I've been working on a simpler, more extensible Markdown variant I'm calling Zd. I've also been working on a simple, embeddable language that's something in between Lua, Koka, and Erlang.

In both of these projects, I'm trying my best to find a balance on AI usage. So far, they both have held strongly to the principle of only accepting code I've written personally. At the same time, as I begin to land on stable design decisions I've found that speculative experiments with AI can be helpful. I can benchmark potential future decisions prior to committing.

I'm not saying that AI couldn't be used to do more, but I'm trying to ride a line where things still feel personal, local, and deeply my own design. And more than that, I'm practicing holding on to the joy in the work. It's too easy to just know that AI could be used to plow through a problem. It makes my scope tend to creep and my estimates shrink. So, pushing back involves continuing to scope small changes and remembering to feel positive about each win.