You are past capacity. The trade press says agents can replace a whole staff. Your instinct says hire someone. Only one of those three statements is a decision, and it is not the one you think.
A dispatch on the staffing question — the one every solo operator hits somewhere between the second good year and the first burnt-out quarter.
TL;DR — the answer first: most independent operators reach for a hire when the real problem is an undocumented system, and most of the rest reach for an AI agent when the real problem is that nobody is accountable while they sleep. The decision is not human-versus-agent. It is a sorting job: work that needs judgment stays with you, work that needs accountability needs a person, and work that needs neither — assembly, drafting, formatting, triage — should have left your desk already. Hire only when the task has repeated for a quarter, you have written down how you do it, an AI system has already failed at it for a reason you can name, and the revenue it unlocks exceeds the person's cost plus your management hours. Four conditions, all of them, or you are buying yourself a second job.
Because the right answer depends almost entirely on where you are standing, here is the specific position this dispatch is aimed at.
You have twenty to thirty years in one field. You left, or were left by, a company that had your name on a door. You now run a service practice — fractional work, advisory retainers, project consulting — inside a niche narrow enough that people in it know your name. Revenue is real and roughly predictable. You are competent with AI in the way most experienced operators are competent with AI: you use it daily for drafting and thinking, you have not built anything that runs without you watching, and you suspect you are leaving something substantial on the table. You are working more hours than you left corporate to work, and the calendar is the binding constraint on every growth idea you have.
That is the position where the hiring question arrives, and it arrives disguised as a capacity problem. It is usually not one.
The instructive version of this is not a story about a solo practice at all. It is a story about what happens when you answer the capacity question by adding people, before the systems exist.
In The Mentor Economy, Chapter Four lays out what it cost to build the original MentorMe platform in the years before AI: a copywriter at $3,600 a project, a web developer at $7,000, a campaign manager at $3,000 a month, an SEO specialist at $4,000 a month, a video team at $15,000. Then the freelancers hired to stretch the budget on top of that. The line that matters is not the total:
We spent more time managing the people we hired to help us build than we spent actually building. Reviewing drafts. Re-explaining vision. Catching mistakes. Approving revisions.
Read that as an equation rather than a complaint. Every person added to a project consumes founder hours in a fixed ratio to their output — briefing, review, correction, context transfer. Below a certain scale, that ratio is bad enough that adding capacity reduces the founder's effective output. The work gets done; the person who was supposed to be freed is now a manager, and management was not the job they wanted or the job that produces the revenue.
This is the trap the one-person business is specifically designed to avoid, and it is the trap most one-person businesses walk into on their first hire — because the hire happens at the moment of maximum overwhelm, which is precisely the moment you have the least capacity to train, document, or supervise anyone.
Sort every recurring task in your practice into three piles. The sort is the whole method.
Assembly. Work with a knowable right answer and a specifiable process. Formatting the deliverable, drafting the recap, summarizing the transcript, first-pass research, turning your notes into a structured document, chasing the scheduling. This pile should have left your desk already, and in 2026 it goes to software and AI systems rather than to people. This is the terrain I mapped in The 80/20 Delivery Layer.
Accountability. Work where someone must own an outcome across time, while you are unreachable, with authority to make small calls you have not pre-specified. A client who needs an answer on Thursday when you are on a plane. A vendor relationship that requires someone to chase it. A commitment that must hold whether or not you remember it. This pile is where a human still wins, and it is much smaller than most operators assume — because they file judgment work here by mistake.
Judgment. The call only your two decades produce. What the client's actual problem is, as opposed to the one they described. What to cut from the scope. When to tell someone their plan is wrong. This pile never leaves you, and the moment it does you no longer have an expertise practice — you have a small agency with your name on it, a different business with worse margins and a far worse failure mode.
The trade press has spent this year running the arithmetic of the first pile against staff salaries — a monthly tool subscription against a five-figure payroll — and concluding that the one-person company has arrived. The direction is right and most of those comparisons are published by people selling the tools, so treat the specific multiples as advertising rather than evidence. What the honest version of that argument establishes is narrower and more useful: assembly work no longer justifies a hire. It says nothing about the other two piles, and the other two piles are where the decision actually lives.
Chapter Ten of the same book describes what it looks like when a founding team gets this order right — building the system first, and only then adding people:
The team has since grown — but every hire is now positioned to amplify Oliver, not replace him.
