Every time I’ve watched a company get built from the inside, the same thing happens. The founders pick a product direction, start shipping, and for a few quarters it feels like the hard part is the product.
Then the surface area starts to open up.
Formation paperwork. Banking. Payroll. Equity ledger. Invoicing. Vendor contracts. Brand. Domains. Positioning. Website. Analytics. GTM motion. Hiring loop. Offer letters. Onboarding docs. Compliance posture. Security questionnaires. A board deck nobody has time to build. A founder’s calendar that was supposed to be for product and is now 40% unblock-everyone-else.
The product you can see. The company you can’t until it’s eating you.
I’ve watched this movie three times
At Weedmaps we scaled from 30 to 300+ engineers through IPO. At ShopKeep we sold for $550M. At BuildOps I watched a unicorn take shape from the inside. Three very different companies, three very different stages, and the same structural pattern at every one of them.
The thing nobody has a named role for is what breaks first.
Not the product… never the product. The product tends to have a team, a roadmap, and an Agile board. It has an owner! What doesn’t have an owner is the second category of work. The founders assumed they’d “figure out” because every founder before them also figured it out.
It’s the non-product surface area of running a company, and for most of tech history, the fact that it existed was so obvious nobody named it.
The cognitive overhead tax, revisited
I’ve written before about the cognitive overhead tax which is the mental background processes that leak founder attention in every direction at once. Tool fatigue is one flavor of it, but the non-product surface area is another, and it’s the heavier one.
Tool fatigue costs you a Tuesday afternoon.
Non-product surface area costs you a company.
The way the industry has solved this for the last two decades is people. You raise money, partly to build the product, and partly to hire the humans who will absorb the surface area so the founders can stay on the thing only founders can do. COO. CFO. Head of People. Head of Marketing. Controller. General counsel on retainer. An agency for brand, another for PR, a third for paid. Every new hire is a bet that absorbing one slice of surface area is worth a seat at the table.
It’s not a bad solution. It’s just expensive, slow, and, if you’re honest about it, a little old school. The act of assembling the team that can absorb the surface area is itself a significant slice of the surface area.
Why I’m writing about this now
Because the ratio has changed, and almost nobody has updated their mental model for it.
For the last decade of my career I watched capable founders drown in the second category of work while the product sometimes waited. I identified it, I complained about it to other CTOs at dinners, and I occasionally tried to help a founder patch it with a better ops hire.
Over the last eighteen months I started noticing something different. The leverage a single builder can bring to it has quietly, and then suddenly, gone up by an order of magnitude.
Not because any one AI tool is miraculous but because the set of things a founder used to need a team, an agency, a legal retainer, and a controller for — formation, early finance, brand, first GTM motion, vendor selection, compliance posture — has collapsed into a set of problems that a disciplined builder with the right scaffolding can actually hold in their head.
That changes who gets to start a company!
What this is not
This is not “AI replaces the COO.” That’s the tokenmaxxing version of the argument and it’s wrong the same way lines-of-code thinking was wrong. Replacement isn’t the interesting number… capacity is.
I’ve been calling this Minutes Added to Workforce (the honest unit for AI value). The right question to ask about any AI investment isn’t “how many heads did this save.” It’s “how much more non-product surface area can one builder carry before the quality of their product decisions degrades.”
Answer to that, today, on April 21st, 2026: a lot more than a year ago! Enough more that the shape of what a single builder can start has changed.
What I’m doing about it
The short version is that once you see the gap clearly, the decision to stop writing about it from the sidelines and start building for it yourself gets very hard to argue with.
For now, two asks for any engineering leader reading this:
One: the next time you hear a founder in your network say “the product is the easy part,” believe them! They aren’t being modest they are describing the surface area problem and they don’t have a name for it yet. Help them with the name, help them size it honestly. The sizing is the first step toward owning it instead of being owned by it.
Two: when you evaluate your own AI adoption this quarter, stop asking what got replaced. Start asking what got carryable, how much more surface area can one of your builders hold before quality breaks. That ratio — capacity per builder — is the real number.
The product, in the end, was never ONLY the hard part. For the first time in my career, that company is getting a lot cheaper to build.