Donkeycorn: Grind Like a Mule, Party Like a Unicorn

AI is redefining rules of entrepreneurship and investment, enabling "donkeycorns" — low-headcount, high-revenue startups.

An old friend called me the other day. “We should start a startup,” he said. I could see why. We know each other well and our skills are complementary. However, what to start is a good question.

Before AI, that is, before ChatGPT broke into mainstream, I’d try to think of a good idea. Then I’d probably try to raise an angel round, then an early seed, followed by a seed round, aiming to get to Series A and beyond. To do that, I’d look at the size of the potential market and think about the story I’d tell to venture capitalists.

AI changed all that.

Henrik Werdelin

came up with a term “donkeycorn”:

It grinds like a mule and parties like a unicorn. It's a new type of AI start-up that makes more than 2 million dollars in revenue with no more than two staff.

I’m not fussed about specific numbers: it could be 1 person and a million dollars or three people and three million. But the idea is simple: low headcount and high revenues.

A donkeycorn?

Historically, achieving this was hard and rare. Now, AI is quickly changing the equation because AI-first software generation tools like Cursor allow building products in hours and days instead of weeks and months.

This has big implications for the investment landscape. On the founder side, why raise money to build a product and prove the PMF (product-market fit) if you don’t need to hire a team of software engineers to build it anymore? Sure, you’ll need a team of engineers to scale later if the product is successful, but not to find the PMF and get early revenues — possibly even achieving profitability.

On the investor side, why invest in a business without a solid traction if getting to PMF doesn’t require a large sum of money? And if there is traction, is it defensible if anyone can copy your product in a matters of hours?

But many founders will wonder why raise at all. While every founder out there learned how to pitch a billion-dollar vision to VCs, privately many admit that they’d be happy with a far more modest sum. Let’s say, money that allows you to never work again. That’s far less than a billion. Maybe that’s a donkeycorn or two.

Maybe the ease of building new products will be counterbalanced by the difficulty to defend them from copycats, but then it still means that founders who know how to learn and build very fast will have an edge over those who don’t.

Sean Ellis

recently wrote:

In a recent blog post, I mentioned that I likely wouldn’t want to be a founder again, primarily because I know just how difficult it is to achieve product-market fit. But with AI-generated code, the cost of experimentation drops dramatically—especially since engineers are often the most expensive team members in an early-stage startup. The speed of experimentation also shifts from several months to potentially just a couple of days. This completely changes the equation. My confidence in being able to find PMF quickly—so that I can then lean on my growth expertise—goes up dramatically.

I feel the same. A decade ago I bet my thirties to build one company, Makers, and learned along the way that while it’s possible, it’s hard. But if the cost of experimentation drops dramatically… maybe that changes the entire experience of entrepreneurship?

New entrepreneurship playbook

The old entrepreneurship playbook, the lean startup approach, says that we need to run lots of cheap experiments to see what customers might pay for. For example, build many landing pages and which ones do well. Then, build several minimal viable products (MVPs) to learn how users actually behave. Then, armed with this knowledge, commit to building a proper product that someone will pay for.

That’s a wonderful approach, but if AI makes software easy to build, doesn’t it make sense to build a bunch of products instead of landing pages? And then double down on those that make actual money and happen to be defensible¹?

This is a very new entrepreneurial landscape. I see entrepreneurs aiming for donkeycorns that pay dividends without VCs instead of founder building unicorns with VC cash hoping to sell them one day.

This will undoubtedly change the VC landscape. I agree with

Azeem Azhar

who writes:

Capital may pivot back to where it started: funding breakthroughs in hardware, materials science, advanced computing, biotech, and complex automation. These domains, where atoms matter as much as bits, are not so quickly bulldozed by fancy algorithms and agentic whatnots.

If you’re building a fusion reactor, go raise VC (unless you’re this Canadian guy who built a fusion reactor in his kitchen with a bit of help from AI), but many companies that until recently would have required VC money, should now bootstrap themselves.

Golden age of generalists and specialists…

I think we’re entering the golden age of generalists and a golden age of specialists. It’s a golden age of generalists because with powerful AI research tools and AI software development tools, a founder can quickly orient themselves in a new field and build products to see if they get any traction.

But it’s also a golden age of specialists. How so? If building things becomes cheap, then knowing what to build and how exactly it needs to be build suddenly becomes very valuable. This comes from knowing one particular area very, very well and using that knowledge to build innovative solutions in a defensible way.

So if it’s a golden age for both generalists and specialists, does is mean everyone will have it easy? No.

…as long as they have agency

In fact, it’s a golden age for people with agency who know how to take initiative, learn fast and adapt quickly. My friend Rachel Carrell recently wrote:

Agency > Intelligence
I had this wrong for DECADES

Our culture reveres intelligence.
There's a weird obsession with IQ.
But 'agency' is significantly more powerful - and much less discussed.

By 'agency', I mean people taking initiative, making decisions, exerting control over their actions and environment.
Someone with agency sets goals, pursues them, figures stuff out.
Someone with low agency is a passenger in their own life, waiting for external forces (luck, other people) to dictate what happens next.

Now AI is turbo-charging the importance of agency.
It's turning intelligence into a commodity.
For the first time ever in history.
This is a HUGE shift.

Rachel is right. Both agency and intelligence have always been important, but until recently it was okay to be smart and not have much agency. You could get a nice job at a big company, solve big problems with your smart brain and lead a safe and comfortable life. You could rely on other people’s agency to deal with uncertainty and change.

I think these times are over. Intelligence still matters, but agency becomes increasingly important because the world is changing faster and faster. Our ability to adapt, both as entrepreneurs and professionals in general, is only becoming more important.

Back to donkeycorns

I’m excited. When I talk to my founder friends and my coaching clients, I sense that the power is shifting away from people who know how to impress VCs with big pitches to people who know how to build fast, learn fast and move fast.

I’m excited to see founders around me suddenly focusing not on visions of unicorns, but on building businesses that make more money than they spend and pay it out in dividends, the old-fashioned way.

I’m excited to see entrepreneurs learning how to build a portfolio of donkeycorn shots instead of betting the next decade of their life on one venture scale company. There’s nothing wrong with that, to be clear, except that it’s not the only model on offer today.

As for me, I should give my friend a call back.

Found this interesting? Forward to your friends.

1. I write happen to be because I don’t think anyone has figured out what defensibility means in the new AI world. Data, community, brand, regulations… all these barriers still seem to make sense, but I also sense that some barriers are going away and no one really know how things might shift in the coming years.