At a moment when nearly every CEO is telling employees to embrace AI, there is a clear “do as I say, not as I do” problem: According to a recent study of 6,000 senior executives, nearly 70% of CEOs and other senior execs use AI less than an hour a week, including 28% who never use it at all.
I am not one of them. I don’t ask my teams to do things I won’t do myself, including becoming fluent in AI. But to be fair, this isn’t a burdensome task. While I trained as a doctor, I have always been a technologist first.
As a 5 year old, I took apart TVs to understand how they worked. At 10, I taught myself to code on a Commodore 16 with a tape drive. As a teen, I learned assembler, became obsessed with computer chips and hacked my machine until it felt like mine. In medical school, I built and sold an early messaging product that moved emails from Munich to America’s West Coast in five days (fast for dial-up FidoNet at the time). And in Zocdoc’s early days, I wrote the original database schema and first doctor-facing interface.
I had been tinkering with AI, but when Opus 4.5 launched in late 2025, it was clear that AI’s transformative possibilities were becoming realities. From then on, I became AI-first in how I work and made it a priority for Zocdoc to do the same.
I use AI in many expected ways and will spare you the productivity-maxxing version of this post. But more interestingly, AI has meaningfully changed how I operate as a CEO in three ways.
I Built AI to Dissent
An occupational hazard of being a CEO is that you receive more agreement than is healthy. Over time, that can narrow thinking. As a result, I am allergic to compliments, flattery, or anything adjacent to sycophancy.
As much as I encourage my teams to push back and challenge me (and believe me, they do in spades), it’s useful to build systems designed to challenge my thinking, not protect my ego.
I regularly use AI to stress-test assumptions, argue the opposite side of a decision, identify weak points in proposals, and surface second-order consequences. Sometimes the responses are insightful. Sometimes they are shallow or completely wrong. Both can be useful. Even weak counterarguments force me to sharpen my thinking.
I can now compress cycles of reflection and iteration that previously required multiple meetings, delayed follow-up, or exchanges extended by human niceties and hidden power dynamics.
AI Changed my Economics of Beginning
A blessing and a curse of being a founder is that I always have ideas. I am drawn to the novel, the new, the next frontier. My team regularly tells me that I live (too far) in the future.
Historically, many of my ideas died before they started because gathering context took time. Analysis took time. Building a model took time. Even getting to a first draft required enough effort that many threads were not worth pursuing.
AI changes the economics of beginning. I can pursue far more questions, analyses, and ideas because the activation cost is dramatically lower. I routinely explore strategy questions, acquisition opportunities, operational edge cases, workflow ideas, technical concepts outside my expertise, and market sizing and analyses that I would have previously deferred or dropped entirely.
I still catch obvious mistakes, shallow reasoning, and hallucinations. But I also find myself pursuing lines of inquiry I never would have explored before.
I believe that the supply of worthwhile problems to solve is infinite, especially in healthcare. The limiting factor has always been bandwidth. AI changes that equation. It makes me, and Zocdoc, much more generative.
I Have Strong Firsthand Instincts About AI
Using AI firsthand has also given me a much more realistic view of the technology itself. I have a feel for where it breaks down, where context collapses, where hallucinations appear, and where significant supervision and judgment is required.
At the same time, many still underestimate how effective these systems already are in everyday work. And how quickly they are improving. Using AI directly has given me sharper instincts about where meaningful leverage actually exists today, and a pattern recognition about where it will exist tomorrow.
I’ve developed a strong intuition for what should be built, where expectations are unrealistic, where I need to push teams harder, and where the technology is advancing faster than most realize. I could not build these instincts from demos, presentations, or summaries.
Being a CEO is not a Spectator Sport
The most important thing AI has given me may not be productivity or leverage, though it has. It is trust and credibility that I am asking my teams to go somewhere I am willing to go myself. They see that I am experimenting alongside them, wrestling with the same limitations, having the same breakthroughs, and forming opinions from firsthand experience versus abstraction.
If CEOs want their organizations to embrace AI, we cannot experience the future secondhand. We must go first.