This is not a case study. This is my story with AI.

I'm from Lamesa, Texas. If you've never heard of it, don't feel bad. Most of Texas hasn't.

It's a low-ceiling, big-sky kind of town. I left because I wanted to build something, and the most exciting new business in Lamesa that decade was a second Dollar General.

I didn't leave with a plan. I left with a chip. A chip is decent fuel if you don't let it burn the house down.

Devslopes was a coding school. Online. Seven years on paper, three years of real revenue, and the first stretch was me, a laptop, and a desk that was also where we ate.

Every coding school on earth was saying “we teach coding better.” So was I. Saying it louder did not make it more true.

Then I changed three words.

“Earn While You Learn.”

Instead of selling the classroom, we sold the job site. Students did real freelance work for real clients on a real marketplace while they were still students. Portfolio before diploma. Same product. Different shelf in the brain.

In 2019 we did about $200,000. It felt like the Super Bowl and the lottery on the same afternoon.

Here's the revenue line, because I like a scoreboard.

2024 was our best year ever. In 2025, we did a couple of million, then a flatline. The kind you hear on a hospital show.

Somewhere on that climb I got a mentor. Patrick Bet-David, the guy who built PHP Agency into a giant and then built Valuetainment on top of it. Not a podcast guest. A close mentor, the kind who asks one question and your whole plan has a hole in it. I learned more from him about how people actually make decisions than from every business book on my shelf combined. Some of what's on this page is his fault.

Now here's the part that doesn't fit on a scoreboard, and it's the part that made us who we are.

While the top line was doubling, the money actually reaching us was shrinking. Coding schools run on financing. A student borrows tuition from a third-party lender, the lender pays the school, the student pays the lender back over time. When we started, the lender paid us about 90% of what the student financed. By the end, it averaged as low as 50%.

Then August 2023 happened.

One month, 49% of the students who applied for financing got approved. The next month, 26%. No warning. No phase-in. Half our pipeline vanished between one Monday and the next.

So here was the situation. Half as many approved students. Half as much money per student. A forty-person payroll. A lead queue that was already too slow.

I had two choices. Shrink the company to fit the new math, or find a way to do more with a lot less.

I'm not built for the first one. I'd already burned the boats. Cortés landed in Mexico with a few hundred men, sank his own ships so nobody could sail home, and told them the only direction left was inland. That's not a metaphor I picked for this page. It's how it felt. There was no Lamesa to go back to. There was only forward.

And forward, it turned out, was software.

Here's where this page earns its name.

When OpenAI opened up their API, I didn't read a think piece about it. I called my best friend. He's a genius software engineer, the kind who reads documentation for fun, and I'm the kind who reads it because I have to. Good pairing.

We had a problem worth solving, and after August 2023 it was a problem we had to solve. Devslopes was running paid ads at scale. Thousands of leads a month. Every lead was going stale in a queue, waiting for a human to text them, qualify them, and book a call. And now every approved lead was worth twice as much, and we had half as many. Speed to lead wasn't a nice-to-have anymore. It was oxygen.

So we built what I'd now call the first pure agentic texting system I'd ever seen. Back then it didn't have a name. We just called it the bot, and then we stopped calling it anything because it became part of the furniture.

It texted every single lead. Instantly. At 2 a.m. On Christmas. It qualified them, answered the “is this legit” question for the ten thousandth time without getting tired, and handed our closers warm, booked conversations instead of a spreadsheet of maybes.

No rep on earth could touch it for speed. And it never had a bad Tuesday.

This is the part I want to be careful with, because it gets told wrong.

The system replaced the work of fifteen-plus employees. It did not replace fifteen people.

Those were setters, qualifiers, follow-up reps. Good at their jobs, and their jobs were now being done faster by software. So we moved them. Onto closing. Onto student success. Onto the parts of the company where a human being with judgment was the whole point, and where one good hour was worth ten hours of texting “hey, just following up.”

That's the year after the approval rate got cut in half. With the bank taking a bigger cut of every dollar. Same size team, roughly. Different math.

Thousands of hours a year, gone from busywork and reinvested in the stuff that actually needed a human.

Once the texting system worked, we couldn't stop.

The product got better. Cash flow got healthier. And the people who worked there got to do work they didn't hate. That last one was my favorite.

I want to be clear about one thing, because “AI guy” can sound like someone who's never had to make payroll.

I didn't just build a sales team. I hired and fired well over sixty sales reps. I hired fifteen-plus virtual assistants across every operations need we had. I hired the entire executive team: Director of Operations, VP of Finance, VP of Sales, VP of Marketing, Director of Paid Media, Director of Education. And I hired the ground floor too, all the way down to the student ambassadors who answered the “is this real” DMs.

Every role, sales to on-the-ground operations. Hired it, trained it, and in a lot of cases, replaced it and hired it again.

Here's what all that hiring taught me, and it's the thesis of this company:

AI can be implemented at every level of a business. Not just sales. Every level.

And when it's done right, it improves quality of life for the people in the seats, it improves productivity for the company, and it lets you go further, faster, with less.

Less headcount doing busywork. Less cash burned on hours nobody wanted to work. Less of you stuck in the weeds.

I know that because I ran it, at every level, with my own people.

Now the part where a normal founder page cuts to a testimonial.

You already know the bloodstream was somebody else's bank. In 2025, that bank collapsed. Not a slow wind-down. The bridge was there on Tuesday and gone on Wednesday. We were midway through state licensing in two states, compliance firm hired, doing the grown-up things. None of it mattered, because the money that made enrollment possible stopped existing.

October 2025. The call. Forty-plus people.

I read The Five Temptations of a CEO afterward, which is like reading the instructions after the bookshelf falls on you. The temptation I fell for was status over accountability. I liked being the guy who built the thing. When the numbers needed a harder look, I didn't always take it. The bank was the cause of death. That part was mine.