Every founder has heard the advice to work on the business, not in it, and almost nobody has finished acting on it. The reason is honest enough: the work in the business is the work you are good at, and being needed feels like the point of the whole thing. But the job is the opposite of being needed. Your real job is to build something that runs without you, and the measure of whether you are doing it well is how little the day-to-day depends on you.

A founder’s job is to make the business able to run without them, which means doing the few things only they can do and designing how everything else gets decided. That is the whole job stated plainly, and it is smaller than most founders treat it. A veteran executive coach lists the duties that genuinely require the person at the top, the strategy and direction, the culture and values, the senior team, and the capital, and then says the quiet part: everything else about the job is optional. Optional does not mean unimportant. It means it does not have to be done by you.

This frame is old and solid. The work-on-not-in idea is Gerber’s, the observation that owners routinely do the wrong work and far more of it than the return justifies. The job was never to do all the work. It was to build the thing that does.

Keep the four duties only you can own. Delegate everything else, but delegate the judgment for it, not just the task. The clean line is not by seniority or by how important the work feels. It is whether the work requires you specifically, or whether it requires a standard that currently only exists in your head. The first stays. The second gets moved out.

Holding on past that line has a measurable cost. Wasserman’s founder’s dilemma, across 212 startups, named the tradeoff founders avoid saying out loud: you can maximize control or maximize value, and rarely both. Keeping every decision is choosing control, and the price of that choice is the larger company you could have built. The other cost is your calendar. Even at the very top, the study of how CEOs spend their time found them working 62.5 hours a week, on 79 percent of weekend days and 70 percent of vacation days. The job eats the calendar by default. The only way it does not is if you design it not to.

You can maximize your control over the company, or you can maximize your wealth from it. Rarely both.
Harvard Business Review, 2008

You move the judgment, not the to-do. Delegating a task is handing someone the thing to do. Delegating a decision is handing them how to decide it: the objective the decision serves, the constraints it has to respect, what good looks like, and the line where a genuinely hard call comes back to you. When that lives somewhere the team can reach, they make the call the way you would have, and it does not route back to your desk.

This is the part that has changed in the AI era, and it is the heart of it. Delegation used to mean documenting steps for a person to follow. Now the work is encoding judgment so a team plus AI can run the process without you. One operator who has spent a career inside the work-on-not-in idea puts the shift cleanly: the bottleneck is no longer execution, it is judgment design. The question stopped being who will sit in this role and became what judgment does this role require and where should that judgment live. That is a different job than the one the canon was written for, and it is the one in front of you now.

The bottleneck is no longer execution. It is judgment design.
Matt Hopkins, 2025

You keep control of the standard and give away the doing, which is the opposite of what most founders do when they delegate and then re-approve everything. Work bounces back for one of two reasons: either the standard for the decision was never made explicit, so the team had to guess and you had to correct, or the escalation line was drawn at every step instead of only at the hard ones. Fix both and the work stops returning.

The way to think about it is to communicate the why and the what, and let the team and the system own the how. As one serial founder frames the CEO’s job, you articulate the why and the what and let people figure out the how. Losing control is when the why and the what live in your head and the team is guessing. Keeping control is when the why and the what are explicit and the how runs without you. The first feels safe and bounces everything back. The second feels exposed for about a week and then runs.

So the test for any answer is not whether it gets the task done. It is whether it holds the judgment so a team plus AI can run the work without it routing back to you. That is the bar, and it is the one JynAI built Works to clear.

Works is an AI Business OS, the operations layer where the objectives, the standard, and the rules you would have carried in your head become how the work runs. A few of the ways that shows up:

• The decision is encoded, not just the step: Expert-Grade Workflows are built on real operating playbooks, so a process carries the judgment of an experienced operator, the decisions and the standard, instead of waiting for you to supply them each time.
• The team and the AI run it at the line you set: Work That Actually Ships runs in Copilot, Pilot, or Autopilot mode, so routine calls execute and only the genuinely hard ones escalate, the escalation line drawn where you want it rather than at every step.
• The why travels with the work, so it stops bouncing back: Learns Your Business keeps the objectives, customers, and context in one place every workflow and agent reads, so the team is not guessing the standard and you are not re-correcting.
• You keep control without holding the work: Receipts log and version every run, so designing how the work gets decided does not mean losing sight of how it went.

It is priced for a founder-led business, not an enterprise: the Pro tier is forty-nine dollars a month. And it is the shift we lived ourselves. At Machintel, the work moved off a couple of people carrying the surface area of six teams and into how the business runs.

Build a business that doesn’t need you. Get early access. Or see what your business still routes through you first.

If he was a good entrepreneur, his only job is to make sure he doesn’t have a job. You stop being the answer and become the one who designed how answers get made. The diagnosis behind this, why you became the constraint in the first place, is in the founder bottleneck, and the test that proves the job got done is what happens if you take two weeks off.

What does it mean that my real job is to make myself unnecessary?

Making yourself unnecessary means moving from being the answer to being the one who designed how answers get made. Four duties stay non-delegable: strategy, culture, senior team, and capital. Everything else gets moved out, and the measure of how well you are doing the job is how little the day-to-day routes through you. Stever Robbins names those four as the only ones the person at the top genuinely cannot hand off, calling the rest optional.

Am I holding on because the business needs me, or because I need to be needed?

This is the question worth sitting with, because the two feel identical from the inside and have opposite fixes. If the business needs you, that is a design problem you can solve by moving the judgment out of your head. If you need to be needed, that is a story you can choose to stop telling. Most founders who will not let go are carrying some of both, and naming which is which is most of the work.

How do I document the why, not just the what?

You write down the decision, not the step. For a recurring piece of work, capture the objective it serves, the constraints it has to respect, what a good outcome looks like, and the point where a hard call should come back to you. That is the why and the what. The how then runs without you, because the team and the system have what they need to decide rather than guess. This is the difference between delegating a task and delegating a decision.

Does making myself unnecessary mean my quality drops when I step away?

Only if you hand off the task without the judgment. Quality drops when the standard lived in your head and left with you. It holds when you encoded the standard into how the work runs, so the team and the AI apply it whether you are there or not. That is the whole point of moving judgment rather than just tasks, and it is the diagnostic behind experiments versus operations.

1. Harvard Business Review, Noam Wasserman, The Founder’s Dilemma, 2008
2. Stever Robbins, What Do CEOs Do? A CEO Job Description
3. Rosenberg Associates, Marc Rosenberg, E-Myth Revisited: Work ON Your Business, Not IN It, 2013
4. Harvard Business Review, Porter and Nohria, How CEOs Manage Time, 2018
5. Matt Hopkins, The E-Myth, Revisited Again, 2025
6. Obvious Startup Advice, Eric Marcoullier, The CEO Has Three Responsibilities, 2019
   
   

Keep reading:

• The founder bottleneck
• If I took two weeks off
• Why I spent a year and a fortune on AI
• Team Runs With Me