Local AI

I am going to try out local AI again now that things have developed a bit more. The last attempt was not good!

Trying MSTY Studio, it seems easy enough, no Docker, etc. Qwen3 worked well, but Qwen 3 Coder needs 18 GB of RAM so I will have to upgrade the Lenovo Think thing. Anyway, that’s enough info to restart the project when I have enough RAM.

Note to self, don’t get carried away with what these local models claim. Even Claude messes things up so this is a long way off from being possible yet.