Canonical /story. Revision 1. A standing page, not an article. No date, no read-next.
The employer is unnamed throughout. Permission items and brackets are listed at the foot.
On November 30, 2022, I was sitting at my desk at a construction chemicals company, where I was the director of marketing. I saw that something called ChatGPT had been released. I opened it that day.
I asked it about the difference between wisdom and information. I have no idea why that was my first question. I expected something like the chat toys from years earlier, a keyword match dressed up as a conversation.
[THE REAL ANSWER GOES HERE, if you can recover it. Placeholder below, marked as reconstructed until you find the original.]
Reconstructed, not the original: it said information is data given form and context, the kind of thing that can be transmitted and looked up, while wisdom is different in kind rather than degree. Wisdom is knowing which information matters in a situation and what to do with it. A person can hold a great deal of information and still act unwisely. It said wisdom develops through lived experience rather than study alone.
Read that now and it is flat. Slightly textbook. The kind of draft you would send back.
In 2022 it stopped me cold. Not because it was deep. Because it was sane, in complete sentences, in answer to a question I had asked in my own words. Nothing had ever done that before.
So I kept asking.
Day five
Within a few days I was telling it about myself. Enneagram, personality type, how I think. I wanted to see whether it would hold any of that and answer differently because of it. I started pushing on voice and style.
Then I went and did something that turned out to matter more than any of it.
A close friend of mine works in occupational safety. I know his work, and I know what is hard about it. So I spent about five days doing deep dives on OSHA documentation and training, inside a context window the size of a napkin, so that when I showed him this thing it would already be standing inside his problem instead of next to it.
The intuition was this. Only if I could meet him where he actually worked would he be able to judge whether it was any good. A demo of a clever machine proves nothing. A machine that knows your world is a different claim.
Eight days after launch he came back from vacation and I pulled him into a conference room, which I had never done in all the years I have known him, and I told him the world had changed while he was gone.
He read it. He was impressed. He seemed to miss the entire point.
He did not use AI again for another two years.
Where that instinct came from
The going-and-learning-OSHA move is the only part of the first week I would defend as skill, and it did not come from marketing.
Before any of this I did bilingual frontline work with immigrant and Hispanic families in Kansas City, in family services and early childhood and ESL. That is where my Spanish came from. I was called to ministry in 2007. From 2011 to 2022 I founded and led Mission House Network, an intentional community here, and about a hundred young adults came through it over eleven years. I learned digital marketing during those years for a plain reason, which is that I wanted to tell that community's stories and nobody else was going to.
Eleven years of that teaches you one thing above all others. You do not learn what a person needs by studying the category they belong to. You learn it by being in the room with them for a long time, in their language.
That is why my fifth day with a general purpose machine was spent learning occupational safety documentation for a friend. And it is the reason the company I run now is built on context rather than on capability. Capability is the part everyone has. Context is the part that makes it worth trusting.
The stack
By early 2023 I believed this thing was going to change content marketing, and that it would land hardest in a business like ours, where the writing is technical and precise and nobody wants to do it.
So I started drafting. Articles about concrete chemicals. And for about a month I walked around that building carrying a stack of them roughly an inch thick, knowing that not one person there had any idea a computer could write now.
I was not worried about getting in trouble. I was worried about something else.
I knew there were two easy readings available. The first was that I had lifted the articles off the internet. The second was that I had gotten distracted by a tech fad. If either one landed, the idea would not just be rejected. The organization would be inoculated against it, and I would have spent the one chance anybody gets.
So I waited for a version of the reveal they could accept, which is the opposite of what I wanted to do. When I finally moved, I told a senior executive I had done some internet research and wanted to know whether the drafts might have any legitimacy as blog posts.
He said they were pretty great.
I showed the president almost as an afterthought. He called it our first breakthrough.
I want to sit on that for a second, because it is the strangest thing in my own story. To get the most advanced work in the building read at all, I had to obscure who wrote it. I now run a company whose entire argument is that authorship has to be visible and checkable. I did not arrive at that from theory.
The sandbox
What happened next looked like a demotion and was not one.
I went from director of marketing, running a team, to building everything digital and AI. Off the org chart, into a sandbox. My title got smaller and the work got much larger, and I preferred it. I had trust and a strategy people could see working and could not entirely follow, and I did whatever I thought was right.
I built a content operation. Hand-fed context at first, then Whisper, then the API, then retrieval. Organic traffic went from around fifty thousand visits a year to around a hundred and forty thousand.
It took me three more years to notice that I had built myself a sandbox with real trust and real stakes and no audience watching, and that this is the exact thing I now tell organizations they need before they touch any of it.
