MAGI (Machine Augmented GTK Interface) is a local open-source AI powered operating system. For now, there’s a built in LLM with context awareness, voice transcription you can use anywere, and an e-reader built-in that can read anything you select/highlight. The goal is to continue integration so that this can be used as a full AI agent OS. Potentially allowing you to interface with your computer fully with voice and onscreen visuals/verbal feedback. All while remaining fully local and private!
Status
You can check out the code here, but the ISOs are no longer hosted due to a lack of interest. This project is being shifted towards a desktop environment rather than a full-blown Linux distribution. Today it’s tested on Debian, but should work on other distributions.
Okay but How?
By using as much of Mate (The desktop) and just replacing the panels. This lets me rely on a full set of Desktop tools that cover a wide range of basics like Monitor setup/themes/compositing (Thanks marco). For the LLM support I’m suing Ollama. For voice transcription, I’m using Whisper.
Support
Like what you see? Support me here! Or jump in and push some code! I’m also looking for a job if you want to pay me full time. 😉