AI is everywhere in 2025—but venture capital has been a strange exception. That’s the paradox Humberto Treviño, AI and Data Analytics Director at Katapult VC, set out to solve. In his conversation with Tim Freestone on the Objective Hiring podcast, he explained how his team built an AI scout to supercharge startup sourcing, tackle false negatives, and create a more data-driven way of investing.
“Very few venture funds have integrated AI and these new technologies into the way they operate. Venture continues to operate with Excel, PowerPoint and PDF—and that’s it.”
At Katapult VC, Humberto led the development of tools that help identify high-potential startups using AI. But it’s not just about automating sourcing—it’s about doing it better.
“We’ve essentially built an AI scout. We’re able to augment the amount of information that’s gathered from the different deals we’re evaluating.”
The traditional VC approach often relies on short-term interns or junior analysts manually reviewing companies. The result? Limited time, limited energy, and limited experience—leading to missed opportunities. Katapult’s AI scout addresses these issues head-on, processing far more deals in less time while learning from senior team members to improve its recommendations.
“Our algorithms learn from experience. You don’t lose that capability when your intern leaves. You're honing a more and more capable AI scout.”
One of the biggest problems in venture? Missing out on unicorns too early in the funnel.
“We found that they were losing the biggest deals in the earlier stages, when more junior profiles were evaluating. That’s where the largest opportunity is.”
False negatives are a nightmare in VC—turning down the next billion-dollar company because someone overlooked the signal. Humberto’s analysis revealed that these mistakes typically happen early, when undertrained team members screen out promising startups. AI helps mitigate this by providing consistency and scale.