Can AI Write Content in Your Voice? (The 10-80-10 Rule)
Not really. And if you're a coach, it's the one job we wouldn't hand it. AI is great at research, reminders and the grunt work. Your voice is the thing people are buying. Here's how we use AI for content without sounding like AI.
Watch on YouTube: Can AI Write Content in Your Voice? (The 10-80-10 Rule)
This comes from Leland's conversation with Alex Changho, host of the Cool Shit for Coaches podcast. Thanks, Alex, for the great questions.
Can AI write content in your voice?
Mostly no. AI can copy your words, but it can't bring your unique perspective, and that's what makes a coach worth following. It gets closer if it's trained on lots of your video transcripts, so it has your cadence and rhythm. Even then, Leland doesn't think your voice is worth outsourcing. The risk is too high.
Leland's answer, when he was asked what AI doesn't do well: "Especially for coaches and experts, writing emails and commenting, it's never going to get your voice. Basically, unless you're doing lots of videos and it has transcripts and it has your cadence and your flow, your rhythm, everything."
The video opens on the same idea: "My AI bot's not going to show my unique perspective. If it tries to show unique perspective, I'm going to look dumb. Otherwise, it's going to look generic."
So you get two bad options. It guesses at your opinion and gets it wrong. Or it plays it safe and sounds like everybody else.
Why shouldn't coaches let AI write their comments and emails?
Because for a coach, your voice is the product. Your perspective, your questions and the way you work with people are what clients pay for. Comments and emails are where people first meet that voice. If those come out generic, people decide you're generic. And one bad AI comment can stick to you for a long time.
Leland calls it the big mistake: "Big mistake coaches make is that they try to outsource their voice." The whole reason to leave a comment is to show how you think. "I should be leaving comments that are insightful, because that shows my unique perspective."
Here's the damage he's seen: "Leaving one burner AI slop comment, you become the AI slop comment person in the eyes of everybody who sees it."
That changed how he works. "I don't take that as lightly as I used to, where now I would never outsource that. My voice is my voice."
People notice, too. In a YouGov survey for Pangram of 2,557 US adults in April 2026, 69% said they trust AI-generated content less than content made by people. And 61% said they're unlikely to read or engage with content if they think it's AI. Your comment doesn't have to be AI to get skipped. It just has to sound like it.
What is the 10-80-10 rule for AI?
Definition: The 10-80-10 rule splits a task three ways. A human does the first 10%: the idea, the direction, the raw material. AI does the middle 80%: the research, the drafts, the busywork. Then a human does the last 10%: checks it, fixes it, and is the one who hits send. Nothing goes out without you.
Leland's version, word for word: "Human does 10%, Grok does 80%, human finalizes and sends everything, because you just don't want to make any mistakes." (Grok is the AI we run our bots on. The rule works with any tool.)
The last 10% is the part people skip. It's also the part that protects your name.
Here's what that split looks like for a piece of content:
| Part | Who does it | What it looks like |
|---|---|---|
| First 10% | You | Say it on camera or in a voice memo, in your words |
| Middle 80% | AI | Transcribe it, research it, shape it, draft it |
| Last 10% | You | Read it, fix what isn't you, then send it |
It's how we make posts like this one. We talk on camera first. The transcript does a lot of the heavy lifting. And nothing goes out until we've read it.
What can AI do well for coaches?
Plenty, as long as it's not your voice. Leland uses AI for reminders and keeping people on schedule. It finds things worth doing, like broken links, missing playlists and keyword fixes on our YouTube channel. It does the 80% of a task that's busywork. Then it hands you the short list of things only you should touch.
On voice versus everything else, Leland's line is clear: "I think it's really good for reminders and just keeping people on schedule. But branding and voice, you can do so much more damage by doing that badly."
In another clip from the same conversation, he shared what our YouTube bot did. It found four of what it called "mega videos" that "just weren't linked properly, didn't have SEO proper, just little small changes." Then it did the work: "It logged into our YouTube studio and did 95% of it, and then there was 5% that I had to do... took me 5 minutes."
