Aug 2–9, 2026
I released the final external testing build, and I’m now preparing the internal testing build. Things are moving more slowly than I expected because responding to different situations takes a lot of time. I’m adding and refining the planned features while testing, and I keep finding more issues along the way. Using the app I built myself lets me discover and fix bugs firsthand. I probably would have missed them if I didn’t use my own app regularly.
I started using Codex GPT 5.6 Luna after hearing that its token efficiency had improved significantly. It definitely uses far fewer tokens. I’m still on the Plus plan, so I’m trying to make efficient use of Luna Max. I think learning how to use AI tokens efficiently will become increasingly important.
Even in an age when AI is used in almost every field, I still believe I should keep writing and reading. I feel that something truly becomes mine only when I use my own mind and think it through. AI can help me create an outcome, but I’m still responsible for that outcome. That means I need to understand it, learn more, and write better prompts to get the results I want.
In something I read recently, I came across the expression “sunk cost, process capital,” and I liked it.
These days, I like the phrase process asset more than sunk cost. Preparing for jobs for a long time and failing. Building a portfolio and getting no attention. … Spending an entire day on an Instagram magazine with fewer than 100 followers and getting no likes or follows. Days like that easily become pathetic days. It feels like time was wasted, there were no results, everyone else is moving forward, and I alone spent energy in a strange place.
But whether that day remains only a sunk cost or becomes a process asset is ultimately a matter of interpretation.
If you spent the whole day thinking about content and received no response, that can be a failure. … But at the same time, it can be the day you learned which sentence was not read. It can be the day you learned which thumbnail failed to make people stop. … It can also be the day you confirmed which topic you can hold onto longer than you expected.
I’m continuing even while feeling uncertain about whether what I’m doing is right. I want to believe that it will help me someday, and I’ll try to leave this time behind as process capital rather than sunk cost. Fortunately, there are still more things I can do.