Of startups that raise a seed round, only about 15 to 20 percent go on to close a Series A. Four out of five fall short — and most of the time, the idea isn't why. In CB Insights' breakdown of startup failure causes, "no market need" ranks first at 42 percent, and running out of cash ranks second at 29 percent. A flawed idea doesn't even show up as its own category.
So what are investors actually looking for in a pitch? Vanessa Larco, a former NEA partner, has been unusually direct about it. Her published bar is simple: does AI let you deliver the same value dramatically faster, far cheaper, or far easier? If none of those show up, she doesn't keep evaluating. The word "dramatically" is doing the real work here — a solution that's just 10 to 20 percent better than what already exists doesn't clear the bar.
What This Test Is Really Checking For
Simply bolting on AI proves nothing on this test. What it actually reveals is how deeply the founder has broken down their own problem.
Take an example. Say a lawyer typically spends four to six hours reviewing a single contract. A legal-tech founder who explains that "AI speeds up the review" doesn't pass this test. What passes is a number: review time cut to 20 minutes, or cost down to a tenth of what it was. And to produce a number like that, the founder has to already know which step in contract review is actually eating the time and money. Only someone who has broken the problem down knows where to point the AI.
When a founder claims AI made "a dramatic difference," the gap between a team that pinpointed the exact bottleneck in value delivery and aimed AI directly at it, and a team that simply bolted on AI features without being able to say which bottleneck they actually solved, becomes obvious within five minutes of the pitch.
Why Investors Look at the Founder Before the Idea
At the seed stage, the idea itself is hard to evaluate on its own merits. By the time any given idea goes public, there are usually already three to five other teams moving in roughly the same direction. Good ideas tend to occur to multiple people at once. The gap opens up in execution.
One of the first things Larco checks in a founder is how long, and how closely, they've watched the market in question. Someone who has observed it for a long time sees structure, not trends. They can explain why customers aren't fully adopting the solutions already on the market — more tellingly than they can recite competitor headcounts. Ask an unexpected question, and the gap between someone who memorized numbers and someone who's spent real time in the field shows immediately.
She also looks at how the first customer was won. A first customer landed through direct persuasion, with no marketing budget, reflects a different intensity of execution than one who arrived through a friend's network. A founder who can describe exactly who that first customer was and how they were won over is signaling that they're doing sales and product refinement at the same time.
How a founder reacts to signals that they're wrong matters too. A founder who gets defensive and one who absorbs the feedback and adjusts course end up, after the same stretch of time, in very different places with their product. You can't observe that reaction directly in a pitch meeting, but asking what pivots they've made since founding, and why, brings it to the surface.
Few startups grow into a Series A on their original seed-stage idea. Instagram started as Burbn, a location-based check-in app. Airbnb started as an air-mattress rental service. What determined the eventual product wasn't the idea — it was how the founders responded to signals from the market.
The Same Bar Applies to Solo Founders in Korea
The VC funding world and the context of a solo founder in Korea running on a shoestring aren't the same. VCs build portfolios around 10x and 100x outcomes. Someone bootstrapping has to prove sustainability before they prove scale. The thresholds differ.
But the underlying question in this test applies regardless of scale: somewhere in the value you currently deliver to customers, is there a point where AI can make it dramatically faster, far cheaper, or far easier?
For a solo content creator, "AI helps me write faster" doesn't pass this test. The real question is where in the content-production process the most time actually goes — research, drafting, editing, or analyzing audience response after publishing all point to different places to aim AI. Six months out, there's a real gap in output volume and cost structure between someone who knew exactly where to apply AI and someone who sprinkled it across every step.
How precisely a founder can define their own problem reveals their capacity for execution far more consistently than any list of AI features does. Larco's test is one way of checking whether that precision is actually there.
The question was never which AI tool you use — it's where you use it. And you only know where if you've actually broken the work down.



