There's a case to be made that the resource-intensive piece is training/creating the LLM and just running a fully trained model is not that intensive
Try it let us know how it goes
1 Like
I’ve done a bit already, it’s not hard to research just a Lot of Vibe Reading (very different from vibe coding) on other people’s experiences
Subredddits LocalLLM LocalLLaMa. People are mixed on qwen 3 and love some 2.5 offshoots and that’s just one random one. If you don’t have bad internet and you’re not doing anything with the bandwidth it’s not too annoying to pull a model down and try it a few times. Waste that energy good
1 Like
By the way did you see open-webui?
No
Pretty nice. UI work is pretty much irrelevant these days though - claude code can build all of this for you in a few hours
1 Like
yEAH im sure it functions just as well haha
Big ass indirectly comfirming the "work" he does is incredibly elementary as we all expected.
2 Likes
He really is just a nepobaby fraud isnt he? Wow
1 Like
just the classic manager forciblky exposed to ai and sold on the dream of drop-in replacement for skill
Nobody can compare with your parental compensation thats pretty true. Also the invaluable work of prompting claude cursor you do is truly invaluable.
the productivity is nuts. claud does his tickets within an hour including checking the work. then he passes the slop to someone who hopefully knows what theyre doing
2 Likes
Putting this in my signature.
4 Likes
holy shit