Hi from Germany, I’m not much into AI and with my pathetic 3GB VRAM I’m not sure, if it would be feasable to get useful local AI running. But recently I read about a very interesting project: llama conductor From what I understood it provides a way to handle ground truth, weave it into your llm interactions and get a confidence level for every answer (to avoid the confident halucinations). I’m not sure, if it can hold, what it promises. But I certainly would like to try.