The standard advice about your first ten hires was written for a world that no longer exists. That world assumed you needed people to do the volume: a support rep to answer the tickets, a junior marketer to produce the content, a coordinator to move the work around, an analyst to pull the numbers. The early org chart was mostly a plan for adding hands.
AI does the volume now. Not perfectly, and not everywhere, but enough that hiring your first humans to do commodity throughput is spending your scarcest resources, cash, headcount, and your own attention, on exactly the work that has just been commoditized. The question for your first ten is no longer who do we need to do the work. It is where does human judgment create leverage once the volume is handled.
That reframing changes who you hire, in what order, and how many of them you need.
Hire for what AI does not have
Start from what AI cannot do, because that is where your headcount belongs.
AI has no ownership. It can produce work, but it cannot be accountable for an outcome. It does not get called when the number is wrong, does not carry the consequence, does not lie awake about the customer. Accountability is human, and a company runs on it.
AI has no taste. It generates plausible output at infinite scale, which means it generates plausible mediocrity at infinite scale unless someone with judgment decides what good actually is. The bar is set by a person, and holding the bar is now more valuable than clearing it, because clearing it got cheap.
AI has no relationships. The trust a customer extends, the credibility a partner offers, the read on a room that closes a deal: those attach to people. AI can support the relationship. It cannot be the party the other side is trusting.
Ownership, taste, and relationships are the three properties your first ten should be selected for. Bias hard toward them and away from raw throughput, because throughput is the thing you no longer need to hire.
The shape of the first ten
In practice, this makes your early team smaller, more senior, and more judgment-heavy than the old playbook produced.
You need an operator: someone who runs the business day to day so the founder is not the only load-bearing decision-maker. This is the highest-leverage early hire in almost any company, and it is more true now, not less, because someone has to own the system that all the AI leverage runs through.
You need a few builders who are AI-native, not a bench of doers. The right builder now ships five or ten times what they could a few years ago, because they wield AI as a force multiplier. Two or three of those are worth more than a room of people doing volume by hand, and they cost you less in management overhead, which is the hidden tax on every early hire.
You need someone who owns quality. Call it taste, call it standards, call it the person accountable for what good looks like. In a world where anyone can generate a plausible draft in seconds, the constraint moves from production to judgment, and judgment needs an owner.
You need customer-facing judgment: sales, success, the relationship roles where trust is human and the stakes are real. These do not collapse into AI, because the other side is deciding whether to trust a person, not a tool.
And you need someone who owns the data and the systems, because AI is only as good as what it operates on. This is the least glamorous hire and often the one that determines whether any of the AI leverage actually lands.
What you do not need early, in most cases, is the layer the old playbook filled first: a support team sized for volume, a coordination tier that exists to move work between people, a bench of juniors hired to do tasks. Those collapse into AI plus one accountable owner. Hiring the old layer and then bolting AI on top is how you end up paying for both.
A lever needs a fulcrum
There is a trap in all of this, and it is worth naming because it is expensive.
AI is a lever, and a lever needs a fulcrum. Drop AI-augmented people into an operation whose data is a mess, whose processes live in someone’s head, and whose systems do not talk to each other, and AI does not fix that. It amplifies it. You get faster production of work built on a broken base, which is worse than slow work on a solid one, because now the mess scales.
So some of your first hires exist to build the fulcrum: clean the data, document the process, make the systems usable. That is not overhead you tolerate before the real work. It is the thing that decides whether the leverage pays off at all. A company that hires for AI leverage without building the fulcrum has bought a lever and no place to stand.
The mistake to avoid
The single most common error is to hire the old org chart and add AI as a feature. You end up with a team structured for a world of human throughput, running expensive AI tools that mostly automate work you also hired people to do, and wondering why the leverage never showed up.
The first ten in an AI world are fewer, more senior, and more judgment-heavy than they used to be. Each one should be someone who wields AI rather than competes with it, who owns an outcome AI cannot be accountable for, and who is set against the parts of the business that got more valuable precisely because the volume got cheap.
Where to start
Ready to find out which of your first hires should be people and which should be leverage, and whether your operation is actually ready to stand on that lever? The Forge Assessment is the 30-day diagnostic that maps it, including an AI-in-ops opportunity map: where AI compresses cost or time, what is safe to automate now, and what has to be built first. $6,500. A ranked 90-day roadmap at the end. Book a discovery call →
Jason Bonito is the founder of Crucible76, a fractional operating partner practice helping scaling businesses find and remove the self-inflicted friction before someone else does. DATA · DECISIONS · GROWTH.