I Built 26 Products. The One That Worked Rents Washing Machines.

Seventy-three repositories. Twenty-six distinct products. Thirty-four months.

That's my GitHub since August 2023. I pulled the numbers expecting to feel proud. Mostly I felt caught.

Thirty-one of those repositories were the same idea wearing two names — PodPulse, then PostMachine — turning podcast audio into transcripts, show notes, and posts. I rebuilt it thirty-one times. Five more were a homeschooling app. Three were AI virtual staging for real estate. Three read faces. I finished most of them. I sold almost none of them.

Here's the part that took me three years to see: the building was never the problem.

The exploration wasn't wasted. It was tuition.

I don't regret a single repo. Each one taught me a tool, a model, an API, a way the pieces fit together. You don't learn that from a course. You learn it by shipping something that breaks and fixing it at 1 a.m.

An AI model is a malleable baby with the reach of the entire internet — it builds anything you can describe in enough detail. The model matters; anyone who tells you they're all interchangeable hasn't pushed one to its edge. But a model with no context is a stranger doing your job off a one-line brief. The leverage is everything you feed it about how you think — your taste, your brand, the way you want a customer to feel.

I finally wised up about that part: every repo's context now lives in an Obsidian vault, my philosophy encoded as skills the model recalls on command, updated as I read and change my mind. A knowledge base that compounds instead of resetting to zero every session.

And it taught me something I didn't expect: I'd solved the context problem and never touched the customer problem. They were never the same problem. I spent three years hoping they were.

The grail isn't just for engineers anymore

In Q1 2026, worldwide app releases jumped 60% year over year. Apple's App Store alone saw an 84% surge in new submissions — its first real spike since 2016. Collins named "vibe coding" its 2025 word of the year. Most of the people riding that wave never wrote a line of code.

I was early to it — shipping since 2023, which means I hit the lesson on the far side of the wall before most people reached it.

In Zero to One, Peter Thiel named the trap years ago, in a chapter called "If You Build It, Will They Come?"

"The engineer's grail," he wrote, "is a product great enough that 'it sells itself.' But anyone who would actually say this about a real product must be lying."

That used to be an engineer's private delusion. AI handed it to the rest of us. I'm not an engineer — I'm a determined AI enthusiast with a hunger to build I still can't fully explain — and I inherited the grail along with the ability to act on it. I believed it completely: make the product good enough — safe, polished, actually useful — and the selling takes care of itself.

Thiel's hammer: "If you've invented something new but you haven't invented an effective way to sell it, you have a bad business — no matter how good the product."

Twenty-six good products. Twenty-six bad businesses.

Then I sold washing machines

Three years in, I did the one thing I'd never done: I sold it before I built it.

Koerner's rule is to put the offer out before the product exists and let the response tell you what to make. For three years I'd run the opposite play — build first, then go hunting for someone who wanted it. This time I posted the offer before I owned a single machine. The replies came back: real people ready to pay for a washer and dryer in their unit today, not a clever app they didn't know they wanted. That traction is what compelled me to build. Demand first, product second. After twenty-six products built in the dark, doing it in the right order felt like cheating.

So I bought the machines. I built a dashboard to track every unit, every customer, every payment — and running that database taught me more about business in three months than twenty-six MVPs did in three years.

It's not sexy. There's no conference talk in it. It clears about $900 a month — but it's the first thing I've built with real customers paying real money on a schedule. It worked for the exact reason the software didn't: I sold to people who already needed it. Koerner says it flat — sell to people who already buy. I'd heard it a hundred times. A washer-and-dryer rental is the first time I did it.

He also preaches cash flow over paper wins. Nine hundred real dollars a month rearranged my sense of what counts.

The podcasts gave me the nerve, not the knowledge

I didn't learn this from books. I learned it from podcasts I lived inside — My First Million, Chris Koerner's Koerner Office, Greg Isenberg's Startup Ideas, David Senra's Founders. Hundreds of hours.

They didn't teach me how to sell — I'm still bad at that, still learning. What they gave me was the nerve to try, and the permission to ship something before it was finished. Every founder in those episodes shipped things that weren't fully there, on purpose, to find out if the thing was worth finishing at all. That reframed building from a performance into a test. The Koerner Office went further and handed me the theory: sell first, build second.

The one thing none of them could give me is the part that's mine. In The War of Art, Steven Pressfield names the force that stops you from doing the work "Resistance," and gives one rule for finding where it matters most: "The more important a call or action is to our soul's evolution, the more Resistance we will feel toward pursuing it."

I feel no Resistance toward building. I've done it twenty-six times — building is my comfort, my hiding spot. All of my Resistance points at selling. By Pressfield's rule, that's not a weakness to route around. It's an arrow. The thing I flinch from is the thing I'm supposed to walk toward.

What I'd tell you, and myself

Starting is hard. Everyone says so. Nobody says the harder part out loud: continuing — grinding through the boring middle of one thing instead of starting a shinier new thing — is the actual skill. I'm only learning it now, at repo seventy-four.

If you've never shipped anything, the fear shows up as "I don't know where to start." If you've shipped twenty-six things, it shows up as "I don't know how to sell." Same fear, two outfits. The cure is the same: do the thing you avoid, badly, this week.

I'd give one of my skills away free to get one person off zero. That's not a marketing line — it's why I built the storefront. I'd rather you do something with a tool than admire it.

Seventy-three repositories taught me what no book could: building was never the wall. The wall was everything I kept telling myself I'd handle later.

Handle it now. Then don't stop.