Turns out prompting is the boring part

I'd been doing the prompt thing. collecting them, saving them,

thinking that was the skill. then you sit in one of these and

realise the people actually building treat prompting as the boring

part. everything interesting is around it — what you feed the

model, catching when it's confidently wrong, making it run without

you babysitting it.

went back and looked at what they've run over the last year.

context engineering, rag, advanced rag, mcp and agents, langchain

and langgraph. that list is basically a syllabus if you're trying

to figure out what to learn next. I've been working through it in

that order.

haven't done their paid cohort so I can't speak to that. free sessions are open to anyone — thegenacademy.com

where are you in that list? and if you've built anything agentic that actually runs unsupervised, I want to hear what broke first.