Just what the post says. Looking to make local AI easier so literally anyone can do “all the things” very easily. We built an installer that sets up all your OSS apps for you, ties in the relevant models and pipelines and back end requirements, gives you a friendly UI to easily look at everything in one place, monitor hardware, etc.

Currently works on Linux, Windows, and Mac. We have kind of blown up recently and have a lot of really awesome people contributing and building now, so it’s not just me anymore it’s people with Palatir and Google and other big AI credentials and a lot of really cool people who just want to see local AI made easier for everyone everywhere.

We just finished automatic multi GPU detection and coordination as well, so that if you like to fine tune these things you can, but otherwise the system will setup automatic parallelism and coordination for you, all you’d need is the hardware. Also currently in final tests for model downloads and switching inside the dashboard UI so you can manage these things without needing to navigate a terminal etc.

I’d really love thoughts and feedback. What seems good, what people would change, what would make it even easier or better to use. My goal is that anyone anywhere can host local AI on anything so a few big companies can’t ever try to tell us all what to do. That’s a big goal, but there’s a lot of awesome people that believe in it too helping now so who knows?

Any thoughts would be greatly appreciated!