The same conversation keeps finding me — in Nashville, Ottawa, Glasgow, Lisbon, and most recently, at SXSW in Austin.
“How is the startup CEO job changing in the age of AI?”
I suspect most people are really describing how agents are reshaping the way they run their companies. But I take the question literally.
And my answer: not much.
The CEO’s responsibility is threefold:
- Choose the mountain the company must climb.
- Choose the people to climb it.
- Ensure those people have the tools to summit safely.
When the company is young, the CEO does these things directly. As it matures, that responsibility shifts from direct oversight of the mountaineers to the creation of systems: goal-setting, budgeting, hiring practices, communication architecture, you get the idea.
So long as there are humans to inspire, align, allocate, and support, the CEO job looks the same in an AI world as it did before.
But isn’t go-to-market changing? Isn’t SaaS dead? Aren’t agents replacing tasks that humans once performed?
Sure. Just like assembly lines did. And then the personal computer. And then the mobile phone.
The way we attract, win, deploy, and serve customers is being actively reimagined by what feels like AI-induced magic. But when I look at the CEO’s role, I still don’t see a fundamentally different job.
What I do see are two tensions that feel new.
Tension 1: The Two Tribes
You now have two tribes of employees.
The first comes from the “before times,” used to sprints and PRDs, writing and editing their own work, asking questions of humans or sorting through page after page of Google results.
And then there are the mutants. With a wave of their hand at J.A.R.V.I.S., applications go live, content is written, videos are crafted, research is completed.
(Quick aside: A vegan, a CrossFitter, and a Claude-coder walk into a room. Who talks about themselves first?
None. The one using Claude had an agent beat all three of them into the room to announce his impending arrival.)
Jokes aside, one side argues for the value of unaided thought; the other is living in the future.
Both are right. But history is going to look back on the former the way we smile at a candlemaker in an old-timey festival. A beautiful art, not actually useful.
Tension 2: Effort vs. Output
Some companies are already mandating the use of AI tools. The logic is simple: If headcount stays constant, and AI multiplies output, revenue per employee should explode.
But employees don’t see it that way. They see a slew of new tools as a path to deliver their historical output at a fraction of the effort.
We know who wins this war. But there will be many battles before a new equilibrium settles.
The pandemic democratized remote work in 2020. AI is doing something similar now, but the sorting mechanism is starker.
Employees who work hard with AI to produce exceptional output will be employed. It won’t matter where on the planet they live. Employees who use AI as a shortcut rather than an amplifier will be terminated.
This shift is different from the technological changes of our past. It doesn’t require technical skills to reach proficiency; it just requires discipline.
You don’t have to train a new generation to use these tools. The user experience for all of them is converging toward a conversation with your computer, in your own voice, with an infinitely patient teacher on the other side. Unlike learning to code, which is a real hill to climb for much of the world, this mountain just requires someone to talk.
There will be no empathy from those in charge for employees unwilling to make such a small leap.
So, is the CEO role truly different?
I don’t think so. So long as there are humans in a company, what a CEO does feels eternal.
But the arrival of AI in late 2022 is a dividing line: before and after. And it will be the CEO’s job to integrate the tribes and hold the line. There is no going back.