going all-in on AI
April 2026 was the month I stopped lying to myself.
I had been watching my friends move ahead while I stayed still. Some of them were already running agents to mine AWP and pulling real profit with almost no manual work. I wasn’t against the idea. I just kept putting it off. I told myself I would learn it when I felt ready. But that moment never came on its own. The longer I waited, the more obvious it became that I was choosing to stay behind.
I regretted it. Not in a dramatic way, but in a quiet, uncomfortable way. I knew I was falling behind and I still didn’t feel like doing anything about it. There was this strange comfort in not learning. Like I had accepted that I would just get the scraps while other people took the real opportunities. That part felt pathetic. I wasn’t curious. I wasn’t even angry at myself yet. I was just… fine with being left behind. And that felt worse than actually failing.
What finally pushed me was the same feeling that had kept me still for so long. I didn’t want to keep getting the leftovers. So I started with Claude Code. Sonnet first, then Opus. I paid the $17 just to see what it felt like to have something that could actually do work for me. A lot of people get overwhelmed just looking at setup tutorials. I get it. But with Claude Code, I could tell it what I wanted and it would handle the installation on my VPS. That simplicity made it easier to start.
Still, the first thing that stopped me was the money.
I didn’t want to spend on something I didn’t understand yet. What if it didn’t work out? What if I just wasted it? That hesitation lasted longer than it should have. Once I actually started using it and slowly understood how it worked, everything felt much simpler. But getting past that initial fear of spending without knowing took time.
I’ve used computers since I was a kid, but mostly just for games. So when I started dealing with agents and servers, a lot of it felt foreign. I wasn’t completely clueless, but I also wasn’t comfortable. The hardest part wasn’t asking questions. My friends were open to helping. The hardest part was understanding how the models actually worked underneath. You can buy the tools, but you can’t buy the understanding. Without that, you end up slowing the agent down instead of using it properly.
My motivation was never noble. I just didn’t want to keep watching other people move faster than me.
There’s a selfish part of me that hopes this advantage stays uneven for a while. I know how it sounds. In this economy, thinking like that feels somewhat normal. I wanted to get ahead first before the gap closed. Once everyone figures out how to use these tools properly, the edge disappears. I wanted to use that window while it was still open.
The feeling was different from when I used to trade.
Back then, I started curious but eventually became anxious. There was always this low-level tension. With AI, I’m still mostly curious. I want to see how far I can push it. I know that could change. Everything shifts eventually. For now, I’m just letting it flow and seeing where it goes.
I see a lot of real coders on X being pretty harsh toward vibecoders. Some of the criticism makes sense, especially when it’s about security or bad practices. I actually appreciate that. But some of it feels unnecessarily destructive. Everyone starts somewhere. I respect people who’ve been doing this properly for years, but I also respect the ones who are serious about learning even if they’re still rough around the edges.
I didn’t quit trading completely. I just stopped making it the center of everything. My focus moved toward building small things that could actually help me. Wallet management, minting tools, simple automations. Nothing big. Just things that would make my day easier and maybe generate some return.
I don’t have a big vision for where this will take me. I just want to keep getting better at using these tools so I don’t have to rely on other people as much. That’s enough for now.