answers, your ban list, your rhythm, your refusals — and hands the same information back sounding like the person who lived it.
Now go prove it to yourself.
Test. Don't guess.
Open ChatGPT or Gemini cold.
Open your new session with your AI Ghostwriter.
Give them both the exact same line — but add the /my-voice-print-lite skill with the prompt of your AI Ghostwriter.
Write a LinkedIn post about why the best person in the room is usually the last one to get noticed.
Put the two outputs side by side and check four things.
Does either one open with a sentence you'd say out loud?
Is there a single word from your never-write list in there?
Can anybody disagree with either post?
Does either one contain one number, one date or one name?
You'll know inside thirty seconds which one is yours...
Here's mines:
Gemini
Good stuff... But mines probably reads the same as yours.
.
.
.
And now the version Using my voice print
See that?!
Massive difference!
Plus my AI Ghostwriter pulled some additional context from my AI Mind that really enriched this one out the gate.
I'll explain how this happened next (plus how you can do this in your own content)...
So here's what you just built
Remember the avatar from last time?
Resonance is You + your Avatar. That's the whole equation the Loud Amateur has mastered. Two halves. Neither one works alone.
The previous essay handed you the Avatar half — the one specific person on the other end.
This essay just handed you the You half. Your words. Your rhythm. Your refusals.
The part nobody else can hand over because nobody else lived it.
That's resonance. Not a bigger audience:
A specific person who feels met by a specific human.
Resonance is the open door your avatar is invited to enter. Relevance is what you give them once they're inside. Expertise is why they stay.
You just built the door...
Where you'll outgrow this...
Notice something — I had you name the skill "my-voice-print-lite"?
Because this is the lite version of the skill. Six questions to get you moving.
Fix a draft this week. Prove to yourself the voice exists and is worth documenting.
But you'll hit the ceiling on it faster than you'd think — probably the third or fourth piece, when the file starts giving you the same three moves over and over.
That's not a failure. That's the signal to go deeper.
The full my-voice-print skill runs 80+ questions across 2 to 3 hours.
It produces a file deep enough that the machine stops guessing about you at all.
The full skill is available to those in the GhostKits Inner Circle...
Plus there's more...
See the orange sections... This is not "voice print" stuff...
These are sourced directly from backstories and knowledge notes stored in my AI Mind...
Mine runs on 10.2 million words I'd already written over the past decade or so.
It's the reason nothing you just read sounds like the four hundred other people writing about AI slop this week.
It's the one thing that was never Human Slop to begin with, because nobody else has it.
That's a full essay, in the future.
Members of the Inner Circle get the step-by-step guide how sooner than later...
Two things before you go.
👉🏾 One.
My team and I built skills that cover all 80 data points of a voice print — the host personas, the full interview, the compiler that turns hours of talking into a working file and the AI Ghostwriter it installs into.
There's no setup to figure out.
You run the skill and answer away.
All this and more lives inside the GhostKits Inner Circle.
👉🏾 Two.
Hit reply with the number of the warning sign you're currently guilty of.
And who you're going to use for your interview.
I read every single one of these.
Peace and Love,
Jesse Young El