Investing in AI for Hard Tech, with Eric Vishria of Benchmark and Sergiy Nesterenko of Quilter

Circuit board design represents a "done-for-you" AI problem rather than a copilot-assisted task, as manual layout processes are inherently inefficient and prone to human error. Quilter utilizes reinforcement learning to optimize circuit board architecture, leveraging physics-based simulations to achieve performance levels that surpass traditional human-led design. This shift highlights a broader investment philosophy where durable AI businesses prioritize specific, vertical applications over general-purpose foundation models, which often suffer from rapid asset depreciation. Successful ventures require founders who have navigated the "idea maze," demonstrating deep, iterative engagement with a problem rather than merely riding current hype cycles. Bridging the gap between conceptual research and scalable, production-ready engineering remains the primary challenge for startups aiming to transform industries like hardware design, where rigorous physics constraints dictate success.