It's already been a little over a year since I joined Coxwave. A few days ago, I decided to move into the 'Engineering Manager' role. Since I was changing roles anyway, I started slowly looking back over the past year — and the scene that came to mind first was, unexpectedly, a weekday last December.

Early December 2025. One day, I finished in two days what I had thought would take me two weeks. To be precise, I didn't really finish it. I had to quickly build a web app for a demo, and most of the code was written by Claude Code (Opus 4.5).

For a while now, every time I worked on the frontend, I'd been stacking up session retrospectives, jotting down my own judgment criteria one line at a time. As those notes piled up, I think at some point I started feeling that there wasn't much of a gap between the code I wrote and the code AI wrote. When I had AI review my code, the heavy-handed touch-ups grew rarer; conversely, when I reviewed code AI had written, the cases where I just hit the approve button without much editing kept increasing.

But that day was different. I had come to terms with the fact that the texture of the code felt similar, but I hadn't quite thought about the speed. I watched, with my own eyes, work that would clearly have taken me two weeks in the pre-AI era wrap up in two days. I think the words "This actually works?" surfaced in my head just once. The overwhelming sense of speed I felt that day. The day I knew I'd meet someday was being shown to me right in front of my eyes. I think reality kind of hit me right then.

That evening, I stopped by a wine shop near the office and brought home a bottle of port. I sat down, sipped the wine, and quietly watched AI churn out the code. Someone once said there's a kind of romance in that scene.

After that day, daily life went on as usual. But somewhere along the way, the center of gravity in my head had shifted to AI first. I shared with the team the frontend engineer persona I'd been shaping all along, and together we set up an environment where other engineers could borrow that persona and try developing full-stack as well.

And then, while I was building rapport with AI and trying to slowly push productivity up, I received an offer from the CEO to take on team leading.

Team leading isn't actually new to me. At Swingvy, my role was frontend engineer, but as a co-founder I naturally ended up taking on team management and lead duties together. Back then, I think I came to feel deeply just how hard managing people really is, and how much mental energy it demands.

After I left that company, I wanted to be free. For a while I worked as a freelancer and just wrote code. The time I spent purely focused on development — looking back now, I think it was a really free time. Then, with the thought that I needed to learn about AI, I joined Coxwave, and a little over a year has passed since.

When I got the offer, I was, unexpectedly, calm. The first thought that came to me was: here it is, the thing that was bound to come. Every time I looked into the team's mood and where each member sat, I had this feeling that someone needed to take on the lead role — and I had been wondering whether I should draw on my past experience and step into the leading position myself.

I'm feeling the weight of it. I know what leading carries with it. Still, since I wish for every team member I've crossed paths with to do well, I'm trying to take on the mindset of "let's lead this well, just once." At times like this, there's a thought I always come back to: "It'll work out somehow." When I just focus on the present, things have always somehow ended up okay, and I think it'll be the same this time too.