This is the thing. AI doesn't fix operational chaos, it just speeds it up. I've watched founders automate a broken process and end up doing the wrong thing faster, feeling great about it the whole way.
"What happens today" is the question everybody skips. Building was never the hard part. Knowing what deserves to get built, that's the job. Good piece, Juan.
The bit about the old cost being a filter is right, but it was also cover. "We don't have the budget for that" is a much more comfortable sentence than "I don't think this is worth doing." Strip out the budget and people have to say the second one out loud.
So yeah, judgment is the scarce thing now. I'd just add that the failure mode isn't only building things that don't matter — it's building eleven of them in a fortnight because each one costs an afternoon. A notebook used to be where bad ideas went to be forgotten. Now they get a staging environment.
The cost of building something has dropped so far that the real discipline now sits in deciding what deserves to exist at all. I have watched people spend a fortnight on an internal tool that saved them an afternoon a month, and nobody thought to ask your closing question until it was already sitting there.
When creation becomes easy, knowing which problems truly deserve time, attention, and resources becomes the real advantage.
Yep, easier said than done for many operators out there.
This is the thing. AI doesn't fix operational chaos, it just speeds it up. I've watched founders automate a broken process and end up doing the wrong thing faster, feeling great about it the whole way.
"What happens today" is the question everybody skips. Building was never the hard part. Knowing what deserves to get built, that's the job. Good piece, Juan.
Thank you Mark. You are right. You can improve operations and still not accomplish your goal.
That's the biggest challenge now, not can we build but should we build.
Buy/lease decisions. Also low barriers to entry. A deeply strategic question.
Like I always say, JUDGEMENT is now king! Those who have it will excel in the age of AI!
I’ve fallen into this trap with AI agents myself. It’s easy to spend hours (or days!) building something that saves you 15 minutes.
The bit about the old cost being a filter is right, but it was also cover. "We don't have the budget for that" is a much more comfortable sentence than "I don't think this is worth doing." Strip out the budget and people have to say the second one out loud.
So yeah, judgment is the scarce thing now. I'd just add that the failure mode isn't only building things that don't matter — it's building eleven of them in a fortnight because each one costs an afternoon. A notebook used to be where bad ideas went to be forgotten. Now they get a staging environment.
The cost of building something has dropped so far that the real discipline now sits in deciding what deserves to exist at all. I have watched people spend a fortnight on an internal tool that saved them an afternoon a month, and nobody thought to ask your closing question until it was already sitting there.
AI has shifted the bottleneck from execution to judgment.
A working prototype creating false confidence is such an easy trap to fall into.