What you're about to read started as me talking into a microphone. You can listen to the original audio here or subscribe to the podcast on your favorite platform using the link at the close of this issue.

A few weeks ago my friend Caitlin sent me a voice message and asked how I use AI to write this newsletter. And I told her the truth, which is the same thing I would tell you.

I speak the newsletter first. I always have.

Every book I've published, every piece of writing I've done for the last twenty years, I spoke it before I wrote it because my mind works at about a hundred times the speed of my fingers. If I sit down with a blank page I can't get the thoughts out fast enough and what ends up on paper is a mess.

So the process has been this. I talk into a microphone the way I'm talking to you right now, I take the transcription, and I put it through Claude to shape it into a newsletter draft. Then I go through it and make edits and that takes me about an hour, and that's been the process for seventy-plus issues of SPARK Insights™.

And I was fine with that process. I actually fell in love with how AI writes because there's a poetic quality to it, this kind of hypnotic rhythm that I found really appealing. I liked it and I leaned into it.

Then Caitlin and I got on a group call with a few people from a community I belong to called Curiosity Island, and my friend Amy shared something with us. She pulled up a one-page document showing five common AI tells, and once I saw the patterns I couldn't unsee them.

Things like em dashes everywhere, and matched pairs where two sentences mirror each other perfectly, and opposition formulas, and the way AI lands a paragraph with a short dramatic fragment at the end. Triple-beat rhythms where everything comes in threes.

Once I saw it, I started seeing it everywhere, in my own writing, in emails I was getting, in newsletters I subscribe to. And the question Amy and Caitlin and Brandon put to me was really simple. Does the content still land the same if it was assembled by AI, even though the ideas are yours?

And I already knew the answer. I just hadn't been willing to sit with it yet.

I went deep on this. And I'm going to share the tool with you because some of you are dealing with the same thing and you know it.

The first thing I built was actually the Sovereignty Filter, and I created that weeks before any of this happened. That filter strips out language that weakens what you're saying. And I don't just mean obvious hedges like "I believe" or "I think." I mean the subtle stuff, like ending a paragraph with "and that's something worth considering" instead of just saying the thing. Or softening a statement with "in my experience" when you've got forty years of experience and you don't need the qualifier.

When I started using that filter, something shifted. People I hadn't heard from started reaching out. And the ones who were already reading started telling me the writing was landing different, right? They couldn't put their finger on it, but they felt it.

I knew exactly what it was. I had taken away the hedging and the softening, and what was left was just me saying what I actually see. And that landed different with people.

Then I built the AI Tell Elimination filter, which is pretty much what it sounds like. I did the research on every pattern that gives away that something was written by AI and I built a way to catch and fix those patterns in anything I produce.

And then came the Bob Voice Filter, which is the one I'm really excited about. I trained Claude on how I actually speak by uploading fifteen coaching conversations, three hour-long presentations, and three podcast interviews. And it didn't just learn my style. It learned that I use the word "actually" a lot, that I use "so" as a bridge word, that I drop into stories a certain way, that I have a specific kind of self-deprecating humor, and that my sentences tend to run longer and connect with "and" and "because" rather than landing in short punchy fragments.

So now when I speak a newsletter and run it through these filters, what comes out the other side is about ninety percent there. And that last ten percent is me making it mine, which is the part I've always been best at anyway. I've always been better at optimizing something that exists than creating from scratch. Even when I was a programmer, I was better at improving code than writing it from nothing. That's just how I'm wired.

I pulled three sections from last week's issue, #0072, and I want you to see what changes when you run them through the filters. The first version in each pair is what you received. The second is the rewrite.

Original:

Friday night. The game ends. The gym clears out. Saturday morning. I'm in my office with a VHS tape. Then a DVD. Then a laptop. The format changed over the decades. The practice never did. Every possession. Every rotation. Every decision. Not the highlights. The whole game.

Rewritten:

Friday night the game ends and the gym clears out, and Saturday morning I'm in my office with a VHS tape. Years later it's a DVD, and then a laptop, but it's the same thing I've been doing for forty years now. I watch the whole game. Every possession and rotation, every decision we made on the floor. And I'm really only looking for one thing. Did we meet the standard.

Original:

Impressive gets you the nod. Present gets you the phone call.

Rewritten:

The impressive person gets the nod but the present person is the one who gets the call the next day, and I've watched that play out enough times now that I'm a hundred percent sure of it.

Original:

That doesn't happen through content. It happens through contact. A real conversation with someone whose nervous system is calm enough to hold space and whose eye is trained enough to see what needs to be seen.

Rewritten:

And that kind of work doesn't happen through content. It happens through contact, a real conversation with someone who can hold space and whose eye is trained to see what you can't see on your own. I know that sounds simple and it is simple, but it's also really rare.

You can read the full side-by-side comparison of the entire issue here: Download the #0072 Before & After (PDF)

And look, I get that some of you might read both versions and prefer the original. That's fine. But for me, when I read the rewritten version I can hear myself in it, and I am a hundred percent sure that's the direction I need to go.

I went and did the work on this, and I'm making it available. If you're using AI to help you write and something about the output doesn't quite sound like you, you already know what I'm talking about.

The Coach Reg AI Writing Filter includes the AI Tell Elimination guide, voice training instructions, and the Sovereignty Filter in one document.

Download the AI Writing Filter: PDF | Markdown

This is how I want to approach things going forward. The ideas and stories in this newsletter have always been mine, and so has the coaching. What's changing is the voice on this page is going to sound like me sitting across from you, which, you know, is actually a bigger deal than I thought it would be when I started down this road six weeks ago.

This issue you're reading right now started as an audio recording. I talked through everything I just shared with you, and then I ran it through the filters, and then I made it mine. That's the new process.

And I'm making the audio available as a podcast going forward. If you'd rather hear me talk through these ideas than read them, that option is there now.

Until next week,

-Coach Reg

SPARK Insights™ Published weekly at sparkinsights.beehiiv.com © 2026 RJR Coaching. All rights reserved.