# Alicia

Catch a thought now. Come back with a better question later.

A thought usually arrives while I'm doing something else. By the time I've found the right window and decided where it belongs, I've lost half of it.

Alicia is the Mac companion I built for that moment. I can capture a note, ask a question from my archive, or give an app command. And yes, the name is because she always uses the best notes.

## Three things, kept separate

```mermaid
flowchart LR
%% id: alicia-overview
  Me[Something on my mind] --> Capture[Capture a thought]
  Me --> Ask[Ask a question]
  Me --> Command[Give an app command]
  Capture --> Notes[My notes]
  Notes --> Ask
  Ask --> Answer[Answer with sources]
  Command --> Action[Known action or clarification]
```

The idea is to make capture cheap and understanding revisitable. My original thought stays around even if its category changes. Later, a question can bring the relevant notes together into an answer, with the excerpts alongside it. I can follow the answer back to what I actually said.

Commands need a different kind of judgement. Jev can help interpret which known action I mean; Alicia's local rules decide whether to act or ask for clarification. A model's interpretation should have a clear boundary before it changes my desktop.

Those three jobs share a convenient entry point, but each has its own responsibility: **preserve the thought, explain from evidence, act within a known scope**. Alicia runs as a native app on my Mac, with its own note archive.

[Back to all work](/work)
