Why I build in public

I've always believed that actions speak louder than words — so I put my code where my mouth is. This is the first post on a small corner of the internet I control, and it felt right to start with why I'm writing at all.

Shipping beats talking

Most of what I know, I learned by building something, breaking it, and fixing it in front of other people. Open source made that possible: every pull request is a tiny public record of what I understood on a given day, and every issue is a reminder of what I didn't.

Writing here is an extension of the same idea. Instead of polished retrospectives written months later, I want to capture decisions while they're still warm — the trade-off I made, the thing I got wrong, the approach I'd repeat.

What I'm working on

Right now most of my energy goes into voxton, along with a rotating set of open-source and AI tooling experiments. A few themes keep showing up:

• AI as a building block, not a bolt-on. The interesting problems are in the seams — retrieval, evaluation, and the boring plumbing that makes a model actually useful.
• Developer tools that respect your time. If a tool needs a tutorial to do the obvious thing, the tool is wrong.
• Small, sharp, finished. I'd rather ship one thing that works than five that almost do.

What to expect here

Short, specific posts. Some will be build logs for voxton and other projects. Some will be things I wish someone had written down before I spent a weekend learning them the hard way. None of them will pretend the process was cleaner than it was.

If any of that is useful to you, good. If it just keeps me honest, that's enough.

— Akshay