How I'd Learn n8n if I had to Start Over in 2026 Nate Herk | AI Automation 465k subscribers Subscribe

This guy basically says most people start learning automation the wrong way—they jump right into trying to build AI agents because it looks cool. He thinks that’s a mistake. His big argument is you should start with boring, regular workflows first. Stuff that just moves data from A to B without any AI. Get really good at that. Learn how data moves in JSON, how to use APIs, and how to handle errors. Once you’ve got that down, then you add in a little AI for small decisions. And only after you master that should you try to build a full AI agent that can think for itself.

I thought his point about the “transition curve” was really relatable. He says when you start you’re super excited, then you get overwhelmed and feel like quitting, and you have to push through that to actually get good. He also stresses that you should build automations that run on their own in the background—like processing a new lead—not ones where you have to talk to a bot to make it work. His advice to map everything out on paper before you even open the software is solid. It stops you from building a messy, broken workflow.

Overall, it’s a good reality check. It pushes you to learn the fundamentals properly instead of just chasing the flashy AI stuff that might break. It’s a slower, more disciplined way to learn, but it probably leads to better results in the long run.