I Built a Fully Automated AI Agency. Then I Killed It.

It worked. That was the problem.

I am my own only employee.

That sentence is the whole problem with my business, and if you work for yourself, it's probably the whole problem with yours. I can only edit so many videos in a day. Shoots eat full days, sometimes several, sometimes in another city. A single standing Monday meeting somehow converts an entire day into half a day — I still can't fully explain the math on that one, but I've measured it enough times to know it's real.

And underneath all of it runs the equation every freelancer knows by heart: if you make $10,000 on a video but it takes you 10,000 hours, you've made a dollar an hour. Time is the business. The work is almost incidental.

Which leads to the catch-22 I was living in: I need more clients to grow. But I can't go find clients, because I'm busy doing client work. I can't afford to hire — not overseas, not locally, there's just no room for another salary. It's me.

So last year I did what any reasonable person drowning in logistics would do in 2025: I tried to automate my way out.

The machine

I've spent my career making content for big brands — the adidas and Walmart tier of the world. But I've always thought there was an opening at the other end of the market: local businesses. The HVAC company. The dance studio. The pest control guy. Nobody's dream clients, which is exactly why there are thousands of them going unserved. They're a lost niche.

So I built an agency for them. An automated one.

And here's the part of this story you're not expecting, because every "I quit my AI business" essay is secretly about a business that didn't work:

It worked.

I built a system that could generate content nearly automatically from a business's existing brand — their style guide, their website, the content they already had. Static posts. Carousels. Short Reels. Even talking-head videos fronted by an AI avatar, so a busy business owner would never have to sit in front of a camera — which, it turns out, is the last thing most of them want to do anyway.

It functioned. I started promoting it. The era was booming. The machine was ready.

The justification spiral

Then something happened that I can only describe as my own brain refusing to cash the check.

I tried to talk myself into it. I'm good at that — I justify things to myself all the time. The internal pitch went something like: I don't even like social media. Why not fill it with junk, if junk is what people want? Maybe AI content floods the whole thing, social media eats itself and dies, and it's a win-win — I make money during the boom, and then the thing I never liked goes away forever.

I want to be honest with you: for about a week, I thought that was a great idea.

But at the end of the day, I couldn't get myself to do it. I don't want to be peddling garbage content onto the internet. Not because of a manifesto. Because it didn't feel good. That's the entire reason. There's no spreadsheet in this story.

So I built the whole thing — and killed it, pretty much instantly.

What I kept

Here's why I don't regret a minute of it: the devil you know.

I know, firsthand, what's possible now — not from a keynote or a YouTube demo, but because I built it with my own hands. I understand what's coming for my business, what I actually need to prepare for, and — this is the part that surprised me — which pieces of this technology are genuinely helpful to the business I already have, rather than to some new business that would have quietly destroyed my soul.

Because some tools survived the purge. A handful of them are doing real work in my shop right now — not making content for me, but clearing the logistics out of the way so I can do the work myself. That distinction turns out to be everything, and it's what this newsletter is about now.

What's changing here

Back in June I wrote about vertical video breaking my brain — how I resisted a format shift, got burned, and swore I wouldn't sit the next one out. Consider that chapter one. This is chapter two, and there are a lot more coming, because I'm making a promise I haven't made before:

The Creative Economist is going weekly. Every Tuesday.

Same subject it's always been — the craft and business of being a working creative — with one sharpened edge: I'm going to document, honestly, what it looks like for a working filmmaker to use AI without becoming a cheerleader for it or surrendering anything that matters. The wins with receipts. The failures with receipts. The parts I refuse to automate, and why.

Not "AI will change everything." Not "AI is slop." The missing middle — from someone who built the machine, looked at it, and chose the craft instead.

If you've got an AI tool you can't decide whether you hate — hit reply and tell me about it. Genuinely. The best of these are going to become issues.

See you Tuesday.

— Alex

P.S. — Next Tuesday's issue is already written, because I'm spending this week in Portland with my family. Phone mostly off, laptop at home. I'm aware of the comedy of announcing a weekly newsletter and immediately leaving town — but building a business where I can leave town is kind of the whole point. See you in seven days either way.