Note

2026-03-17

Hosung's systems depth = moat:

  - RoboEdge (#32) — Optimize AI models to run faster on robot hardware. His

  Nvidia job, directly. 1-month MVP.

  - RoboOS (#29) — Lightweight robot OS between Arduino and ROS. He built OS at

  Nvidia. Open-source growth play.

  - RoboConnect (#30) — Universal API to control any robot brand. Systems

  engineering is the moat.

  Agentic AI that "actually works":

  - RoboAgent (#1) — Tell a robot arm what to do in plain English. Foundation

  models make this possible now.

  - DroneAgent (#41) — Autonomous drone inspection for solar farms, towers,

  bridges. $10B+ market. 1-month MVP.

  - LabBot (#49) — AI agent that runs lab experiments on lab robots.

  Self-driving labs are hot.

  Big money:

  - MineBot (#48) — Coordinate autonomous mining equipment. $10K-50K/mo per

  site. 10% improvement = $10M-100M value.

  Fun/passion play:

  - RoboSchool (#50) — Teach robotics with real remote hardware. Combines

  Hosung's remote debug idea + Angie's NYU ITP education background.

 Okay, now I see the full picture:

  - Hosung is at Nvidia but his team (display/power) is losing relevance as

  Nvidia goes all-in on AI/data center

  - He can't go remote because hardware work

  - You're long distance and want to be together

  - Angie's company won't let her go

  The startup isn't just a side project — it's the exit plan. It solves three

  problems at once:

  1. Hosung pivots from a shrinking team into AI/robotics (where the industry is

   going)

  2. He gets to work remotely (or from the same city as you)

  3. You build something together

  That changes the question from "what's the best market" to "what can Hosung

  start building now, part-time, that gets good enough to leave Nvidia for?"

  Which means it needs to:

  - Let him work from anywhere

  - Be software (not tied to a lab)

  - Use his systems engineering skills but pivot toward AI

  - Show traction fast enough to justify leaving

  Want me to think about it from that angle instead of generating more idea

  lists?