Gradient A.I., spun out of Milliman, looks to midsize insurers for growth

Stan Smith, founder and CEO of gradient A.I., has acquired the company from its former parent, the insurance advisory firm Milliman.

Gradient provides predictive analytics and data warehousing services to insurance companies, powering its underwriting and claims-management models with AI. There's a staff of about 20 who are coming with Smith in the new era of the company. Terms of the deal were not disclosed

Smith joined Milliman in 2013 to start gradient after launching a series of startups including a predictive modeling service for supply-chain management. Milliman decided to spin the unit out because it felt it didn't dovetail with its consulting services.

di-server-stock-08161.jpg
Fiber optic cables, center, and copper Ethernet cables feed into switches inside a communications room at an office in London, U.K., on Monday, May 21, 2018. The Department of Culture, Media and Sport will work with the Home Office to publish a white paper later this year setting out legislation, according to a statement, which will also seek to force tech giants to reveal how they target abusive and illegal online material posted by users. Photographer: Jason Alden/Bloomberg

"There wasn't alignment, so I purchased what I didn't already own from the business," Smith explains.

Gradient's current clients are risk pools, self-insured companies and small mutuals. These include the Connecticut Interlocal Risk Management Agency (CIRMA), The Builders Group and A.I.M. Mutual.

"It's a tough space to start but if you prove yourself you can move up," Smith says. "We're fortunate that a lot of the clients we've worked with are happy to share knowledge and experience."

The database that gradient works off of includes data from its clients, Milliman and other third-party sources. Smith says the company anonymizes information as much as it can, and segregates as needed depending on usage agreements. Now, the goal is to operationalize it for bigger clients.

“We’re seeing a pull from the midmarket -- bigger than we initially targeted. These are carriers up to a billion dollars in premium,” Smith says. “Even companies with hundreds of millions in premium have a small number of anomalous outcomes. Having a big data set, because we started in Milliman, will help.”

For reprint and licensing requests for this article, click here.
Artificial intelligence Machine learning Data warehouses Data Scientist Predictive modeling Predictive analytics
MORE FROM DIGITAL INSURANCE