Venturing into the world of local LLM's, would love some pointers!

Hi everyone! Very exciting times we live in where we can run models from laptops and GPU's which 5-6 years ago would've been SOTA. I have been working with cloud models for years now, and I am now starting to dig into local models. At work, I am leading a few different AI projects across the biz, and with our devs (who all love claude and have seen real value from it), our biggest pain point is the limits at the moment. SO, I have started to have a play to see what the art of the possible is with local models.