Ten Years Ago You'd Have Hired
Now you don't have to. The constraint that shaped every startup for thirty years is gone. Most founders are still building like it's there.
Mike Molinet & Govind Kavaturi
Last week we talked about models are commodities now. The race moved to application. Nobody cares what model you use. They care what it does for them.
This week: what a company looks like when you start it today.
The Old Way
Ten years ago, starting a company meant one thing above all else: headcount.
You had an idea. You needed a developer. Maybe one of you could build, but not fast enough, and not while also doing everything else. Then you needed someone to sell it. Then someone to support the people who bought it. Then someone to write the words that got people to look at it in the first place.
Every function was a person. Every person was a search, a pitch, an offer, an onboarding, a salary.
And you couldn't afford any of it. So you raised.
Raising took months. Then hiring took months. Then coordinating the people you hired took the rest of your life.
That was the job. Not building. Assembling a group of humans and pointing them in the same direction.
If you stayed small, you accepted the ceiling. Two or three people can only cover so many functions. You picked what you were good at and let the rest rot.
That was the tradeoff. Raise and coordinate, or stay small and slow.
The Constraint Is Gone
Companies starting today don't have that tradeoff.
Day zero, you can have a researcher. A product builder. Someone reading customer feedback. Someone handling support. Someone writing marketing. Someone running outbound.
Not eventually. Not after the seed round. Day zero.
Everything required to build and run a company is available before you hire a single person.
The constraint that shaped every startup for the last thirty years is gone. And most founders are still building like it's there.
What Is An AI Teammate
Here's where it goes wrong.
Most people hear this and think: I already do that. I have Claude open. I ask it things. It helps.
That's not a teammate. That's a chat box. Vol 23 covered why that gap matters. The chat box waits for you and hands everything back.
An AI teammate is something else. Four things make one:
A persistent role. Not a session you open and close. A seat that exists whether you're at your desk or not. The research teammate is the research teammate on Tuesday and still on Friday. It doesn't reset. It doesn't become the marketing teammate because that's what you happen to need right now.
An objective. A human hire gets a mandate. Grow the pipeline. Reduce churn. Ship the roadmap. An AI teammate needs the same. Not a prompt. A standing goal it works toward across sessions, across weeks.
Responsibility for an outcome. This is the one people skip. A teammate owns something. When it goes wrong, that's their thing to fix. If every output still lands on your desk for you to decide what to do with, you didn't delegate. You just got a faster typist.
Context that carries. Memory is the past. State is the present. A teammate has both. It knows what happened last month and what's happening right now. Monday doesn't start with re-explaining the company.
Persistent role. Standing objective. Owned outcome. Carried context.
Miss any of the four and you have a tool. Hit all four and you have a teammate.
The Difference Is Not The Model
Same Claude. Same GPT. Same API.
A chat box and an AI teammate can run on identical intelligence. What separates them is everything around the model. The harness.
The teammate has a role written down. Access to the systems it needs. A place its work lives that other teammates can read. Triggers that wake it up without you. Handoffs to the next teammate that don't route through your clipboard.
The chat box has a text field.
You're not waiting on a smarter model to make this work. You're waiting on yourselves to set it up.
Staffing, Not Prompting
The mental shift is from prompting to staffing.
When you hire a human, you don't hand them a sentence and hope. You write the role. You define what they own. You give them access. You tell them who they work with and how the work moves between them. You set what good looks like.
Do that for AI teammates.
Who owns research? What do they own, exactly? Where does their output go? Who picks it up next?
Who owns customer feedback? What triggers them? What do they do with what they find? Who hears about it?
Who owns support? What can they resolve alone? What comes to you?
This is org design. It's the same work founders have always done. The only thing that changed is who fills the seats.
Build the agent org covered the structure. This is about doing it before you have a company, not after.
What This Actually Changes
The old model: raise money, hire people, coordinate people, maybe build something.
The new model: staff the roles, build the thing, find out if people want it, then hire humans where humans are the answer.
You skip the part where you spend eighteen months assembling a team to test an idea that takes three weeks to test.
You get to the real question faster. Do people want this? That's the exercise that matters, and everything before it was overhead.
And when you do hire humans, you hire them into a company that already runs. Not into chaos.
Where Humans Still Belong
This is not "you never hire anyone."
Humans belong at the edges. Intent. Verification. Exceptions. That was true for workflows. It's true for companies.
Someone has to decide what the company is. Someone has to sit across from a customer and hear the thing they won't type into a form. Someone has to make the call when the stakes are real and the data is thin.
The terminal won't reject you. The market will. AI teammates don't get you out of that. They get you to it faster.
The point isn't a company with no humans. It's a company where the humans do human work from day one, instead of spending the first two years doing work that no longer requires a person.
The Close
Ten years ago, the org chart was a hiring plan. Every box was a person you had to find and pay.
The boxes are still there. The company still needs research, product, support, marketing, sales.
What changed is who fills them on day zero.
Most founders are still treating AI like a tool they reach for. Open a tab, ask a question, close the tab. Then they go back to wishing they had a team.
You have one. You just have to staff it.
Stop opening sessions. Start defining roles.
Vol 1: The opportunity exists. Economics changed.
Vol 2: Smart operators doing dumb work. Patterns to spot.
...
Vol 24: Models are commodities now. The race moved to application.
Vol 25: Ten years ago you'd have hired. Now you don't have to.
You have a team. You just have to staff it.
Mike & Govind
This week: Write your org chart. Every function your company needs. Then for each one, ask whether it's a session or a teammate. Persistent role, standing objective, owned outcome, carried context. If it's missing all four, you don't have a teammate. You have a tab open.