Then I take them apart and build them better.
I'm Mann, a founder and incoming computer-engineering student in Atlanta, Georgia. I co-founded and run a detailing business across four cities, operate a hardened self-hosted server stack, and design and manufacture my own hardware. Below is the work that's running now, what's in progress, and the experiments I chose to kill.
Calendar and tasks are live behind one dashboard; building the agent layer that runs them now.
Building small, scoped agents next to the evals that catch where they fail.
Mostly box volume and port tuning right now.
The process is the point.
CAD and planning only reveal so much. The real constraints — tolerances, fit, materials, edge cases — show up in the first physical build. I prototype fast and rough on purpose, so the problems surface early instead of late.
I pressure-test an idea against its real limits — platform rules, unit economics, regulation — before I build. Killing an idea on those grounds takes a weekend; learning the same thing after months of work is the expensive way.
Running a business while building its software and its hardware taught me to work across the whole stack. The marketing funnel, the server config, and the printed part are the same problem seen from different angles.