When The E-Myth Revisited was first published, it named a failure mode so common it felt almost universal.

Small businesses fail, Michael Gerber argued, because they are built by technicians. People who are good at doing the work assume that running a business is simply doing more of it. What they end up with is not a business, but a job they own.

The remedy was structural, not motivational. Define roles. Document processes. Design the business as a system that can operate without the founder's constant involvement. Gerber framed this as a shift between three internal personas — the technician who does the work, the manager who organises it, and the entrepreneur who designs the business itself. Most founders, he observed, are trapped in the first role and rarely reach the third.

I used these principles when I built Vertical Leap. The org chart exercise, the process documentation, the gradual transfer of work from founder to team — it worked. The business ran without me, and eventually sold. So when I say the E-Myth still holds, I mean it literally. I built on it. If you want the playbook for that endgame in one sitting, John Warrillow's Built to Sell is where I'd start.

What has changed is the environment in which it operates.

What the E-Myth actually solved

At its core, The E-Myth is a book about labour substitution. It assumes that work is done by people, that people are costly and variable, and that growth means replacing the founder with other humans who can execute defined processes. The famous E-Myth exercise — the org chart with the owner's name in every box — is a tool for gradually transferring work from the founder to employees, role by role.

In a world where every role required a human, every decision consumed attention, and every handoff introduced friction, this was exactly the right abstraction.

AI changes that abstraction.

AI doesn't replace people — it replaces the question

Most conversations about AI in small business focus on tools: marketing, admin, support, analysis. That framing misses the deeper shift.

AI doesn't just make labour cheaper. It changes what delegation means. Tasks that once required memory, pattern recognition, coordination, and first-pass judgment can now be handled by systems that do not tire, forget, or wait for permission to act.

This makes the traditional E-Myth question — "Who will eventually sit in this role?" — insufficient. The more useful question is now: "What kind of judgment does this role require, and where should that judgment live?"

That is not a tooling question. It is an architectural one.

The technician trap didn't disappear — it evolved

One of the ironies of AI is that it makes the original E-Myth failure mode easier to miss.

The modern technician is no longer just the baker or the plumber. It is the founder who knows how to prompt, knows how to wire automations, knows how to "get results" from AI tools — and therefore stays central to everything. They are still doing the work, still holding the logic in their head, still acting as the bottleneck. Only now the work looks strategic, so the dependency is harder to see.

AI doesn't eliminate the technician mindset. It upgrades it.

Systems are no longer just procedures

Gerber's systems were procedural by necessity: step-by-step instructions, checklists, scripts. These worked because humans need explicit guidance to behave consistently.

AI systems do not operate that way. They require objectives, constraints, trade-offs, and escalation rules. You cannot hand an AI agent a checklist and expect coherent behaviour across novel situations. You have to define what matters more when goals conflict, when speed beats accuracy and when it doesn't, and what outcomes are unacceptable even if they are efficient.

This shifts the founder's work again. Instead of documenting what to do, the founder must encode intent — the principles that govern decisions when no one is watching.

What this looks like in practice

To make this concrete, consider a three-person consultancy.

In a classic E-Myth framing, the founder documents sales, delivery, and admin. Junior staff execute defined processes. Consistency is the goal, and the system's job is to make human effort predictable.

In an AI-first version, the shape of the work changes. AI drafts the first version of client proposals, pulls together background research, and summarises meeting notes. A junior consultant reviews the output, adjusts the framing, and flags anything that feels off. The founder no longer writes proposals from scratch — but they did decide that proposals must always state what the firm won't do, that research summaries must surface conflicting evidence rather than smooth it over, and that any recommendation involving redundancies gets escalated to a human before it leaves the building.

None of that is documented as a step-by-step procedure. It's encoded as policy: principles the system follows, constraints it operates within, and tripwires that pull a human back into the loop.

The bottleneck is no longer execution. It is judgment design.

The uncomfortable implication

If The E-Myth were written today, the central danger would not be founders who refuse to delegate.

It would be founders who automate without intent, delegate cognition without constraints, and allow systems to optimise locally while drifting globally. The old failure mode was "I must do everything myself." The new one is "The system is doing everything — but I no longer know why."

That is not a smaller risk. It is a larger one.

Where this series is going

The E-Myth taught founders to move from technician to manager to entrepreneur — to work on their businesses rather than in them.

In an AI-first world, that work has shifted again. Founders are now responsible for designing bounded intelligence, deciding where judgment belongs, detecting assumption drift, and holding accountability for system-level outcomes. The entrepreneur role doesn't disappear. It becomes more demanding, not less.

This series revisits Gerber's core ideas with that shift in mind — not to discard them, but to follow them to their logical conclusion.

In the next article, we'll return to the technician, manager, and entrepreneur framework and show why those roles no longer map cleanly to people — but still map precisely to capabilities.