Insurance provider Hiscox has announced its creation of a new AI model capable of automating the lead underwriting process for specialist risks. Built in collaboration with Google Cloud, the platform was recently tested in a pilot where it acted as the lead underwriter for a property sabotage and terrorism policy – the first time, Hiscox said, that this type of process had been automated. The insurance provider added that it would likely begin using the tool in late 2024.
“We are delighted to be working with Google Cloud to explore the applications of AI technology to augment our underwriting teams’ knowledge and expertise,” said Kate Markham, chief executive of Hiscox’s market division. “This early stage work suggests that there are areas of the business where automation can support better, faster results for our brokers and clients, and we look forward to developing the programme further.”
The platform itself uses Google Cloud’s Vertex AI and BigQuery services to extract and analyse data from emails submitted by insurance brokers, including draft contracts – a process, Hiscox said, that can take up to three days using a human underwriter. The provider then uses an in-house AI model system to decide whether the risk associated with the product is appropriate for the insurance provider to assume, before writing a draft email summarising that decision that is reviewed by a human underwriter before it is dispatched to the broker.
“We don’t want to remove the human in the chain at the moment,” Markham told the Financial Times. “We want to keep the underwriter so they can confirm that the models are behaving.”
Markham also implied that the tool is unlikely to lead to any reduction in headcount at Hiscox. Instead, it would allow its team of human underwriters to “do more from a sales perspective” and not prevent the insurance provider from continuing to hire new underwriters or grow its headcount generally.
AI insurance advances
This is not Hiscox’s first foray into AI. In 2018, it partnered with natural language processing firm Eigen Technologies to create assistive programs for staff. Again, it emphasised the goal of its drive for more AI in its business as creating more efficiencies for its existing headcount, and not for devising new ways of reducing it. “Our goal is to be digital wherever possible, but with a human touch where it matters to the customer,” said Hiscox’s then-group head of strategy and corporate development, Lidia Bozhevolnaya.
Neither have Hiscox’s competitors proved slow in adopting AI. Zurich UK has also announced a new “customer copilot” powered by generative AI. The platform, explained its head of UK commercial insurance Frank Streidl, would ingest policy, submission and risk engineering data to produce solutions to “challenges some customers face in consolidating vast amounts of data”. Allianz, too, is using the technology to improve the claims journey for customers, claiming in a recent blog post that was actively using “AI-based tools” to trawl through available data for fraudulent activity.