Soteris targets a structural blind spot in P&C underwriting: insurers traditionally average policies into segments, so negative-profit or high-loss policies can remain hidden even inside healthy books, while Soteris says its models score policies individually. The bet is that policy-level predictions can turn existing books into more EBITDA without changing rates or headcount; the company says its system has scored more than 100 million submissions covering over $180 billion in premiums, customers have seen 5-to-15-point loss-ratio improvements, and proofs of concept showed 70% to 125% EBITDA increases.