When it comes to AI adoption, many insurers have lagged behind their counterparts in other industries. But waiting is not a sustainable AI strategy. Insurtechs are adept at using natural language processing, machine learning (ML), deep learning and other AI technologies. These upstarts are leveraging AI to introduce a new range of innovative products, such as instantly customizable life insurance and on-demand property coverage.

Haven Life, a subsidiary of MassMutual, for example, applies machine learning to third-party data such as prescription and driving records to offer real-time underwriting, allowing customers to buy life insurance online in just minutes without a medical exam.

A handful of established insurance companies are also investing aggressively in AI and applying it throughout the insurance value chain. This includes chatbots that deliver customized product recommendations and manage customer service inquiries; underwriting that occurs in minutes by analyzing a broader array of external data sources; and automated claims processing that analyzes images of damage provided by customers and even via drones.

To remain relevant, insurers will need to move quickly to infuse AI throughout their strategy and operations.

The Legacy Company Advantage

Established insurance companies actually have a head start when it comes to deploying an AI strategy, in the form of customer data. When a leading property and casualty insurer wanted to improve its first notice of loss (FNOL) process, for instance, it based its solution on existing customer phone interactions. We worked with the insurer to transcribe 25,000 phone recordings using voice-to-text technology and analyzed these calls to identify the most commonly occurring activities. We built chatbot functionality to capture these processes automatically. The session also includes recommendations for service rep dialog.

With our solution, the insurer achieved 92% voice-to-text transcription accuracy, up from 67%. All calls are transcribed in real time. After a call is completed, the transcription is summarized and automatically presented to service rep supervisors for review and approval, enabling them to focus on quality and efficiency. With its new post-call approval process, the insurer reduced average call length by 20% and lowered call costs by 22%.

Meanwhile, Zurich Insurance Group has partnered with the Swedish insurtech Greater Than and the Irish division of car rental company Sixt to analyze potential customers’ individual driving data and compare it to a set of reference profiles created from more than a decade’s worth of collected data. Doing so allows the company to customize premiums based on driving behavior.

AI applications can also enhance the customer experience by providing personalized product recommendations, rapid underwriting and quick claims resolution. U.S. insurers Allstate and Farmers use image recognition software or computer vision to settle auto claims without the need for a visit from an adjuster. Home sensors, drones and smart devices will often generate an FNOL before the customer needs to contact the insurer. Rather than address straightforward claims, adjusters are freed to focus on analyzing complex claims and investigating potential instances of fraud. Amelia, IPsoft’s virtual agent, is used at insurers such as MetLife and Credit Suisse to combine ML with NLP to make decisions based on real-time conversations.

Getting Started with AI

Although each company’s situation is unique, the following guidelines can help insurers develop an AI game plan:

  • Cast a wide net. Insurers should assess each aspect of their organization to identify how best to deploy AI. Rather than exclusively focusing on technology, the effort should begin with the company’s business needs and opportunities and where AI can generate business value.
  • Look for opportunities to leverage data. For each business process, insurers need to identify the data required to take advantage of AI, including data from external sources such as wearables and from connected devices on the Internet of Things (IoT). Most insurers will need to develop stronger data governance to ensure they have access to accurate, timely data.
  • Acquire AI expertise. Additional AI skills will be required through a combination of hiring additional talent, partnering with third-party experts and partnering with, or even acquiring, insurtech start-ups.
  • Encourage experimentation – and discipline. Insurers must develop a tolerance for experimentation but combine that with rigorous measurement of ROI so that failures can be terminated quickly and successful pilots can be moved into full implementation.
  • Prepare business processes for digitization. Layering an AI solution on top of a poorly designed business process will squander its potential. Insurers should first optimize the process by making system changes, as well as standardizing and consolidating fragmented systems.
  • Design responsible AI. Applications that make inappropriate or biased decisions can inflict significant reputational damage and loss of shareholder value. Just as they have ethics officers, insurers will need to establish AI ethics policies and procedures to ensure their applications continue to operate appropriately as they learn and adapt over time.

To ensure AI efforts achieve their desired business outcomes, we recommend the “5 E” process (educate, embrace, evaluate, establish and explore).

AI is moving to the mainstream in many industries, and the technologies and tools are evolving quickly. It’s time for insurers to accelerate their efforts and realize where AI fits into their digital strategy.

To learn more about advancing your AI strategy, visit us at Guidewire Connections 2019, Nov. 3-6, 2019, at the Gaylord National Resort in National Harbor, MD.  

Steve Naish

Steve Naish

Steve Naish is Guidewire Practice Leader at Cognizant. He has 25 years of experience using technology to help leading insurers drive profitable... Read more