Just thought I'd share this use case. I was setting up a miniPC as a home theatre with Archlinux (It's the OS I'm most familiar with). I needed to twiddle some things and am not yet familiar with wayland (I'm trying our hyprland, but normally rock i3). So, I installed pi coding agent, pointed it at my desktop/AI server thing with Qwen, and then ... just told it what I wanted.
Setting up bluetooth became "Can you connect to my bluetooth speaker. It's a panasonic soundbar". Changing HDPI scaling became "Can you fix the screen resolution" and then it just did it, occasionally telling me to run a sudo command to install something. I wasn't quite brave enough to give it root/sudo directly, but I really don't know why. It's not like there was any private data or keys on that machine, it was the very freshest of installs.
I'm now considering putting hermes on the machine with full root access and some sort of voice input. I mean, why not?
This experience definitely raised questions on the future of computers for me - and what interfaces we will use in 5 years time. I don't know what it'll will look like in 5 years, but yolo mode with agents on your local hardware are epic!
[link] [comments]