New Founder Archetypes in the AI Era: One Person Can Start, but the Million Customer Meetings Are Still There

a16z's Andrew Chen recently wrote an article about what kinds of founders can break out in the AI era. At a16z's Speedrun program, he writes the first check to new companies, up to $1 million, and looks at founder backgrounds every day. His observation: AI has lowered the initial barrier to starting a software company, but the demands on founders have changed—the barrier is lower, but the test is different.

He lists several emerging founder types and gives the case for and against each one.

Solo founders

One person doing 100x the work—this is the most direct change AI has brought. A v1 that used to require an entire team can now be shipped by one person.

But on the flip side: co-founder teams may still move faster; agents can write code, but no one can attend a million customer meetings for you; and from the very beginning, code may not even be the main bottleneck for a startup.

Non-technical founders

Finance and consulting are full of smart people who don't write code but have sharp business instincts. People in creative industries create all day long—they just don't code. Now they can all build things themselves.

But there's still a chasm between a demo and a reliable product. When everyone can build a first version, technical judgment becomes more valuable—deciding what to build, what's feasible, and how to stay ahead as complexity grows. This is especially true for categories like developer tools.

Domain experts

Doctors working in hospitals understand the problems they face every day best; real estate agents understand the market and customers best. Same for military personnel and teachers. What if they could use software to solve their own problems and build a business around that?

The counterargument: hotel operators didn't build Airbnb, and the finance industry didn't build Kalshi. Disrupting an industry often requires an outsider's perspective, not incremental improvement within the existing system.

Kids, high schoolers, college students

Young people who grew up playing Minecraft mods, Roblox games, and Discord bots have already learned how to make things used by thousands of people. With AI, they can do more, earlier.

But something your friends like and building a company are two different things. Hiring, fundraising, managing, and making money don't go away. Young people understand their peers when it comes to consumer products, but in industries like enterprise software and healthcare—where experience and relationships matter—their advantage is limited.

Content creators

They already create content, and now they can apply the same skills to software. And they have distribution: they know what their fans care about and how to get attention without buying ads. In the past, turning an audience into a product meant finding a technical co-founder or hiring a team; now creators can build it themselves and push it directly to millions of potential users.

The counterargument: an audience provides an initial distribution spike, but a sustainable growth engine is another matter. The most valuable products aren't memes—they require long-term iteration and long-term growth. Is the DNA of a content creator suited to that?

The comments section was even better than the article itself.

Several people mentioned the same thing: AI can't replace meetings with customers. One commenter said code is no longer the bottleneck, but talking to customers eats up most of the week. Others would rather open Claude Code and write code than ask users what they actually need.

Another frequently raised issue: a solo founder has no dissenting voice. The agent nods at everything; no one tells you the idea sucks. Someone suggested treating building in public as an external feedback mechanism and letting Reddit be the one to say 'no.'

One commenter was more radical: founders don't need investors now. Adding engineering staff only slows you down; marketing should start guerrilla-style; buying ads is easy; creating content is basically free. A company can quickly get to positive cash flow, pay its founders salaries, put the rest into marketing, and stay small and lean. Unless it's hard tech, seed funding is more like a liability than a blessing.

Not everyone will agree with that view.

The code bottleneck is disappearing, and other bottlenecks are emerging. For example: who's going to attend those million meetings.