Background 6

Insurtech Insights: When AI becomes the best recipe for underwriting

Jul 04, 2018

Delivering AI underwriting can enable insurers to scale their business into new and expanding markets.

After Cytora CEO Richard Hartley spoke at the Insurtech Insights conference in London, we wrap up some of panel’s thoughts on underwriting with AI.

What are the opportunities for AI in the underwriting value chain?

Hartley emphasised the opportunity to use AI underwriting in risk selection. In the SME insurance market where premium is typically low, insurers can use AI to automate decision making, similar to the current situation in personal loan underwriting, where it is very easy for an applicant to get a decision.

For more-complex risks, Ashish Umre, Head of Artificial Intelligence at XL Catlin, emphasised looking at how decision making can be improved by data to help underwriters better select and manage risk.

When AI is used for risk selection and pricing, insurers can automate the entire underwriting flow and optimise their processes for customer experience and engagement. One example here is replacing large cumbersome question sets with a small number of targeted questions.

Is there always a need for people in the workflow? 

Michael Natusch, Global Head of AI at Prudential, offered a perspective on the role of AI underwriting for a multinational life insurer (Prudential operates worldwide through a network of 650,000 agents). It provides health insurance to 65m people in Asia. While impressive, this number represents only ~2% of the Asian population. Natusch spoke about how data-driven approaches could deliver growth without adding additional agents to the business.

For the Asian health insurance customer, the Prudential buying journey currently comprises a standard set of 27 questions (potentially blowing up to 156 questions), and a 90-minute appointment with an agent to answer them. Many of the answers are rarely used for underwriting, and the company is exploring ways of bringing down the number of questions asked.

What lessons can legacy insurers learn as they build the insurance company of the future?

Hartley highlighted the benefits to legacy insurers of experimenting outside of their legacy processes. He offered XL Catlin as an example of an insurer that does this really well by experimenting end-to-end outside of legacy and then introducing a complete solution to the organisation (Disclaimer: Cytora work with XL Catlin’s innovation team, Accelerate, to embed the Cytora Risk Engine).

Satadru Sengupta, General Manager at DataRobot, suggested that the most productive partnerships between legacy and insurtech are the ones that focus on bite-size delivery. This means adopting a start-up-like mentality and making a series of adaptable 3-month plans rather than a rigid 2-year plan.

John Pyall, Head of MGA Cockpit at Munich Re mentioned that many insurtechs struggle to link up with their market and that ultimately, insurtech AI has to make an impact on the insurance value chain to be relevant.

Hartley echoed this view, saying that while many emerging insurtechs are developing impressive new technologies, they aren’t doing so well to quantifying their impact in terms that matter to insurance leaders: combined ratios, premium growth and underwriting performance.