I noticed this three days ago. I was mid-feature on my Agentic Development Environment Classis when I realized it had crossed a threshold. It could be used to develop itself. Not perfectly, but it could. I’d spent almost four weeks poring over every feature and edge case, and somewhere in that stretch the tool had become capable enough.
And instead of triumph, I felt something closer to dread.
Not because Classis isn’t good. It’s good. I’m having a ball building it. The kind of deep-focus month where you’re still thinking about architecture in the shower, where every solved problem feels like cracking a safe. This is the real thing, and I know it, which makes the next sentence harder to write:
I have no idea what I’ll build with it once it’s done.
Another month of polishing, maybe six weeks, and Classis ships. Then I’ll hold a powerful developer tool, and think: okay, now what?
Build a SaaS startup. With what idea? I didn’t build this to solve a specific problem. I built it because building it was interesting.
Contribute to open source. Maybe. But that’s finding excuses to use a hammer you forged yourself, not the same as having a project that needs it.
Just keep improving the tool. This one’s dangerous. I could spend another year on Classis. There’s always one more feature to add, one more layer of polish. And every week spent improving the tool is a week not using it for anything except improving the tool.
You fall in love with the process of making. Not the thing you’re making. The making itself. The 2am debugging sessions where a problem finally cracks open. The architectural decisions that click into place and make you mutter “yeah, that’s right” at an empty room.
I’ve been here before. Most developers have. You finish a project, the high drains out, and you’re staring at a repository thinking that was fun, what now? The difference this time is scale. A month of focused work. Real opportunity cost.
So the question keeping me up isn’t is this tool any good?
It’s: Am I just developing for the sake of developing?
What if I’m not someone who builds tools to make things? What if I just love the act of building so much that I’ve convinced myself I need a reason for it?
Part of me says: that’s fine. Building for the joy of building is honest in a way that building for metrics never is. One month crafting something, learning a lot along the way, and I’m the only user. I am ok with that.
Another part says: cop-out. That’s the story you tell when you’re afraid to ship, afraid to put work in front of people who might not care. “I build for the love of building” sounds noble until you notice it also protects you from judgment.
Both parts are right, I think.
In four or five weeks Classis will be polished enough to hand to another developer and say “here, try this” without cringing. I don’t know what happens after that sentence.
Maybe the right problem appears and the tool turns out to be exactly what the problem needed. Maybe this whole spiral is just pre-launch nerves, the feeling before any release when you’ve invested enough that of course you’re questioning whether it matters.
Or maybe I finish, sit in front of it for a week, and realize the best part is already over. Building done. Fun done. And I have to decide: find something new to build, or figure out what “done” feels like when you’ve been running toward it for two months.
I’ll let you know which one.
Assuming I figure it out myself.