AI took the effort. Watch what goes with it.
Effort used to be the filter that quietly told you what was worth doing. AI removes the effort. The danger is letting it remove the judgment that was the actual value.
When a task used to take three hours, you had a built-in filter. You didn’t even realize it was there. You just looked at the afternoon it was going to eat and asked if it was worth three hours of your life. Half the time, the honest answer was probably no, so you killed it. The friction did the deciding for you.
AI took that friction down to zero. That feels like a massive relief right up until you notice what else left the room.
When everything is cheap, you do everything
Once the cost drops, the filter drops with it.
A task that used to cost half a day now costs you a single prompt. So you fire it off without asking if it matters. Then you fire off nine more. None of them move the needle. But because an answer came back sixty seconds later, it feels like real work happened.
That’s the small trap: busywork on steroids. You end up twice as busy and pointed at absolutely nothing. The friction you’re proud of eliminating was the only thing keeping you pointed in the first place.
The bigger trap: outsourcing the part that was yours
The second trap is the one that actually costs you.
When a model will happily spit out an answer to anything, it’s dangerously easy to let it make the call. Which lever to pull. What product to kill. What to build next quarter. The answer arrives formatted like a finished task, so it feels like something you can delegate.
Those are judgment calls. And judgment is the only reason you’re in the chair.
Hand it over because deciding is hard, and you’ve automated away the one thing that was creating value. That’s how a company gets screwed slowly. It doesn’t happen with a bang. It’s a quiet drift where nobody’s steering.
I catch myself doing it too. It’s wild how natural it feels. You open a tab intending to get a little leverage on a problem, and ten minutes later you’re letting the screen do the thinking. The line between “help me think” and “think for me” is almost impossible to feel in the moment.
The discipline
Most of the noise about prompt engineering is nonsense. Prompting is cheap. The real discipline is deciding, on purpose, what stays yours.
Use the machine to strip the effort out of the grunt work. Let it summarize the call notes, format the data, generate the boilerplate, write the first pass. But the moment a decision has teeth, pull the work back. Sit with the friction. Do it the hard way.
The operators who win with this stuff won’t be the ones generating mountains of cheap output. It’ll be the ones who keep a very short, very stubborn list of things they refuse to hand over.
Know where this has to go, and not how?
Twenty minutes. Tell me where you want this company to be in two years and I’ll tell you what I’d go look at first.