I recently quit a job I’d been at for three months. On the surface, this sounds impulsive. But the timing—as AI reshapes every corner of knowledge work—made the decision feel almost inevitable.
This isn’t really a story about quitting. It’s about a fundamental mismatch between how most companies still organize knowledge work and what actually matters now.
Table of contents
Open Table of contents
The Old Mental Model
For decades, the efficient way to build software was specialization. You hired frontend devs, backend devs, QA engineers, product managers. Each person owned a lane. Coordination costs were real, but so were the benefits of depth.
This model assumed knowledge work was mostly execution—the more hours you put in at your specialty, the more value you produced.
What AI Changed
That assumption no longer holds.
Coding assistants handle boilerplate, tests, and increasingly complex implementations. Marketing copy, research summaries, data analysis—all commoditized rapidly. The things that used to justify a narrow specialist are becoming table stakes.
What remains genuinely valuable is harder to automate: taste, judgment, the ability to see across domains, the initiative to ask the right questions. A person who understands the full stack—not just technically, but across product, market, and technology—can now execute at a pace that would have required a team two years ago.
The Suffocation of Forced Narrowing
After years of working in a role where I was expected to think broadly—about industry direction, what we should be building and why—I joined a company that wanted me to execute narrowly. Just do the task. Don’t ask too many questions. Ship the ticket.
I have a phrase for this: forced cognitive downgrading. Expanding your thinking is hard—it requires experimentation and failure. But having your thinking artificially compressed is worse. It feels physically wrong, like breathing through a straw.
The bitter irony: this happened at exactly the moment when broad thinking is more valuable, not less.
996 Is the Symptom
The company also had a 996 culture (9am–9pm, 6 days a week). Someone smarter than me wrote that a successful company isn’t a sprint, it’s a marathon. But there’s something more specific here: 996 made some sense when output was proportional to hours. It makes no sense when the leverage is in how you think, not how long you work.
Overworked, narrowly-scoped engineers in 2026 aren’t a productivity asset. They’re a misallocation of human intelligence at exactly the moment when human intelligence is the scarce resource.
The Super-Individual Thesis
I believe we’re entering the age of the super-individual.
With AI as a force multiplier, a single person with taste, judgment, and cross-domain fluency can build things that would have required teams. The leverage is real. The companies still optimizing for the old model—specialists in lanes, lots of them, working very long hours—are burning fuel in the wrong engine.
What I’m Doing Instead
I’m taking time to think. Working on side projects. Writing. Staying close to where AI is actually moving.
I’m not in a hurry to find the next job, because the next interesting opportunities look different from the last ones: smaller teams, more autonomy, global scope, genuine AI leverage.
Uncertainty without clarity is scary. Uncertainty with a clear sense of direction is just… interesting.