Hey y’all. Sam Altman says his tech CEO group chat has a betting pool on the first year a one-person company hits a billion dollars in value.

That used to be a thought experiment. Now it's a… market? And whether or not anyone collects on that bet, the question underneath is a practical one right now: how much company can one person actually run?

My own business isn’t a one-person startup, but I am a solo founder with a small, fully part-time team. We could fit at a single table.

So this week we’re discussing how big a one-person startup can realistically be, and whether (and when) to still make hires.

The receipts are real:

For context, great SaaS companies used to brag about $200-300k per employee. It is legitimately crazy but the best AI-native companies are already running at 10x that.

Headcount used to be a milestone. Even though it was misguided (bigger tends to equal more complexity more than it equals “better”) the tradeoffs were necessary because serving more customers effectively was impossible without somewhat linearly scaling headcount.

Now it's a choice.

The truth is more nuanced than “agents replace humans.” It’s more like “agents replace tasks with a definition of done while humans own ambiguity.”

If you can write one sentence describing what finished looks like, that work is a candidate for an agent to own: support triage, QA passes, research briefs, outbound drafts, bookkeeping prep, content repurposing… this is most of what junior hires used to do, and it's why the bottom rungs of org charts are disappearing.

What agents don't replace is the work where deciding what "done" means IS the job.

When an NYU Stern professor ran an experiment with Microsoft having MBA students build startups entirely with agents, he found that they’re great at discrete tasks, but no substitute for specialized judgment. His warning was that you're "taking it on faith" that the output is good, because nobody in the room knows better.

That matches my experience exactly. Our agents ship real work every day, and the work still gets reviewed by a human who knows what good looks like. Reviewing and editing and giving feedback on outputs literally is the job now, in some cases.

Everyone running this type of experiment hits the same three walls, in roughly this order:

Let’s apply this practically.

If agents eat the bottom of the org chart, what do your first five hires look like?

When you’re considering hiring, write the role as agent workflows first. Whatever's left over is the actual job description. If nothing's left over you didn't need the hire yet.

So how big can a company of one get? Bigger every quarter.

But the honest answer is that "company of one" is the wrong target. There’s no revenue in ego-based decisions. No one really cares if your company is run by one person or a small team as long it’s, you know, working well for customers.

The right target is minimum viable team to build the “something” people want from you.

Today, before the week starts, list out every distinct thing you personally did last week. Try to write it fully in one sentence, including what “done” looks like. Then, turn that list into a backlog of tasks to delegate to an agent. Ask your team to do the same.

Advertise in my newsletter to get in front of 60,000+ founders.

Founders: treat yourself for once.

A few of our partners are offering $$$ for taking a quick call with them (that’s it).

Honestly, block off a couple hours, knock them all out, and walk away with some cash to get off the ramen diet for the weekend.