Last week, I opened an email from one of my favorite online creators, someone I’ve followed for years. Her emails are usually long, soulful, messy-in-a-good-way. I read every word because I love being in her energy.

But this time?

Something felt off.

It was polished. Flat. Empty.

I’m sure by now you can tell too. It’s getting almost impossible to miss, right?… The slightly formulaic structure, the soft-sell tone, the bland little “perfect” cadence that doesn’t quite say anything at all.

There was no doubt in my mind. The woman I idolized for her uncanny ability to show up and keep it real had used AI to write this email.

I was… (perhaps to an inappropriate degree) devastated.

I unsubscribed on the spot.

Not because I’m against using AI. I mean, that would be a tad hypocritical, seeing as I use it every single day.

I unsubscribed because

I wasn’t there for the content.
I was there for her.

Her voice. Her weird little side tangents. Her humanness.

And that email wasn’t her. So it became instantly irrelevant to me… like white noise in an inbox already drowning in AI-isms.

We can’t fully blame AI, though. I mean, it’s doing an adequate job. But when everyone uses the same tool the same way, everything starts to blur.

And when everything blurs…

“human” becomes the thing people are actually hungry for again.

Your voice becomes the differentiator. Writing that feels like a person. Thoughts that feel like they came from an actual human brain. Sentences that are slightly imperfect… but unmistakably you.

So what’s the solution?

How do we use a tool that seems to have the capacity to help us, but spits out copy that sounds like it was approved by a committee of robots trying to be likable?

We use it, we just shift HOW we use it.

Back in the day, I hired Tracy, a copy editor. Not a copywriter, an editor.
(for those not in the know… copywriters write for you, copy editors refine what you’ve already written.)

She was incredible. Truly magical. She had this uncanny ability to take what I would write and elevate it into something that made me feel like a Pulitzer Prize-winning writer.

Here’s how the process would go:

Step one: The first pancake

I would get all my ideas out in the most sloppy, disorganized mess. I called it the first pancake because — and let me know if you can relate to this — even though I’ve made Sunday pancakes about a thousand times in my life, I always seem to burn the first one. A crispy, slightly disfigured mess that is NOT fit for human consumption.

Just like my first draft.

No one with a pulse should ever lay eyes on it.

Step two: The second pancake

Now that the grill was at the appropriate temp, I was ready for pancake #2. I would tighten sentences, add in my dry, self-deprecating wit, maybe whip out my trusty thesaurus if I couldn’t think of that perfect word that would make everything click.

Then, when I thought it was perfectly acceptable to send to an actual human, I would send it off to Tracy.

A little while later, she would return it “gently” marked up. Just kidding — it actually looked like the Google Doc version of a Game of Thrones episode, with red strokes all over the place… my ego would definitely get a good beating.

RELATED: 

And then I would review what she had done.

Despite a page filled with (what I perceived as slightly aggressive) red marks, she hadn’t really changed anything at all.

My voice was there (albeit better). My thoughts were all intact, just tighter. And the entire thing would flow in a way that made reading it actually enjoyable.

Turns out a second set of eyes on your ideas is never a bad thing. It can turn a decent draft into something people actually want to read (which is why I paid her the big bucks).

Tracy has since moved on to more lucrative endeavors.

(I don’t actually know that for sure, but she was a genius, so I choose to believe she’s on a beach somewhere, lightly tipsy, wildly unbothered, and making dolla dolla bills by being hotter and smarter than the rest of us.)

Now I use ChatGPT the same way I used Tracy,
as an editor, not a ghostwriter.

Because I know how the process is supposed to go (pancake one, pancake two, ego punch), I know exactly how to instruct it to give me exactly what I need.

To recap,

here’s what you need to do if you want to steal my process

(which, by the way, is how I wrote this very email you’re reading):

01. Start with your first pancake

Write out (or voice-note) everything you want to say. Do not worry about making it good. Do not worry about making it make sense. This is not the time for excellence. Just talk and talk and talk. Let it be messy. Let it be unhinged. Let it be… your first pancake.

02. Give it a light pass (your second pancake)

Now go back over what you wrote or said and edit it the way you would if a friend texted you for help and you were being both helpful and slightly ruthless. Honestly, this part takes the longest, but it is so worth it, and if you do it well, this is the blueprint your AI copy editor will use, ensuring it sounds exactly like you.

03. Give ChatGPT (or your AI of choice) this prompt:

🤖 I want you to work as my copy editor, not copywriter. You are taking what I wrote and elevating it, not turning it into something new. Keep my rambles, my tone, my rhythm. It should still sound like me, just with fewer typos, better flow, more impact where appropriate. Here is the copy I want you to clean up:

04. Let Chat do its thing…

Then refine as needed. This is the final pass. At this point, you should have a pretty stellar piece of writing. Now you’re just doing one last sweep to catch any weird little AI-isms that may have snuck in when Chat got a little too excited.

This is how you stay human in a sea of people handing the mic over to robots.

This is how you stay relevant, and more importantly, interesting to an audience that is so, so tired of cookie-cutter.