Week Note 34 | Joohee Kim

Aug 17–23, 2026

My left arm has been feeling unusually sensitive. Maybe because I’ve been overworking it, I get tired easily and feel a faint pain. I decided to keep an eye on it for a little longer and go to the hospital if it starts happening more often. In fact, I may have surgery to remove the metal rod sometime between mid-September and early October. The date hasn’t been set yet, but it will probably happen by October at the latest. So I’m being as careful as possible and trying to avoid demanding schedules or too many plans. Maybe that’s why I’ve been paying even closer attention to how my arm feels. I really hope it doesn’t get worse…

Earlier this week, I used up all my tokens and ended up writing code by hand. It was a simple task to adjust the height ratio at the top, but I enjoyed it. It reminded me of when I used to code everything myself. I think I once read that writing code by hand might become a hobby, or something reserved for people who are really good at coding. Maybe that really will happen. People who say they built an app entirely by hand seem impressive, and part of me wants to follow them. But in reality, I still need to deliver results quickly. I’m still figuring out where I stand. For now, I’m in a situation where I need to use AI to produce results, so I’m trying to spend tokens efficiently and only where they’re needed. I’ve become more efficient with tokens than before, and I’m able to work on the things I want to do in a reasonable way.

I finished fixing the bug that had been tormenting me, but the code review is still ahead. Part of me wants to trust that the AI-generated code is good enough and merge it immediately. But the more I review it, the more opportunities for improvement I see. It’s surprising that I can spot those things, and it also makes me wonder whether I should have instructed the AI better from the beginning. I had already written a spec document together with the AI before asking it to write the code, so the result was reasonably solid. Still, as I looked at and understood the code more deeply during the review, I noticed things I hadn’t seen before. Some of the changes I asked the AI to make were simple, such as renaming variables for readability. Others involved removing unnecessary extensibility or trimming away over-engineered parts. Those issues became visible once I approached the code with the mindset that I would be responsible for maintaining it. Maybe you could call it coding style. Some of this may have happened because I haven’t yet written those standards down as explicit rules. I think I can reduce these problems by refining them later and incorporating them into the initial planning stage, which is what I consider the spec-writing stage.

I’m still making changes as I go through the code review. There are more code changes than I expected, so checking them one by one is taking time. But I want to finish it properly. I’m eager to show the result soon, but I also want to see it through carefully. I’m finishing with the goal of releasing it within this month.

On August 18, August’s performance surpassed the entire month of July. In just 18 days, the number of downloads and sales revenue had already exceeded July’s totals. July was already the app’s third-best month since launch, so I’m starting to wonder whether August might even outperform October and November, the months when revenue was particularly strong. I’m excited, but also anxious. Will this momentum continue?

Things are a little difficult in real life right now, so I’m holding on and getting through it. Meanwhile, my small, precious app seems to be saying, “I’m getting better and better.” I want to recover quickly so I can help it grow even more. The fact that I still have things I want to do feels almost like a small miracle.

My week…

Listened to:

Watched:

Currently reading:

「그만 배우기의 기술」 (finished)
「요즘 당근 AI 개발」