Builds a simulation-driven risk quantification platform that uses probabilistic modeling, Bayesian safety validation, and neural network verification to enable insurers to price and underwrite coverage for physical AI and autonomous systems (e.g., autonomous vehicles, robots, drones) in the absence of historical claims data.

Valgo has publicly stated its mission to build the risk quantification layer for insuring physical AI. As a YC W26 company, their near-term roadmap centers on refining simulation-driven probabilistic risk models for autonomous vehicles and robotics, engaging pilot insurer partners, and expanding coverage to additional physical AI asset classes. Their website and YC profile emphasize route-, task-, and environment-level risk modeling as the core product offering.

Founder publication activity in 2024-2025 on constrained POMDP planning (ConstrainedZero), neural network backward reachability, and importance sampling for rare-event failure estimation strongly signals development of a proprietary simulation engine for tail-risk quantification. GitHub and academic profiles show active work on surrogate modeling and black-box validation tooling. Both Robert Moss and Sydney Katz completed Stanford PhDs in AI safety under Professor Mykel Kochenderfer at the Stanford Intelligent Systems Laboratory (SISL) and Stanford Center for AI Safety. Sydney Katz also teaches Validation of Safety-Critical Systems at Stanford. Conference appearances and postdoc affiliations hint at partnerships with AV and robotics companies for real-world validation data.