AI Founder Mode Means Hiring Micro-CEOs

Brian Chesky's Founder Mode, rebuilt for the AI era: why founders must hire Micro-CEOs who manage agents, not just teams,

TL → DR

Brian Chesky named founder mode after losing control of a 7,000-person Airbnb and clawing it back through 35 hours of meetings a week. He now sees the next version, AI founder mode. Meetings were only a workaround for an information problem, and AI removes it. Management layers collapse toward four. The real shift is who you hire: stop hiring polished people managers, start hiring micro-CEOs, people biased toward execution who own outcomes and run agents. The founder's hours move from meetings to building workflows. Get the hire wrong and the company dies faster than before.

---

What is Founder Mode?

Brian Chesky made founder mode a public idea after Paul Graham coined the term. In a recent interview, he went further. He described what comes next and called it AI founder mode. He has not finished working it out. But its shape is already clear enough to act on, and the version that matters most is not the one running inside a 5,000-person company. It is the version for the founder with 600, 60, or 6 people.

Start with what founder mode actually solved:

• By 2019 Chesky ran 7,000 employees and could not turn the company. He described feeling as if he were in a car without a steering wheel.

• He had over-delegated to professional managers and was being managed instead of managing.

• The pandemic cut revenue by 80% in eight weeks and forced the correction. He took control, worked 100-hour weeks for two to three years, and reviewed every detail of the business.

• Chesky bought visibility with meetings. He ran about 35 hours of meetings a week, group meetings only, no one-on-ones, the full chain of command in the room.

• He reviewed everything weekly, biweekly, monthly, or quarterly.

• He spoke last and ratified every decision.

Founder mode was a workaround for an information problem. A founder who cannot see the company pays for that sight in calendar time.

---

How Does AI Change Founder Mode?

Chesky said AI founder mode means even more detail, because the information is available on demand. The meeting was never the value driver. It was the toll the founder paid to know what was happening. When the work itself becomes legible without a meeting, the toll drops to near zero. He expects the culture to move from meeting-based to asynchronous, the layers of management to shrink toward the four that the Catholic Church has run on for 2,000 years, and every job to change. His hard line: pure people managers do not survive.

Everyone becomes a hybrid manager and individual contributor with what he calls contact with reality. An engineering manager still codes. A lawyer who manages still reads case law.

Chesky is not forecasting theoreticallyically. Harvard Business Review and MIT Sloan Management Review have spent 2026 documenting the same shift, and large companies are already collapsing their org charts. Bayer cut its management layers from roughly twelve to six and removed close to 5,500 roles, most of them in management.

The direction is set. The open question is what replaces the layers.

New to Operating?

Check out these reader favorites:

• Operating Economics: The Optimization Company (Amazon)

• Operating by Warren & Charlie

• Winning the Loser’s Game: The Creator Economy and the Companies We All Need to Build

• 0 to 55,000 - The First 90 Days Playbook - What I’d Do Differently Knowing What I Know Now

Become a paid subscriber to join the Operating Skool community and attend our weekly live Office Hour Q&A’s w/ John.

---

Operator Perspective

Here is what I see from inside my advisory work at 6AEP.

The traditional founder problem has always been the same:

• The founder starts with the minutiae and the details. The challenge is to get out of it and delegate to a team. Some founders never make the delegation happen. They hire very high-quality people and then do not give them enough autonomy. Those good people leave. The only thing that keeps them is the company being wildly successful, and most companies are not.

AI founder mode does not remove the delegation problem. It moves it. The delegation still has to happen, but a large part of it now happens digitally, across agents and platforms. It is embedded in the founder's relationship with the technology. The founder is not just handing off the tasks that the technology now does on its own. The founder is delegating to humans, and the management of those humans is now mostly about one thing…understanding the level of individual productivity achievable and the quality of the direction the human team is giving to the tech.

That changes who is worth hiring.

The person who is an excellent people manager and relationship builder is worth less than they used to be. A founder needs someone more technically capable. Someone who can think across multiple departments and build out the agents, the applications, and the execution layer that gets the work of those departments done. The way I have started to see it is that the founder or CEO wants to hire teams of micro-CEOs. People who think strategically across the whole business, who carry real ownership, and who have a bias toward action and outcomes that a founder has.

Harvard Business Review reached a version of the same conclusion. In February 2026, it argued that the AI era needs a new role, it calls the agent manager, the way the software era produced the product manager. Here’s one version using my thinking. The micro-CEO is the ceiling. The agent manager runs the agents. The micro-CEO owns the outcome. And in some orgs, the Micro-CEO and the agent manager will absolutely be one person overseeing multiple divisions or departments.

This shift necessarily changes the hiring signal. It makes it more of a matter of the candidate's industrial psychology:

• You want people heavily biased toward execution.

• You want people who have actually created outcomes, not described them.

• You want technical know-how across multiple departments.

• You do not just want leaders.

• You do not just want people with a lot of education.

• You want enough education and enough smarts, paired with a heavy bias toward having been in the execution of a company’s work, not only its management.

Look hard at personality and at how a person thinks about their life. I tell founders to stay far away from anyone whose first concern is work-life balance. The green flag used to be the polished manager who builds relationships. That is now a red flag.

The founder’s own day changes as well. A majority of the founders’ hours should go to building better workflows. The founder uses agents and technology to create workflows that get every department's work done. Better customer experience. A better product in the customer’s hands. Fires are put out faster for customers. Faster prototyping. The needs vary by company, but the work is the same shape. The founder builds the workflow, shares it with the team, manages it, and measures its execution against the business's growth goals. The meeting was where the old founder spent the hours. The workflow is where the new founder spends them.

