Week one was mostly thinking before building. Before I wrote much of anything, I had to decide what NEURA actually is and what it should run on. That feels slow, but getting it wrong now costs months later, so I took the time.
The biggest call was the shape of the whole product. NEURA is a personal AI you own. It runs on your own hardware, not somebody else's cloud that can change the rules or quietly read your life. There is an optional robot body if you want NEURA on your desk looking back at you, but the companion comes first and the robot is the bonus. Locking that in on day one set the direction for everything since.
For hardware I settled on a Raspberry Pi as the brain, with a small round screen up front for the face. The round screen matters more than it sounds. A face you can actually look at changes how a companion feels, and I wanted NEURA to feel like something, not just a box that talks.
With the brain picked, I mapped the software stack around it. Nothing is locked yet, but I know the rough shape of the pieces and how they fit, and that is enough to start moving.
A lot of the week was reading. I went back through my reference material and my earlier DeckOS work to find the best starting point instead of reinventing everything. Some of it carries over cleanly and some does not, but knowing what I already had saved real time.
It feels good to have the vision clear. NEURA is real to me now in a way it was not on Monday, and next week I finally get to build on it.