Three years of showing people
The rest of it I did for free.
First that friend. Then a team at a large manufacturer. Then a philanthropy group. Then a Kansas City nonprofit. Then movement leaders, including Alan Hirsch. All of it free, all of it alongside a full time job.
People ask why. The honest answer has three parts. I had a job that paid better than anything I had ever made. There was no market yet, and trying to charge in 2023 would have cost me most of the chances I got to build. And I do not really think about money. I did not start thinking about it until I decided to start a business.
But the timeline had to be this way, and I only understand that in hindsight. This is a market built on trust, and it is opening now. Three years of building in front of the people who knew me best meant that when it opened, they were the first ones to ask me to build for them.
Here is what those three years actually taught me, and it is not what I went looking for.
People were astonished. People did not adopt.
They would move their schedules to watch me work. They would pass the thing around. And then nothing. Not my friend. Not the smartest published authors I know, after years of watching me do this constantly. Almost nobody.
I had assumed anyone would see it once and understand. I was wrong about that for three years running, in front of a sample size of everyone I know.
For a long time I read that as a failure in people. I do not anymore. It is the fastest spreading technology in history and also the hardest to actually grasp, which is a combination nobody has dealt with before. It looks like a party trick, so you try it, and then it makes a mistake, and some part of you is relieved. Or you sit in front of a blank box and cannot think of what to ask. Or you can, and you do not have the stamina for the back and forth, which is tiring and looks like nothing from outside.
It was not that people could not. It was that it was too early. It is just starting now.
That result is the reason my company sells a hundred hours of in person training and not a login.
What it cost me
By 2025 I was building with agents, running two builds at once, and shipping at a rate I would not have believed a year earlier. I had gone from vibe coding a front end to assuming a full stack app would work the same way.
It does not. I walked straight into it. A week into building a custom agentic CRM I gave a demo that did not work, in front of a room I had spent two years teaching to expect things in a week. I delivered the full system in about six weeks against a three month timeline and it did not matter, because the room had decided in week one.
I have written the whole of that one out properly, including the part that took months, which was not diagnosis but learning back end architecture from a standing start.
Two things came out of it. A rule, which is that I do not test in public anymore. And the thing I would not trade, which is that the same day it became clear I had to build a real data chain was the day it became clear I could build one from the prompt level. That idea is under everything I have made since.
Sometimes you say you will be back in a week and it turns out to be a longer trip, and you come back with a map nobody had.
The Sunday call
Alan and I had a standing call, Sundays at seven, for years. One week I asked Brad to join. I did not describe the idea. I built it in an afternoon and let them click on it, and what I said on that call is close to what the business is now. That call has its own piece.
Alan asked me to build his platform for real, and I thought side project. Then it was Alan and Brad. Then it became clear that building for leaders and building for organizations were the same work, and that leaders come first only because leaders can decide faster. Then a nonprofit I had known for years had grant money and wanted their people trained.
In March 2026 I went full time on Movemental. Brad is CEO, because we sell trust and he is the one who already holds it in the rooms that matter. Alan is the voice the network gathers around. I built the engine, and I own it.
What I hold to
AI does the gathering and the drafting. It never does the discernment, the care, or the human voice.
You can take that line and use it without ever speaking to me. It settles most of the hard questions on its own.
The rest of it. Every word that goes out under someone's name is theirs. Sources get named. Nothing gets invented. If a claim is not checkable, it does not get made. Where vulnerable people are involved the bar goes up, not down. And nothing gets built for an organization before its people are ready for it, which is why the order is safety, then sandbox, then training, then tech, and never the reverse.
Those are refusals. They cost me work I would otherwise take.
Who this is for
If you lead an organization whose work depends on people trusting you, and you are trying to work out what to do about this without either ignoring it or handing it the parts that matter, this is written for you.
I have been at it since the first day, mostly for free, mostly in front of people who were not ready. I am not going to sell you urgency. I already know how that ends.
[Newsletter invitation, one line.]
Before this page goes live
Two brackets. The original ChatGPT answer from 30 November 2022, if it can be recovered. The reconstructed version is currently marked as reconstructed, which is honest, and it stays marked unless the real one turns up. And the newsletter line.
Two permission items. The friend who received five days of OSHA preparation is unnamed here. He should hear about this before he reads it, and if he agrees the name goes back in. The nonprofit is anonymous until the engagement is finished and they have agreed to be named.
One structural note. The CRM story and the Sunday call are both told properly in I started with two people, not a product. They are compressed here on purpose. Do not let them grow back.