How AI Is Changing How We Evaluate Early-Stage Founders

The way we evaluate founders used to be simple. Pedigree. Previous exits. Ability to raise capital. Track record.

Those signals still matter. But they're becoming less reliable. And AI has changed what matters instead.

The old founder playbook

For twenty years, the founder playbook was consistent: start at a big tech company, learn how things work, leave to start your own company. If you had a previous exit, even better. If you could raise money from top-tier VCs, that proved you were serious.

This playbook worked. It created a pipeline of founders who understood business, who had networks, who could recruit teams.

But it also filtered out a lot of good founders. People who didn't fit the mold. People from different backgrounds. People who came to the problem through a different path.

What AI changed

AI changed three things about founder evaluation:

First, the pace of learning accelerated. You can now learn AI, build with AI, and ship AI products in months instead of years. That means a founder without a pedigree can demonstrate technical capability quickly. You can evaluate them based on what they've actually built, not where they previously worked.

Second, the distribution of capability flattened. You don't need to work at Google to understand how to build AI systems. You need to understand the problem you're solving and be willing to grind. Those are learnable. Pedigree matters less.

Third, the bar shifted. We used to evaluate founders on their ability to raise capital and execute inside a known playbook. Now we evaluate them on whether they understand a problem deeply enough to solve it differently than everyone else.

How to evaluate founders now

Technical depth matters more. Not because engineers are better than operators. But because if your founder doesn't understand the technical problem, they'll make bad decisions. With AI, that's even more true.

Problem obsession matters more. If your founder doesn't actually care about the problem, they'll pivot when AI hype shifts. The founders building something real are building it because they're obsessed with the specific problem, not because they think AI is the next thing.

Coachability matters more. The playbook is less stable now. Founders need to be able to learn and adapt. That means they need to be willing to listen, to change their minds, to take feedback seriously.

Network matters less. You still need distribution and connections, but you can build those as you go. Early capital used to flow to well-connected founders. Now it flows to founders solving real problems.

The inverse risk

The flip side of this is that we're now taking risks on founders we wouldn't have backed before. Some of them will fail spectacularly. Some will build things that matter.

That's the trade-off. The old system was safe. Safe founders, safe problems, predictable outcomes. The new system is riskier. Riskier founders, harder problems, unpredictable outcomes.

But the upside is higher. The founders who don't fit the mold are building different things. They're solving problems in different ways. They're creating new categories instead of competing in existing ones.

What founders should know

If you're building something AI-related, here's what we're actually evaluating:

Do you understand your problem deeply? Not theoretically. Have you talked to customers? Do you know their constraints? Can you articulate why your approach is different?

Can you build it? Not eventually. Now. Can you demonstrate that you understand the technical problem well enough to at least ship something?

Are you obsessed? Not just interested. Do you care enough to grind when things get hard?

Those three things matter more now than your resume ever did.