Why bother? "We just wouldn't have done it. We have other stuff to do. I would rather be playing with my kids."
That's the split we want. AI handles the stuff that eats your week. You keep the parts that are you. If you're still working out which tasks those are, start with your numbers. Our content scorecard post shows what we track.
Free training
Want AI doing the grunt work, not your voice?
The training shows what to fix first.
Watch the trainingWhat is the "dog in the room" problem with AI?
It's when you don't know enough to judge the advice AI gives you. Leland credits the idea to Scott Adams. Your dog has no idea how much smarter you are than it is. A little smarter or a ton smarter, the dog can't tell. When you're the dog in the room, you can't tell if AI's advice is any good.
Leland's version: "Your dog doesn't know how much smarter you are than the dog. It only knows how smart the dog is... And so when you're the dog in the room, you don't know if the advice that you're getting is any good."
This is why handing AI your voice is risky. If you don't know good copy, you can't tell good copy from bad. The draft sounds fine to you, and it goes out.
The fix isn't to stop using AI. It's to get smarter than the dog.
How do you use AI for content without sounding like AI?
Flip the job. Don't ask AI to write for you. Ask it to critique your work the way world-class marketers would. Ask what Ogilvy would think of this. Ask what Frank Kern or Dan Kennedy would change. Then ask why it gave that feedback. You learn the basics, your writing gets better, and the words stay yours.
Leland: "You give AI the prompt to think like the great world-class marketers and advertisers. What would Ogilvy think of this? Even some of the modern ones, what would Frank Kern think of this?"
He goes back to the people who started it all, "the first person who wrote the book on advertising." His reasoning: the modern marketers all point back to the same classic books. "Those are the people that you should have as your mentors with AI."
He says the difference shows fast: "If you have Ogilvy look at any of your stuff... you'll be like, 'Oh yeah, this is actually way better than anything.'" Leland's done this a lot himself: "I've had Dan Kennedy analyze my copy a lot."
Here's how to run it:
- Write it yourself first. Say it out loud or type it fast. Your words, your take.
- Pick a mentor. Ogilvy, Dan Kennedy or Frank Kern. Use the words "world-class."
- Ask for the critique. "What would Ogilvy think of this?"
- Ask why. "Why are you giving that feedback?" This is where you learn.
- Fix it yourself. Make the changes in your own words. Then you send it.
Leland's advice for getting out of the dog seat: "Study those old people and have those old people look at your stuff and give you feedback and ask why it's giving feedback that way."
Can AI get better at your voice over time?
Some. It gets better if two things are true. You understand the basics, meaning the foundational marketers from the 1900s. And it's trained on your voice, meaning you make videos and it learns from your transcripts. Then it can be pretty good. Leland still doesn't think your comments and emails are worth outsourcing.
His full take: "I don't think it's good for writing in comments, but it can get better if you understand the basics... and it's trained on your voice. Meaning you just do videos, it uses your transcript, and then it can be pretty good. But I still don't think that that's worth outsourcing."
So keep making videos. They give AI the best raw material it can get. Your coaching on camera becomes the 10% that starts everything. The same videos can feed a whole content plan, like we show in how to get content ideas from one keyword.
FAQ
Can I train AI on my own voice?
Partly. If you make lots of videos, AI can learn your cadence and rhythm from the transcripts. It gets closer, but Leland still wouldn't outsource comments or emails to it.
What's the 10-80-10 rule?
A human does the first 10% of a task, AI does the middle 80%, and a human finalizes and sends. Nothing goes out without you checking it first.
Should coaches use AI to write social media comments?
We wouldn't. Comments are where people see your perspective. One generic AI comment can make you "the AI slop comment person" to everyone who reads it.
What's the best way to use AI for writing?
As a critic. Write it yourself, then ask AI what a world-class marketer like Ogilvy or Dan Kennedy would think of it, and why. Then make the changes yourself.
Your voice is your voice. Let AI do the research and the busywork, and keep the parts people are paying for.
Free training
Let AI do the 80%. Keep the voice.
The business should free you up to do what you love, more coaching or more running the business, your choice.
Watch the trainingAbout the authors
