The Talk That Started It
In September 2024, Paul Graham published a short essay called "Founder Mode." Within days it was the most-discussed piece of writing in Silicon Valley, and "founder mode" had become shorthand, a compliment, and a punchline all at once.
The essay had a specific trigger. At a Y Combinator event, Airbnb co-founder Brian Chesky gave a talk. It was supposed to be half an hour and off the record. It ran for two hours. Graham, who was in the audience, later wrote that many of the founders present said it was the best talk they'd ever heard.
What made it land wasn't a growth hack or a fundraising story. It was Chesky admitting that he had run Airbnb the way everyone told him to, that it had gone badly, and that fixing it meant throwing out most of the advice he'd been given. Graham realized he'd heard versions of this from other great founders too. They'd all been told to run their companies a certain way, it had hurt, and they'd only recovered by trusting their own instincts instead.
The essay is Graham's attempt to name that pattern before anyone had studied it properly. He was explicit that he was describing something poorly understood, not delivering a finished theory. That humility is part of why it spread. Founders read it and felt seen.
What Founder Mode Actually Means
Graham frames the whole essay as a contrast between two ways of running a company: "manager mode" and "founder mode."
Manager mode is the version taught in business schools and passed down by professional executives. Its core instruction is simple: hire good people and give them room to do their jobs. You build an org chart, you delegate whole areas to the people who report to you, and you treat each of their divisions as a black box. You set targets and check outcomes, but you don't get into the details of how the work gets done. Reaching around a manager to talk to their reports is considered a violation. Skip-level meetings are so unusual they need a special name.
Founder mode rejects the black box. Graham's claim is that founders can, and often should, engage directly across the hierarchy. They talk to people several levels down. They stay close to the product and the details. They keep the kind of first-hand knowledge that lets them smell when something is wrong long before it shows up in a metric.
Graham is careful to say founder mode isn't fully mapped yet. But its shape is clear enough: "There are things founders can do that managers can't," he wrote, "and not doing them feels wrong to founders, because it is." The essay's whole emotional core is that last clause. Founders have been told their instincts are the problem. Graham says the instincts were often right.
The Advice That Nearly Broke Airbnb
The most quoted line of conventional wisdom is also the most dangerous, according to Graham: hire good people and give them room to do their jobs.
It sounds unimpeachable. Who wants to be the founder who won't let talented people work? But Graham reports what founder after founder actually experienced when they followed it. In practice, he writes, it too often turns into "hire professional fakers and let them drive the company into the ground."
This is what Chesky lived through. As Airbnb scaled past its startup phase, he was coached to step back, hire seasoned executives, and delegate. The company grew layers. Chesky lost touch with the details. Decisions he would once have caught slipped through. By his own account, the results were bad enough that he had to rethink the entire model.
The turnaround is the part that gives the essay teeth. Chesky didn't just complain about the advice. He studied how Steve Jobs ran Apple, rebuilt Airbnb around a single functional organization, and got back into the details himself. Airbnb's financial recovery followed, with free cash flow margins that became some of the strongest in tech. Graham's point is blunt: Chesky achieved this against a headwind of bad advice, not because of good advice.
Graham even describes the emotional experience of the transition. As manager mode takes over a growing company, he says, founders feel like they're being gaslit. Everyone around them insists they should behave in a way that feels, to the founder, obviously counterproductive. The essay gave those founders permission to trust the feeling.
What Steve Jobs Did Differently
Graham's clearest concrete example of founder mode comes from Apple. Steve Jobs, he notes, used to run an annual retreat for what he considered the 100 most important people at Apple. The crucial detail: these were not the 100 people highest on the org chart.
That single practice violates manager mode completely. In a pure org-chart world, you invite the top of the pyramid. Jobs invited the people who actually mattered to the work, wherever they sat. He reached past titles to get at reality.
The broader pattern is what people now call skip-level engagement. A founder in founder mode doesn't only talk to their direct reports. They talk to the person two or three levels down who's building the thing, because that's where the ground truth lives. Graham predicts that as founder mode gets better understood, "skip-level meetings will become the norm instead of a practice so unusual that there's a name for it."
Jobs is the archetype, but he isn't alone. Commentators quickly added Elon Musk and Jensen Huang as founders famous for staying deep in the details of their companies long past the point where convention says a CEO should have let go. Huang has around 60 direct reports and famously skips one-on-ones, sharing information with everyone at once instead. The pattern repeats: the founders most associated with sustained excellence tend to be the ones who never fully handed off their context.
