I was not sure it would work. I did it anyway.

Learning by Doing

When I start something I’m never sure how it will work, but that’s how I learn. I learn by doing, and today I want to talk about doing things before you’re fully prepared and how AI helps a lot in this process.

The Gym Script

Last week I went to the gym where I practice jiu‑jitsu, not to train but to build a script. The script was needed because the member data was locked behind a turnstile system that only worked on the gym’s internal network. I had to be there in person to access it.

I’d read a piece by Shane Parrish about probabilistic thinking, and one line stuck with me: you’ll never have full certainty up front, so you only need a bet good enough to move. Most people wait to be sure before they start. For me it was the first time I would write a script like that while the client was watching me.

I knew the data lived inside the turnstile and could only be reached on their LAN. Two days earlier I had interviewed the staff, and they showed me how the system worked. I walked away with a vague idea of the process and bet on that vague idea, trusting that I could figure out the details. Even though I’d never done this before, I usually experiment, and if I fail I learn from the process.

Why Manual Was Impossible

Doing it by hand would have been almost impossible. Exporting the data manually would require dozens of exports each day to stay near real‑time. Simulating a person clicking through the interface would be repetitive and error‑prone. It would have taken hours, maybe days, to code a solution by hand.

AI Accelerates the Build

I turned to AI. I gave it a plan, and it understood how each request to the system worked, then wired the bridge for me. What would have taken me a couple of hours to code took under an hour with AI, even though it was my first time using it for this kind of task. AI worked in ways I hadn’t predicted, and it made the whole process much faster.

Cheap Mistakes

These days a mistake is cheap. You might lose a little time or a bit of money, but AI makes learning fast, so the cost of trying is almost nothing. Because a wrong attempt is cheap and quick, you can afford to bet that you’ll figure it out.

If I had to travel to another city to try this, the cost of being wrong would have been much higher. In my case the gym was a ten‑minute walk from my house, and I went with my brother on his motorcycle. I took my notebook with me, so I could search for solutions on the spot. There was no problem being wrong that day.

Skill of Improvisation

I’ve been developing the skill of pulling things off without being fully prepared for years. It’s not specific to a gym turnstile; I’ve done similar things in other contexts. The point I want to make is simple: you don’t need to be sure. You just need to start, and then the learning begins. AI just amplifies that learning, as I always say.

Closing

The article that sparked this video is linked in the description. I’m recording this minutes before heading to mass, which starts in ten minutes. The church is close to my house. I should have recorded this yesterday, but I was too tired and then had insomnia, so I got almost nothing done. In the afternoon I rested, got motivated, and decided to record now.

That’s it for today. See you in the next video.