AI Didn't End Burnout — It Just Gave Us More Ways to Overcommit

The promise of AI was that automating grunt work would leave people underworked and stress-free. The author, a serial startup founder, found the opposite: because AI slashed the cost of starting things, he spun up roughly 40 proof-of-concept side projects in stolen five-minute windows between meetings. Each one became an open loop demanding attention, and the accumulated weight produced exactly the burnout AI was supposed to eliminate. His diagnosis is that burnout stems not only from overwork but from the endless make-work that cheap productivity invites.

His proposed fix borrows from Greg McKeown’s Essentialism: use AI to go deeper on the few things that matter (vertical) rather than to multiply the number of things you attempt (horizontal). He leans on Garry Tan’s eclipse analogy — the gap between a 99% and 100% eclipse looks like a rounding error but is experientially a 100x difference — to argue that the final 1% of polish, which often consumes the majority of the effort, is precisely what separates good-enough work from work that lands. Most people ship at “good enough” because that last stretch is so costly.

The takeaway for anyone leaning on AI tooling: since AI cheapens the bulk of the labor, there’s no longer an excuse to skip the finishing work. The winning strategy is ruthless focus and follow-through on a small number of projects, publishing less but making each output genuinely excellent — turning saved time into quality rather than into a sprawl of half-finished experiments.