You take pride in your code quality. Your drafts beat anything the AI produces—yours come out clean the first time; the AI's always need work. So today, like most days, you spend two hours wrestling with payment integration yourself. Handing it off would mean more time in review, and besides, it feels wasteful to outsource work you're genuinely better at. Two hours later, the code is clean. And the first customer interview you'd scheduled? Pushed to next week. Again. You spent one more hour doing what you do best—and one fewer hour on the thing that determines whether you have a business at all.
That scene contains the trap solo founders fall into most often: holding on to tasks because you're good at them, and only delegating what you're bad at. Intuitively, it sounds right. But draw the line that way and you consistently spend your most valuable hours on your least consequential work. The delegation criterion needs to change.
Delegating What You're Bad At Is Only Half the Answer
The first answer most people reach for is: delegate what I can't do. Marketing founders hand off engineering and design; engineering founders hand off copywriting and accounting. That's not wrong—you're filling gaps. But it's only half right. The criterion leads to a conclusion that feels obvious: of course you handle the things you're good at. And that conclusion is exactly what created those two lost hours.
The missing half comes from a two-hundred-year-old trade theory. David Ricardo's principle of comparative advantage, first articulated in 1817, centers not on absolute skill but on opportunity cost. Even someone who is better at two things than anyone else cannot do both simultaneously. Hold one and you give up the other. So the question isn't who does it better—it's what you're sacrificing while you hold on. The principle was built to explain trade between nations; now the trading partner has simply changed from a country to an AI.
The Better You Are at Something, the More It Costs You to Keep It
Back to the developer whose code quality outpaces the AI. In absolute skill, the human wins. Delegating coding means a small drop in quality, plus time spent reviewing the output. By the old criterion—delegate what you're bad at—coding obviously stays in your hands. But comparative advantage looks at the adjacent column. That same coding hour could have gone to product strategy or a customer interview. If an hour of strategy creates more value than the gap in code quality, then handing the coding off and keeping only final review expands the total output of the business.
That's where the criterion flips. The right question isn't "Is the AI better than me?" It's "What am I giving up while I hold on to this?" The better you are at something, the more you want to keep it—but the more valuable your hour is, the more valuable the thing you're not doing with it. The real price of two clean hours on code isn't a freelancer fee or a software subscription. It's the first customer you didn't meet. The tasks you grip tightest because you're proud of them are often the ones with the steepest opportunity cost.
Labor economics has already mapped which tasks are least costly to surrender. David Autor and colleagues established that technology substitutes for routine tasks—work whose steps can be written down as rules—while complementing non-routine cognitive tasks: judgment, creativity, contextual interpretation. What's safe to delegate is what can be documented as a procedure. What stays with you is what requires a call. And routine tasks—the ones that produce similar results regardless of who performs them—are precisely the least precious ones. The boundary comparative advantage draws by opportunity cost and the boundary labor economics draws by task type land in exactly the same place.
Field evidence backs this up. In an experiment Erik Brynjolfsson and colleagues ran observing 5,179 call-center agents, introducing generative AI raised average productivity by 14%. The gains were largest—34%—among new and lower-skilled workers. For a solo founder, the implication is direct: the payoff from AI delegation is biggest where you're weakest. If your background is marketing, you stand to gain more from putting AI on your engineering, design, and accounting tasks—the domains where you're still a rookie—than from automating the marketing you've already mastered. The "least precious tasks" that comparative advantage identifies and the "tasks you're a beginner at" that the field data highlights point in the same direction. A two-hundred-year-old trade theory and recent call-center data arriving at the same conclusion from entirely different starting points—that convergence is only visible if you look at both.
This division of labor doesn't reduce the total amount of work—it shifts its composition. As Daron Acemoglu and Pascual Restrepo have shown, automation displaces humans from existing tasks while simultaneously generating new ones. When you hand off drafting, the freed space fills with new work: building review criteria, refining prompts, curating among competing outputs. The work that shrinks is procedural; the work that grows is judgment-based. Wherever you let go of the least precious tasks, the more precious work of decision-making moves in.
Three Ways to Redraw the Line with Comparative Advantage
First, write next to each task what you couldn't do while doing it. Pull up the last two weeks of work and, beside each item, note what got pushed aside. Next to two hours of coding, write: customer interview. Next to reconciling accounts, write: product roadmap. Any task whose opportunity-cost column names something more valuable becomes a delegation candidate—even if you're the better performer. This is comparative advantage's arithmetic, reproduced on a single sheet of paper.
Second, flag the tasks you're keeping because you're good at them. The work you've already let go because you're bad at it is gone. The real question is what you're still holding because you're proud of it. Mark every item in your list where the reason to keep it is "I do this best," then identify which of those can be written down as a documented procedure. If a task is both something you excel at and something that can be scripted, "I'm good at it" is not a reason to hold on. That task is your first delegation candidate.
Third, start putting AI on your weakest areas first. If you're unsure where to begin, look at the domains where you're the novice—that's where the field data shows the gains are largest. One caveat: if you've never done something yourself, you won't be able to judge what the AI returns. So spend enough time doing it once to distinguish a strong output from a plausible-but-wrong one before you hand it off. Starting to delegate in weak areas isn't about eliminating work; it's about building the critical eye to review it.
From Productivity to Profitability
Redrawing the line with comparative advantage isn't just about saving time. The point is to pour the hours you free up into work where judgment compounds and customers multiply. Two hours nudging code quality up by a fraction and two hours understanding a first customer's real problem are not the same two hours. The first produces one artifact. The second sets the direction of every artifact that follows. Trading some of your best technical work for more of your highest-stakes strategic work is the first step in turning faster execution into actual revenue. The moment you shift the delegation criterion from "what I'm bad at" to "what I can least afford to keep," the division of labor crosses over from a productivity tool to a profitability one.
This series works through one manuscript, one installment at a time—drawing from Running a Company Alone, a book that reframes standard concepts from accounting, economics, management, and investing as tools for the solo operator, without the usual disciplinary borders. Next, we'll look at how the task map you build by applying these principles can serve as an org chart for a company with no employees—and why that boundary line has to be redrawn every quarter. If you've ever felt your business stalling because you couldn't stop doing the work you're proudest of, start here: put a sheet of paper in front of you and write, beside each task, what you gave up to do it.
Concept Notes
- Comparative Advantage — Articulated by David Ricardo (1817). Regardless of absolute skill, both parties gain when each specializes in the activity with the lower opportunity cost and then trades. Replace "nation" with "AI" and the principle applies directly to how you and an AI system should divide tasks.
- Routine vs. Non-Routine Tasks — Established in labor economics by David Autor, Frank Levy, and Richard Murnane. Technology substitutes for routine tasks—those whose steps can be written as explicit rules—while complementing non-routine cognitive tasks that require judgment and contextual interpretation. This distinction is the practical guide for deciding what to hand off and what to keep.
- Task-Based Automation Framework — Developed by Daron Acemoglu and Pascual Restrepo. Automation displaces human labor from existing tasks while simultaneously creating new tasks. Delegation therefore changes the composition of work rather than simply reducing its volume.