The sequence in that sentence is the entire lesson. The system existed before the headcount, so each person joined a structure that already knew how the work was done. Hire first and you invert it: the person becomes the structure, the process lives in their memory, and you have manufactured a dependency you will pay for at the worst possible moment.
Before any hire — employee, contractor, part-time assistant — require all four. Not three.
The order matters as much as the content. Run conditions two and three first, in that sequence, and a meaningful portion of hiring questions dissolve before you reach the money — because the act of documenting the task reveals it was assembly work wearing a costume.
It asks you to sit in the overwhelm slightly longer than is comfortable.
The four-condition test is deliberately slow, and it is slow at exactly the moment you most want speed. That is not an oversight. A hire made under pressure is a decision made by the version of you with the least judgment available, and it commits you to a fixed cost, a management load, and a relationship you will find hard to unwind. Documenting a task takes an afternoon. Unwinding a bad hire takes a season and costs you a relationship you did not want to damage.
There is also a quieter thing worth naming. Many experienced operators want to hire because working alone is isolating, and a colleague is a reasonable thing to want after decades in buildings full of people. That is a real need and it deserves a real answer — a peer group, a collaborator on defined projects, a standing conversation with someone at your level. It is a poor reason to attach a salary to your business, because the loneliness returns in about a month and the payroll does not leave. The constraint you are usually feeling is the one I wrote about in The Bottleneck Is You, and it does not respond to headcount.
Do the sort. Document one task this week. Let the system take the assembly, keep the judgment, and hire a human only for the narrow band that genuinely requires a person to be accountable while you are not there. That band is real, and when you find it, hire well and hire deliberately. It is just far narrower than the exhaustion makes it look.
The full accounting behind this dispatch — what building with people cost before AI, and what it looks like when the system is built before the headcount — is Chapters Four and Ten of The Mentor Economy. Get your copy →
When the same constrained task has been repeating for at least a quarter, you have documented how you do it, an AI system has already failed at it for a reason you can name, and the revenue it unlocks is larger than the fully loaded cost of the person plus the hours you will spend managing them. If any one of those four is missing, hiring will feel like relief for about six weeks and then feel like a second job. The usual failure is hiring at the peak of overwhelm, which is exactly when you have the least capacity to train anyone.
Split the work by whether it needs judgment or accountability. Drafting, summarizing, formatting, research triage, first-pass structuring — an AI system handles these now, at a cost that is not close. Anything that requires someone to own an outcome while you are unreachable, hold a relationship, or make a call you cannot pre-specify still needs a human. The honest test is not "can AI do this task" but "am I willing to review its output every time" — because you will have to, and that review time is the real price.
Yes, and it is more possible in 2026 than at any prior point, but the constraint moves rather than disappears. Without staff, your ceiling is set by how much of your delivery can run without your attention. Operators who stay solo at high revenue almost always did one specific thing: they converted their judgment into a repeatable system that other things — software, agents, contractors on defined scopes — can execute against. Without that conversion, staying solo just means capping revenue at your own calendar.
The three things clients are actually buying: the judgment call, the relationship, and the standard. You can delegate the assembly of a deliverable and keep the decision about what belongs in it. Once the judgment itself is delegated, you no longer have a practice built on your expertise — you have an agency with your name on it, which is a different business with different economics and a much worse failure mode when the person carrying it leaves.
Ask what would happen if you doubled your capacity tomorrow. If the answer is "I would deliver twice as much of the same work," it is a capacity problem and a hire or an agent can help. If the answer is "I would still be the one deciding everything, just faster," it is a systems problem, and adding a person makes it worse — because every undocumented decision now has to be transmitted through conversation, which is the most expensive channel you own.
The management overhead, which is routinely underestimated because it never appears on the invoice. Briefing, reviewing, correcting, re-explaining context that lives only in your head, and the switching cost of dropping your own work to answer a question. A useful discipline is to estimate the hours you will spend managing the engagement, price those hours at your own rate, and add them to the quoted cost before deciding. Many contractor decisions that look obviously profitable stop looking that way after that addition.
Author of The Mentor Economy and co-founder of MentorMe. He writes about turning hard-won expertise into AI-leveraged one-person businesses.
The Mentor Economy is the full system — free, you just cover $9.95 shipping.
Get Your Free Copy