MIT Sloan Management Review made the same point in plainer terms. Its researchers found that AI does not create productivity on its own. The gain comes from redesigning the work. Companies that deploy AI tools on top of old workflows get almost nothing for the spend. The founder who treats AI as a tool to buy will lose to the founder who treats it as a company to rebuild.

Get this wrong and the cost is real. The founder makes the old hire, the meeting-centric people manager. That person is expensive and not productive enough. Worse, that person recommends the company take on too many tools, too much software, too many technological assets, to run a people-management culture inside a company that should be built around outcomes.

The company fills up with meetings.

It stops moving fast.

It does not prototype fast enough.

It does not respond to customers fast enough.

It does not see the whole picture fast enough.

It does not innovate fast enough.

It does not grow fast enough.

The company dies.

The founder who does not understand this makes the hires that kills it.

There is a second way to get it wrong. In May 2026 Harvard Business Review published a large-scale experiment on companies that placed AI agents on the org chart and treated them like employees. The result was worse, not better. Treating the agents like people reduced individual accountability, raised unnecessary escalation, and lowered the quality of human review. The lesson holds the whole argument together. Delegation to agents is not abdication to agents. The founder and the micro-CEOs stay accountable for the work the agents produce. That is what contact with reality means.

The honest version of this is not complicated. Founder mode was about the founder staying close to the work when the company grew too large to see. AI founder mode is about who the founder builds the company with. The founder who understands that first will use it to build a company that stays small in headcount, strong in ownership, and close to the customer the whole way up.

- j -

---

Frequently Asked Questions (FAQ)

What is the difference between founder mode and AI founder mode?

Founder mode solved an information problem. A founder who could not see the company bought visibility with meetings. AI founder mode removes the information problem, because the work is legible on demand. The meetings go away. The detail does not.

Does AI founder mode only apply to large tech companies?

No. Chesky describes it from a company that already scaled to thousands of people. The sharper version is for the founder with 12 people. For a small team the value is the option to never build the management layers that detach a founder in the first place.

What is a micro-CEO?

A micro-CEO is a hire who thinks strategically across the whole business, carries real ownership, and holds a founder’s bias toward action and outcomes. They run the agents and own the result. They are not coordinators of other people’s work.

What kind of person should a founder hire now?

Hire for the psychological profile. You want a heavy bias toward execution, a record of creating real outcomes, and technical range across departments. Education and intelligence matter, but only when they are paired with execution experience.

Do pure people managers still have a place?

Less than before. The polished manager whose only output is other people’s output used to be a green flag. It is now a red flag. Everyone needs contact with reality. An engineering manager still codes. A lawyer who manages still reads case law.

What should the founder actually spend time on?

Building workflows. A majority of the founder’s hours should go to using agents and technology to design the workflows that produce better products and faster outcomes for customers. The meeting was where the old founder spent the hours. The workflow is where the new founder spends them.

What is the most common mistake?

Making the old hire. The meeting-centric people manager raises cost, sprawls tools, and fills the company with meetings. The company slows until it cannot prototype, respond, or grow. The wrong hire kills the company, and faster than it used to.

---

About the author

John Brewton documents the history and future of operating companies at Operating by John Brewton. He is a graduate of Harvard University and began his career as a PhD student in economics at the University of Chicago. After selling his family’s B2B industrial distribution company in 2021, he has been helping business owners, founders, and investors optimize their operations ever since. He is the founder of 6A East Partners, a research and advisory firm asking the question: What is the future of companies

Appendix: Links, Media, and Research

Primary Source

The founder mode and AI founder mode framework comes from a recorded interview with Brian Chesky, co-founder and CEO of Airbnb. The interview covers his industrial design training, the 2019 realization that the company had grown beyond his control, the move into founder mode during the pandemic, and his early thinking on what AI founder mode becomes.

The term itself comes from Paul Graham, who named it after a Chesky talk at Y Combinator.

• Paul Graham. “Founder Mode.” paulgraham.com/foundermode.html

Research Cited

• Harvard Business Review, February 2026. “To Thrive in the AI Era, Companies Need Agent Managers.” The argument that the AI era produces a new role, the agent manager, the way the software era produced the product manager. hbr.org

• Harvard Business Review, May 2026. “Research: Why You Shouldn’t Treat AI Agents Like Employees.” A large-scale experiment finding that placing AI agents on the org chart as employees reduced accountability, raised escalation, and lowered review quality. hbr.org

• MIT Sloan Management Review. “Agentic AI at Scale: Redefining Management for a Superhuman Workforce.” sloanreview.mit.edu

• MIT Sloan Management Review. “Want AI-Driven Productivity? Redesign Work.” The finding that AI produces a return only when the work is redesigned around it. sloanreview.mit.edu

• MIT Sloan Management Review. “Stop Deploying AI. Start Designing Intelligence.” The case against layering AI tools on top of old workflows. sloanreview.mit.edu

Supporting Data

• Bayer’s management-layer reduction, from roughly twelve layers to six and the removal of close to 5,500 roles, is drawn from 2026 reporting on the company’s “Dynamic Shared Ownership” reorganization. It is cited here as a concrete example of the management-layer flattening that Harvard Business Review and MIT Sloan Management Review describe.