For a long time, building software felt like a thing I was doing in secret.

I was a physical therapist. I saw patients all day, then stayed late to build little tools that made the hospital’s busywork hurt less. Nobody assigned it. Nobody could help me with it. I had no one to ask. I just kept poking at problems alone, in the gaps between my real job.

Then one day, in a city called Andong, two things happened at once. Both of them changed everything.

Meeting the person on the other end of the phone

There was a software founder I’d gotten to know — until then, only a voice on the phone. As it happened, he had a golf outing near my hospital, and he stopped by on the way. So there we were, sitting across a meeting table for the first time.

I opened up one of my tools. It was an inventory search I’d built for the hospital — hundreds of thousands of supply items, findable in a single search. A few days earlier I had finally fought my way through setting up outside access, so the screen loaded fine, even on the meeting room WiFi. I showed him.

Then he showed me his work. He ran a distribution company, and he had used AI to automate the product photos and marketing images his design team used to make one by one in Photoshop. Now anyone at the company could press a few buttons and get something good.

It was a strange feeling. We were doing the exact same thing. I had put real work into the hands of ordinary staff. He had put design into the hands of ordinary staff. Different companies, different jobs, but the same idea: take something only an expert could do, and open it up to everyone.

The one sentence I still carry

That day he said something that stuck with me.

He told me, more or less: assume you can build almost anything. Good open source is out there for free. Don’t fold before you start. Just begin as if it’ll work, and as you go, it usually does.

That was the first time I’d heard the words “open source.” Somebody, somewhere, builds useful pieces of software and gives them away — pieces sitting there waiting to be used. Half the things I had been grinding out from scratch, alone, already existed in some form.

But the part that moved me wasn’t the technology. It was the attitude. Don’t fold before you start. That was already how I lived, in a quieter way — try it, and if it works, great; if it doesn’t, I learned something. He just said it bigger and clearer than I’d ever let myself say it.

Seeing how far the road could go

Here’s the thing that gave me the most courage: he wasn’t a developer either.

He had never learned to code. And yet this non-developer had built an entire AI research effort inside his own company. He had even built his own custom working tools. (I tried to copy that part later and gave up fast. That’s fine. You don’t have to copy everything.)

Up to that point I’d been completely alone. No one beside me on the same path. And now I was looking at someone who had started from the same place I did and gone much, much further. So there was something at the end of this road. I could get there too.

If you’re building something on your own and feel like the only one doing it, find one person who started where you are and went further. You don’t need a mentor or a teacher. You just need proof the road leads somewhere.

A new desk, and a real weapon

The same day, something else happened.

A director called me in and asked if I could help out in the planning office — the part of the organization that works for the whole foundation, not just one therapy room. There was no formal transfer, no ceremony. I just started showing up there.

But the meaning wasn’t small. Until then, I had built everything on stolen time, after hours, as a side thing I did because I liked it. Now it was becoming work I was allowed to do out in the open. The day my tools changed, my place started to change too.

So I fixed up how I worked. The founder had told me to use Claude in the terminal — that black, blinking screen — because it was more powerful. But that black screen scared me. I barely understood computers. So I found a middle path: VS Code, which looks a lot like Cursor but keeps the power. It felt familiar, and it stuck to my hands fast.

That was when I finally had a real weapon. Looking back, this is where the serious building actually began. Someone who had started with nothing but a few stubborn questions now had a proper tool and a proper desk.

Next came the biggest thing I’d ever attempt: pulling everything I’d scattered across the hospital into one place. But that’s the next story.