There is something incongruent about some devs who claim to have a ton of success with coding-oriented LLMs. They seem to be by far in the minority in my experience, and all the tricks they claim saves them a ton of time don鈥檛 seem to pan out when I try them out. When I get an opportunity to do so, I think I should sit down and watch what they actually do to understand why they claim this. Maybe they鈥檙e fooling themselves? But maybe they鈥檙e doing something I can benefit from.