Hey folks, Kamayani here.
I’ve been thinking about a slightly strange question lately: what happens when one person can do the work that once needed a whole team?
We’ve all had that moment where it feels easier to just do something ourselves than explain it to someone else. AI is making that instinct a little more interesting. It can write code, make the first draft of an ad, answer customer questions, pull together research and keep track of what’s happening across a business.
That doesn’t mean one person can suddenly run everything. But it does make me wonder how much more one person could take on when AI starts working alongside them - and what that could mean for how companies are built and run.
There are about three months left in the year, so I went looking for how close anyone has come. Nobody has cleanly crossed the line yet, but two people have gotten close enough.
Matthew Gallagher launched Medvi, a telehealth company selling GLP-1 weight-loss drugs, from his home in LA in September 2024, with $20,000 and more than a dozen AI tools, which he used to write the software, make the ads and answer customers.
His only full-time hire is his brother Elliot, and he leans on contractors for the rest, so technically it's a company of two (I'll allow it).
It has also never raised money, so it has no official valuation.
The other is Polsia, which Ben Cera, a very early employee at Travis Kalanick's CloudKitchens, launched in mid-December last year.
You give it a business idea, and a set of AI agents builds the product, runs the ads, answers support and emails you every night about what it did. (Polsia is "AI slop" spelled backwards, and he has said that was deliberate.)
On 26 March, Fortune reported it had reached a $4.5 million revenue run rate with Cera as the only employee, and about two months later, he announced a $30 million raise at a $250 million valuation, still with no one else on the payroll.
What amazed me more was what he said when an interviewer pushed him on the numbers.
About half of Polsia's customers leave in their first month, and by his last check about half of the rest left in the second. Of the more than 8,600 companies it was running, roughly 10% had made even a dollar, and the best one he had seen was making $3,000 to $4,000.
So, at least among the examples I could find, the highest-valued company of one is selling the idea of a company of one, and while AI can already do the coding, ads and support inbox, the customer is still the hard part.
Going solo has become normal.
Stripe says solo founders made up 63% of the C corps formed through its Atlas service so far in Q2 2026, an all-time high, and Carta found the share of new startups with one founder rose from 23.7% in 2019 to 36.3% in the first half of 2025, though solo-led startups founded in 2024 got only 14.7% of the cash raised in priced equity rounds that year.
The same Stripe data shows the top tenth of solo founders made 61x what the median one did in their first six months in 2025, up from 34x four years earlier, so more people are starting, and the gap at the top keeps getting wider.
The ones at the top usually arrive with something the agents can't give them.
Tibo, a French founder who had already sold Taplio and Tweet Hunter, met Eugene, a self-taught developer whose AI tool for SEO articles was stuck at around $300 a month, and they repositioned it as Outrank and launched it publicly in January 2025.
"Solo founder" tells you who owns the company, and says very little about what they knew before AI showed up.
I suspect India's version of this won't mostly be a 22-year-old clicking "surprise me" on a tool like Polsia. It will be a business that already exists, with customers, suppliers and a bank balance, run by the next generation of the family that built it.
Siddharth Soni
Siddharth is 23, and his family has made jewellery in Hyderabad for decades.
In 2023 he co-founded Idea Jewellery, which designs pieces with AI and makes the moulds on a 3D printer, and a piece that used to take six or seven months now takes him about a week to make, before hallmarking.
I'd underline the 20 retailers he is in talks with already buy from his family's business. That is what most of Polsia's companies are missing, and a family business already has it.
When Amodei made his prediction, his first guess for where the one-person billion-dollar company would come from was proprietary trading.
The first serious contender turned up in healthcare.
On 20 February 2026, six weeks before The NYT profile, the FDA sent Medvi a warning letter saying its website made false or misleading claims about the compounded drugs it offered, including claims about their source and FDA approval.
By Gallagher's own account in that profile, his customer-service chatbot also made up drug prices, which he honoured.
Somebody still has to check, and the gap I care about is that nobody is training the person who sits above the agents, the one who decides what they're allowed to do, notices when they get it wrong, and answers for it.
Which brings me back to the group project.
For a lot of you, the most interesting company of one may be the business your family already runs, where the customers and the cash flow exist and the departments could get much smaller.
Running it that way needs someone who knows the business well enough to decide which calls an agent can make on its own. That job doesn't have a title yet, and I wanted to put it in front of you while it's still being figured out.
The people already doing this look a lot like you. Siddharth is 23, and Eugene was building products on the side before teaming up with Tibo.
