Hi Dorian, I’m Ethan Reed. My work is adjacent to agent tooling: a hardware project called Lumi. It gives agents a physical way to operate real mobile apps.
I found your work through GitHub and figured this could be useful for developers working with AI agent tools. The problem I’m trying to solve is that agents are becoming powerful in browser and API environments, but a lot of useful work still happens on the phone. That context is valuable, but it’s still hard for agents to access and act on directly. Lumi is my attempt at a simple physical layer for that: an agent observes the phone screen, moves to a target position, and performs real taps, swipes, and typing on the device. I’d love to hear how this feels from an agent builder’s perspective. Would you be open to taking a quick look at the preview? Best, Ethan Reed