Founder Mode vs Manager Mode
The two modes aren't just different tactics. They rest on different beliefs about where a company's judgment should live. This table lays out the contrast Graham draws, plus the risk each mode carries.
| Dimension | Manager Mode | Founder Mode |
|---|---|---|
| Core instruction | Hire good people, give them room, don't meddle | Stay in the details, engage directly |
| Org structure | Divisions as black boxes, strict chain of command | Functional org, fewer layers, reach across |
| Meetings | Direct reports only, skip-levels forbidden | Skip-level conversations are normal |
| Source of truth | Dashboards, summaries, status reports | First-hand contact with the work |
| Who it suits | Professional managers running mature units | Founders who built the thing |
| Failure mode | "Professional fakers" quietly sink the company | Micromanagement that demoralizes strong teams |
| When it shines | Stable, well-understood operations | High ambiguity, fast change, product-led bets |
Read across the bottom two rows and you get the honest version of the debate. Neither mode is free. Manager mode fails by losing touch; founder mode fails by never letting go. The skill is knowing which risk you're closer to right now.
The Micromanagement Trap
The loudest criticism of the essay arrived fast: founder mode is just a rebrand of micromanagement. It's a fair worry. Read carelessly, "stay in the details" is exactly what every controlling boss tells themselves to justify meddling.
The critics have a point, and the best response to it comes from inside the same tradition. In "High Output Management," Andy Grove introduced the idea of "task-relevant maturity," the level of experience someone has with a specific task. Ben Horowitz has built on it for years. The rule is simple and it dissolves the whole argument: when a person's task-relevant maturity is low, detailed, hands-on direction helps them. When it's high, that same involvement is counterproductive and insulting.
So founder mode and delegation aren't opposites. They're two settings on the same dial, and the founder's job is to read the situation and turn the dial correctly. Get deep on the new product line where nobody has done it before. Back off from the finance function that a seasoned CFO runs in their sleep. The founders who get this wrong in the meddling direction burn out their best people. The ones who get it wrong in the hands-off direction wake up to find, as Graham warns, that they've been handed a company running on fumes.
The distinction that saves founder mode from being an excuse is intent. Founder mode is about retaining context and judgment, not seizing control of decisions that others are better placed to make. Chesky's own framing is useful here: he describes it as being "in the details," and says "great leadership is presence, not absence." Presence is paying attention. Micromanagement is refusing to trust. They can look similar from across the room, but they don't feel the same to the people on the receiving end.
Founder Mode in the Age of AI
The essay landed in 2024, but its relevance has grown with the AI boom, and Chesky has been explicit about why.
His argument is that AI collapses the advantage of size. When small teams armed with capable models can move as fast as large ones, the winning move is to run your company like a startup again, no matter how big it's gotten. That means fewer layers, faster decisions, and a leader who's close enough to the work to redirect it in days rather than quarters. Chesky has said that in the age of AI you need to be founder-oriented, because you'll need to move like a startup to adapt.
He practices what he preaches. Despite Airbnb being worth tens of billions with thousands of employees, Chesky says he still runs it like a startup, staying personally involved in hiring, firing, promoting, and managing a core group of roughly 50 people. He wants far fewer layers of management than a company that size usually carries. The whole design is built to keep the founder's context intact as the company scales.
This is the counterintuitive upshot. You might expect AI to make deep human involvement less necessary, since models can summarize and delegate. The opposite case is stronger. When execution gets cheap and fast, the scarce resource becomes taste, judgment, and context, exactly the things founder mode is built to protect. The same logic runs through the emerging playbook for how founders build knowledge systems: the edge isn't doing more, it's knowing more of the right things.
How to Apply Founder Mode
You don't need to be a billionaire CEO to use founder mode. The underlying moves scale down to a two-person startup or a team lead. Here's the practical version.
-
Stay close to the actual work. Talk to the people building the product, using it, and supporting it, regardless of their level. Ground your view in first-hand contact, not just summaries that have been polished on the way up.
-
Normalize skip-level conversations. Make it clear you'll talk to anyone about the work, and that this isn't a threat to their manager. The goal is context, not a shadow chain of command.
-
Calibrate with task-relevant maturity. Before you get into someone's area, ask whether they've done this specific thing before. New task, get involved. Proven track record, set the goal and step back.
-
Protect the details that matter, delegate the rest. You can't be deep on everything. Pick the few areas where your judgment is the company's edge, usually product and the core customer experience, and go deep there.
