I’m not building a chatbot.

I’m not building something that needs to be prompted constantly to be useful. If a system requires me to restate my goals, my context, or my history every time, that’s not intelligence — it’s a loop.

I’m not building for demos. Short-term impressiveness is easy. Persistence is harder. Anything that looks good in a five-minute screen recording but degrades over weeks isn’t interesting to me.

I’m not building tools that maximize output at the expense of thinking. Speed is not the same as clarity. More text is not the same as understanding.

I’m not building products optimized for engagement, virality, or novelty. If a system needs attention to justify its existence, it probably doesn’t belong in the background of someone’s life.

I’m not building something that resets. Memory loss as a default is a design failure, not a feature. Systems should carry context forward, even if imperfectly.

I’m not building a solution for everyone. Generality is often a way to avoid making hard decisions about what matters.

I’m not building on short timelines. Anything meaningful here will take time to understand, time to shape, and time to get wrong.

This list will change.
Some things here may turn out to be wrong.

For now, these are the boundaries.