Less haste, more speed: clear AI instructions for better results

“Less haste, more speed.” That adage has never felt truer to me than now, working with AI.

I work all day with a team of AI agents alongside me. They’re not perfect, but they’re phenomenally capable, and improving every day. Given clear instructions, they can do a hell of a lot of the heavy lifting in my day-to-day work: research, analysis, evaluating options, designing, testing, troubleshooting, and documenting. Work I used to do sequentially can now happen in parallel.

The limit on efficiency isn’t the tech; it’s my clarity of thought and my leadership. The shifting idea of leadership here intrigues me. I’m directing agents trained on vast portions of human knowledge, and they can research anything on demand. They know more than I do about virtually every topic. The one thing they don’t know is the outcome I need right now, and that comes down to my instructions.

When I’m rushed, my instructions degrade, and so do the results. Most failures come down to the same things: I don’t give the AI enough agency, and I don’t bring it in early enough during planning.

Turns out the rhythm for making this hyper-AI approach work suits my life well. Get a good night’s sleep, go for a run, and start with a clear head. Then engage the team from the first idea, super early in the planning process, just like you would a trusted 2IC. It’s the opposite of being “busy”, and it’s a pattern that rewards being well rested and unhurried.

I’m still learning, but so far I’m enjoying the shift. 🙂