U.S. property and casualty insurers had a strong 2025. AM Best, the ratings agency that tracks the industry’s finances, reported that carriers paid out about $92 in claims and expenses for every $100 they collected in premiums, their best margin in ten years. Underwriting profit reached $60.9 billion, nearly triple the year before, helped by a milder year for catastrophes.
Soteris, a machine learning-powered company serving property and casualty insurers, came out of stealth today, with a new product that promises to find the policies inside a profitable book that are losing money. The Richmond, Virginia, company has raised more than $8 million in seed funding led by Spider Capital, with Intact Private Capital, Amplify Partners, Foundation Capital, the Webb Investment Network and Overlook Ventures participating. It went through Y Combinator in 2019 and says it has since scored more than 100 million insurance applications with its first product.
The problem is particular to insurance. A carrier sets a price before it knows what the policy will cost, because the cost is whatever claims get filed later. A single policy reveals little on its own, since it either produces a claim or it doesn’t. So, insurers group similar policies together and judge them by how the group performs on average. The group has to be big enough for that average to be reliable, and individual policies disappear into it. A group that looks profitable can still contain policies that lose money every year.
Soteris says its software divides a carrier’s policy history into millions of overlapping groups at once and compares the results, which lets it estimate what any single policy will earn or lose. The company calls that a “segment of one.” The system goes live in under 90 days and scores a policy at any point in the policy lifecycle in about a quarter of a second.
“Every insurer knows they’re writing policies that will lose them money. They just can’t find those policies with the resources currently at their disposal,” said founder and CEO Sunit Shah, a Ph.D. in economics who built a $750 million insurance business inside hedge fund Pine River Capital.
Carriers using the company’s first product cut the share of premiums paid out in claims by 5 to 15 percentage points within a year, Soteris says, with trials of the new product raising the profit on an existing book by 70% to 125%. McKinsey has pointed to the same kind of opportunity: in an April study, it said insurers covering midsize businesses can lift underwriting profit 30% to 50% by working on the books they already have, including dropping chronically unprofitable policies.
Using the product requires no change to an insurer’s rates or filings, the company says, which keeps it clear of state approval. Regulators in 25 states and Washington, D.C., including Virginia, have adopted guidance from the National Association of Insurance Commissioners on how insurers should oversee AI used in decisions like these. The guidance points to the AI risk framework published by NIST, the federal standards agency, as one model insurers can use, and Soteris says it is working to align with it.
The company is coming out of stealth as investors put more money into AI for insurance. AI-focused companies received 99.1% of the $2.44 billion invested in insurtech worldwide in the second quarter of 2026, the highest quarterly total since 2022, according to broker Gallagher Re.
Featured image: Getty Images via Unsplash+
Disclosure: This article mentions a client of an Espacio portfolio company.
