
Know when to hold em, fold em, walk away, and run when working with AI development agents
Working with development agents is a lot like being the gambler.
Hopefully more than half the time: Hold em - it's like having unlimited junior to mid level engineers on your team that can get what you're asking for done, but in minutes instead of hours or days. Magic.
Fold em: Notice when it's going down a very ill advised path and causing problems down the road on its way to the local solution.
Walk away: This is not working and now you're in your third doom loop in a row. Step away, tinker with the code yourself and think of a better prompt, or get some sleep.
Run: If you have the (understandable) urge to YELL IN ALL CAPS at the AI all day, Maybe take up wood working or consider walking into the forest, never to return. I think for better or worse being a code reviewer, manager, and exploratory tester for AIs is almost certainly our near to intermediate future for most of us in the make software in exchange for money game.
Here's hoping they manage to increase the "magic" rate because when it works, I don't think I've experienced a greater feeling in my career than a working one or two shot new feature that would have taken me at least a full day.
How git worktrees enable effective collaboration with multiple coding agents without the context switching penalties
My workflow makes it easy to spin up 10+ parallel agent threads, but 5 or 6 well-sized tasks is the sweet spot - for my sanity and to keep the final review pile manageable.
How connecting the right Model Context Protocol servers in AI coding tools dramatically improves output quality
Get More Like This
Follow along as I build and share what I learn
Found this helpful? Share it with your network!