But Did We Finish?

AI output feels good. The high of watching an agent knock out in a minute what used to take an afternoon or longer. It’s what makes us say yes, please! Yes to more, yes to faster, and yes to the whole industry’s pitch that we can and should be producing around the clock. If we’re not, someone else is. Right?

But through these highs, I’ve found myself more often sitting with what gets left behind. Saying yes also results in a ton of work in progress scattered across my desktop, terminal windows, agent chats, and more. Like any consumption high, it’s fleeting, and the tabs stay open long after the endorphins are gone.

I doubt I’m alone here, at least not amongst those working at a startup. Most nights before I fall asleep there’s a small panic for all the real work still sitting unfinished. Yet morning comes and the loop continues. Endless fast starts and limited finishes.

I see how we got here. We started keeping score on output. But was output really what we wanted? Anyone else longing for a calm nervous system and the satisfaction of starting real work and finishing it well? Finishing is not the reward. It’s the point. A system built to keep producing never lets you get there.

Skipping that closure is felt. The elevated cortisol and brain fog are real. Decisions get slower, you produce more and trust yourself less. And underneath all of it, ownership drains out of work you did not really make, so you care about it a little less each time.

Maybe the answer for most of us (dare I say the 80% that aren’t output-maxing) isn’t another accelerant. It’s a calmer way to start and finish real work. Slow that down. Give people the closure. Ironically you might get more done, not less.

We were sold productivity. What we got was output. Those were never the same thing.