Eric Siu explains four mistakes that slow AI workflows: using expensive models for simple tasks, skipping automation for repetitive work, ignoring learning loops that adapt over time, and not using the /goal command for autonomous long-running tasks.
GPT-6 Astra mistakes to avoid. Number 1 is to use the highest thinking mode, the most expensive model for everyday menial tasks. That's the equivalent of bringing a gun to a knife fight. Number 2, not using scheduled routines on tasks that you're doing over and over. At a certain point, you have to figure out how to make those tasks into skills and then repeat them as routine. Number 3 is not using loops. So after you've made a skill into a routine, you want to make it into a loop that learns over time and picks up the patterns, and that way you can Number 4 is not using the /goal command for longer running work. So if you don't use it, you're gonna have to sit and attend to your computer more often for every single prompt, every single question. If you use a /goal command, you can get more done without having to be around your computer the whole time and just let the AI figure it out. If you enjoy this, just follow for more on AI, business, and marketing. Thank you.
