oof - got texas deployed on voxels - went for a run and a swim - now have spend 3 hours solid getting all my other sites checked out and up to date so they're current and easy for the robot to deploy to.

main thing i wanted to do before i get back to voxels is release botparty since i really enjoy using it as a chat interface for shit talking (chatgpt and codex are great for work, but botparty is really designed for shit talking).

I have the app running, but gotta get the billing site and API key stuff setup so i can point it at a qwen 27b inference provider.

which means i will have diversified my income stream ($10 a month paying for my own service counts honestly, and maybe i can convince my girlfriend to sign up as customer #2)

but it will also mean that i'll have some serious firepower in place (once it gets stable and i start to get a feel and trust the production infrastructure) so i can use a qwen base model with my own LORA trained to give the botparty-specific assistant tools - and also a lora trained on the new zealand (english and te reo) cultural style.

Chur Bro, gonna be fully sick to have a robot thats as smart as the quant from the big short but talks like Barry C. crossed with Jermaine C.