Of course I learned from it. I mean the reverse engineering part which is basically try and error is something I rather skip. The remaining things like wiring the hardware is still there. The boring stuff is what the LLM can do for me. I still find the process to get stuff working challenging and interesting. It's not only about the end result. It's just a different approach than the old school low level one

This right here. People simp for LLM companies as if their experience of using the out-of-pocket top-of-the-line "team of PhD's" paid models will be what is deployed when trying to contact your bank, insurance, etc. No,... once tech companies stop playing the "no/some revenue until we own the world" VC game, we'll all be stuck trying to talk to GlueSnifferGPT when reporting an emergency.