A.I., Start-ups, and the Shifting Edge of Venture
AI has lowered the bar to become a startup founder.
At the same time, it has raised the bar for competition.
That paradox is where we now live.
A single entrepreneur today can write code, generate market research, build financial models, draft legal documents, and produce world-class marketing content faster than small teams could just a few years ago. AI compresses time. It amplifies capability. It reduces friction. But if it’s easier for you, it’s easier for everyone else too…which means the competitive baseline just moved up.
Now AI is entering venture capital itself. Models are emerging that claim to predict startup outcomes with remarkable accuracy. That sounds revolutionary — until you ask what is actually being predicted. Predicting which companies will raise capital is interesting, but it isn’t the same as predicting which companies will build enduring enterprise value. Capital often follows signal, network, and narrative. It is recursive. Money attracts money. That’s not prophecy. That’s pattern reinforcement.
Angel and venture investors have always lived in the world of forecasting uncertain futures. Historical data gives comfort, not certainty. Past performance is not a lock on future performance — especially at the earliest stages. The real challenge has always been the data.
Startup data are messy, disjointed, incomplete, and biased toward survivors. To call it imperfect would be generous. We don’t have standardized reporting across early-stage companies — nor should we. Mandatory reporting would smother fragile startups in administrative burden. This is where AI becomes genuinely important.
AI can ingest imperfect data. It can identify patterns across noise. It can surface correlations humans might miss. That makes it a natural tool for venture.
But evaluating eight buckets of risk isn’t enough. Reviewing cap tables and TAM slides isn’t enough. Even revenue growth doesn’t tell the full story. The hardest variable has always been the founder.
Psychometric models are now showing surprising predictive power around entrepreneurial outcomes. That should make us pause. Traits like resilience, cognitive flexibility, and risk tolerance can be measured. And increasingly, they can be modeled. Have we crossed a milestone? Yes.
For decades, early-stage investing relied heavily on pattern recognition — often a polite term for refined intuition. We gravitated toward investors who seemed to have unusual luck. We called it instinct. AI is beginning to formalize parts of that instinct. But here’s the distinction that still matters:
Predicting later-stage success is more math.
Predicting early-stage success is still more art.
Later-stage companies produce data stability — unit economics, margin trends, customer retention. Early-stage companies are experiments in motion. They pivot. They evolve. They survive near-death experiences.
AI can assist. It can augment judgment. It can reduce bias. It can flag blind spots. But even the builders of these models say they will augment, not replace, investors. Partly for liability reasons. But partly because if predictive models alone generated venture-level alpha, those companies would run funds instead of selling software.
We are at an inflection point. AI will influence:
Sourcing
Diligence
Portfolio monitoring.
Capital allocation.
The capital markets are not immune.
My thesis is simple. AI will compress the middle. Average founders will become more capable. Average investors will become more informed. That means the real edge shifts again — to judgment, timing, and courage.
The mythology of pure gut-feel investing is fading. The art remains. But the canvas just became mathematical.