-
Keep a written record of what you learn. Founder mode runs on memory and context. The founders who stay effective as they scale externalize what they know so it doesn't evaporate when the company grows. This is where a real learning system earns its keep.
That last point is where most people underinvest, and it's the bridge to the deeper reason founder mode works at all. The founders who apply it well aren't just present. They're relentless learners with a system for keeping what they learn, which is the same discipline behind do things that don't scale and how to do great work.
The Knowledge Base Behind Founder Mode
Strip away the org-chart arguments and founder mode is a claim about knowledge. The founder who stays in the details knows more, in a more grounded way, than a manager who reads reports. That knowledge is the whole advantage. The failure mode of manager mode, hiring "professional fakers," is really a failure of context: the founder no longer knows enough to tell a faker from a builder.
Which raises the practical question the essay doesn't answer: how do you actually maintain founder-level context as the surface area of your company explodes? You can't hold it all in your head. Great founders don't. They build systems for capturing and recalling what they learn, and those systems are boring on purpose.
This is exactly the problem a learning tool like Glasp is built for. When you read the essays that shape your thinking, from Graham's own writing to founder memoirs and market analyses, Glasp's web highlighter lets you mark the passages that matter and keep them in one searchable place instead of losing them to a closed tab. Chesky's ideas mostly live in long podcasts and conference talks; YouTube Summary turns those into structured, timestamped notes you can actually revisit, so a two-hour talk becomes something you can mine in ten minutes.
Over months, that saved reading becomes a personal knowledge base. With Glasp's AI chat you can question it directly, asking what you've collected about hiring, delegation, or scaling and getting answers grounded in your own sources rather than a generic model. And because Glasp is social, you can learn from what other founders and operators are highlighting in the community feed, turning solitary reading into shared context. Founder mode is about not losing touch. A knowledge base is how you stay in touch with far more than one person could otherwise hold. The same principle is why knowledge debt is a startup killer: context you fail to capture is a debt that comes due exactly when you need it most.
Frequently Asked Questions
What is founder mode in simple terms?
Founder mode is Paul Graham's name for a hands-on way of running a company where the founder stays close to the details and engages directly with people across the organization, instead of delegating whole areas and treating them as black boxes. He contrasts it with "manager mode," the conventional advice to hire good people and give them room. Graham argues founder mode is harder but often works better for founders.
Who coined the term founder mode?
Paul Graham, co-founder of Y Combinator, coined it in an essay published in September 2024. The essay was inspired by a talk Airbnb co-founder Brian Chesky gave at a YC event, in which Chesky described how conventional management advice hurt Airbnb and how studying Steve Jobs helped him fix it.
Isn't founder mode just micromanagement?
That's the main criticism, and it's a real risk. The difference is intent and calibration. Founder mode is about retaining context and judgment, not seizing every decision. Andy Grove's idea of "task-relevant maturity" is the guardrail: get hands-on when someone is new to a task, and back off when they've proven they can handle it. Applied without that judgment, founder mode does collapse into micromanagement.
Does founder mode only work for founders?
The instincts Graham describes are strongest in founders because they built the company and carry its original context. But the underlying practices, staying close to the work, running skip-level conversations, and calibrating involvement to experience, are useful for any leader. A hired CEO or team lead can borrow the toolkit, even if they lack the founder's built-in authority to use it.
How does founder mode relate to AI?
Brian Chesky argues that AI makes founder mode more important. As AI lets small teams move very fast, large companies need to run like startups to keep up, which means fewer layers and a leader close to the work. When execution gets cheap, judgment and context become the scarce advantage, and those are exactly what founder mode protects.
Conclusion: Presence, Not Absence
Founder mode struck a nerve because it defended something founders felt but had been told to suppress. The instinct to stay close, to know the details, to reach past the org chart, isn't a bug to be managed out of them as they scale. Handled with judgment, it's the edge.
The honest version keeps both truths. Manager mode fails by losing touch. Founder mode fails by never letting go. The work is reading the situation and choosing the right depth, again and again, task by task. As Chesky put it, great leadership is presence, not absence.
Presence runs on knowledge, and knowledge runs on a system. If you want to stay in the details of a world moving this fast, start by keeping what you read. Highlight the essays and talks that shape your thinking with Glasp, turn the long ones into notes you'll actually reuse, and build the context that founder mode depends on. The founders who apply the same discipline to ideas that they apply to product are usually the ones still in the room years later. For the mindset that pairs with it, read how to do great work next.