Mar 1 • General discussion
What Is Your Single "Best Practice" When Working with AI?
I'll share mine first in case it helps. We all have different blind spots and maybe we can help each other solve for them together.
Mine is that I and my agents all focus on what I call the Prime Directive. If you happen to recognize the phrase as one from 2001 a Space Odyssey (or even Star Trek)... congratulations, you are also a nerd. 🤓 😬 ☺️
My Prime Directive is "Do the thing (or make the choice) that makes all subsequent things/choices faster, easier and more profitable.
If I focused on the thing that makes the next thing faster (Helllloooooooo, Claude!) I miss the larger vision. AI agents (and most people) are often myopic. They fail to see and work towards the big picture, often sabotaging the big picture in order to get to the current goal. That's why benchmark testing is becoming more and more useless since the AI are learning how to game the sytems.
If I focus on what makes things more profitable, I may sacrifice integrity, quality and reputation. The natural consequence is that everything gets harder as I have to work through the fallout and opposition that comes from such things.
Ease is relative. Some math calcuations are easy for some and impossible for others. If a person or agent is responsible for something that is difficult for them, it will take a long time and impede progress. A solution must be found if progress is to happen and momentum increase. If that resposibility can be given to another person or agent because they already possess the ability to perform that task quickly and without strain, that is a native solution and falls within our prime directive.
I pepper this logic throughout my docs and conversation both with humans and ai agents because it is VERY easy to conflate short term goals with actual success of long term goals when they are always quite different.
If an agent asks me (as they do annoyingly often, trying to program it out) what do you want to work on next (posts a list for me to review here) - I ask if they are aware of the Prime Directive. Then they will realize I want them to apply it. They are almost always wrong. When they are right... I know I've improved clarity in our goals and documentation and systems.
So it serves both as an optimization system as well as a litmus test for how well my systems are doing. And it's been incredibly helpful.
3
2 comments
What Is Your Single "Best Practice" When Working with AI?
skool.com/myaifreedomsystems
MyAiFreedomSystems - Maximize Profits - Streamline Operations - Love Your Business
Powered by