Why these notes exist

Sapix is a personal AI substrate. Single-person, sovereign, running on one machine: mine. By design it has almost no audience. So why keep notes in the open?

Because the interesting part was never the code. It was the decisions. Why memory should freeze instead of delete. Why the answer to a smarter assistant is not a bigger model but a sense of when to recall. Why an amplifier beats an autonomous agent for someone who actually likes to think. Those choices do not show up in a diff, and they are the part worth keeping.

What goes here

Field notes, not documentation. Short, technical, honest about the dead ends. When a decision turns out wrong, that note stays up with the correction next to it. The goal is the reasoning, captured while it is still warm, not a polished after-the-fact story.

If you build personal tools, or you are curious what it looks like to treat your own mind as the thing the software serves, some of this may transfer. That is the whole bet of Sapix anyway: the best patterns are the ones that move across domains.

A note on the system writing alongside me

Most of these notes start as something Sapix surfaced. A connection it kept warm, a decision it remembered when I had forgotten the why. The blog is downstream of the substrate. That feels like the right order.

More soon. For now, the rest of the story lives in the repository.