Build the Timeline, Not the Checkbox

This is my personal view on product development, and it really changes from person to person. Some people go documentation first. Others build first and document second. There are many approaches, and with AI we can do a lot more of them than before.

I personally like Claude Code. It feels like a genuinely new kind of tool, not just sitting inside another ecosystem. We already know what we should do, and Claude Code goes and does it for us. The worker is Claude Code, and it largely directs itself. That is the feeling I keep coming back to.

Tools like Antigravity, Lovable, and the rest are fine for starting something quickly and testing it with customers. But I still feel they are not the right fit for larger applications. They may have their own strengths around security and so on, yet I believe keeping control of everything inside your own application matters a lot. I come from an engineering background, and I am still not sure that sitting on someone else's platform and running a business on top of it is a good long-term approach.

Owning your own infrastructure is genuinely hard. Keeping up with security, updates, and uptime is a lot of work. But I think this is exactly where AI should make things reliable, and we can already do it today. A lot is still on the way, and I am really optimistic about where AI is heading for these engineering workflows. We do not need to do the repetitive tasks anymore.

So my conclusion is that product development should follow a controllable approach. First you build what you actually need, not whatever the platform happens to offer. If the platform gives you a checkbox but the job needs a timeline, then you build the timeline. I want that empty canvas feeling, where nothing is decided for me. That is exactly why I loved Figma in the first place.