How autonomous AI can expedite commercial-insurance underwriting
Autonomous AI assistants are coming. Here are four ways insurance organizations can begin using this tech right now.
Autonomous AI assistants are the next evolution of generative AI as this technology can take on distinct personas, self-learn and self-correct.
While AI assistants can benefit the entire insurance lifecycle, they are having a significant impact on underwriting.
There are often capacity challenges for underwriting teams, especially in commercial lines. Deep and intuitive underwriting proficiency requires a specialized skillset, and training professionals in risk evaluation takes time. However, autonomous AI assistants can help to reduce the talent gap as they can be trained as reliable resources on many task-oriented underwriting processes, freeing underwriters up to focus on more complex risk analysis, workflows and priorities.
Implementing AI assistants into underwriting workflows may raise some questions at first. Will these assistants disrupt existing workflows? How will underwriters manage and interact with them? Can their output be trusted?
Insurers should think of AI assistants as well-qualified interns. As a starting point, underwriters can train them to take on simple roles and tasks and then progress to more elaborate assignments. AI assistants can learn to perform tasks quickly and accurately while self-learning as they execute their assignments.
What follows are four ways insurance organizations can begin using role-based AI assistants in their underwriting processes right now to help commercial underwriting teams and boost their capacity:
No. 1: Determine risk eligibility
When a business submits an application for coverage, the underwriter must evaluate whether it meets the insurer’s risk-selection criteria. Autonomous AI assistants can ingest the insurer’s underwriting guidelines and then use those guidelines to review incoming submissions efficiently. The AI assistant can decide on in- or out-of-appetite risks, enabling the underwriter to review only the submissions that fall into gray areas.
The underwriter can also train the AI assistant to incorporate underwriting nuances. For example, often, food establishments that sell liquor and are open past 10pm typically generate more revenue from alcohol than food and would be classified as a bar. Based on its guidelines, an insurer might not wish to write bars and favors restaurants where alcohol is 30% or less in total sales. If an underwriter reviews an application where a food establishment reports alcohol sales as 30% but closes at midnight, this will raise a flag. The underwriter can train the AI assistant to look for similar discrepancies and flag the results for further review.
No. 2: Comprehensive portfolio management
Underwriters will often audit their books of business by selecting sample lists of accounts by class and ensure those accounts tick all the necessary boxes for risk eligibility. Underwriters no longer have to rely on sample audits because their AI assistants can be trained to review entire books of business reliably. Once the assistant is familiarized with the various eligibility parameters, they can quickly go through all policies to ensure in-appetite consistency.
No. 3: Quickly scale new risks
AI assistants can shorten the risk-assessment lifecycle. Insurance organizations typically have a standard set of risk attributes they look for in a business class. For a restaurant, the attributes might be related to grease trap specifications, and for a bar, they could be related to hours of operation. Changing or creating new risk characteristics to capture and understand can be challenging due to complex IT infrastructures and programming requirements. However, it is easy to teach AI assistants such changes.
For example, an insurer might want to change its risk parameters for limited-service restaurants like fast food establishments. It had previously insured operations that were open until 2am but now wants to narrow it and only insure restaurants open no later than midnight. The underwriter can teach this new parameter to the AI assistant in straightforward terms. The assistant learns and can apply it to the book of business, enabling the underwriter to modify the risk attribute without coding/programming changes.
No. 4: Streamline premium audit
Premium audit is a heavily regulated workflow, and specific guidelines must be followed. The underwriter’s premium audit responsibilities can take considerable time, leading to delays in moving the audit downstream. AI assistants can take over some of the more administrative tasks of premium audit workflows, such as workload distribution and document review, to help expedite the process.
Insurers often face challenges allocating auditable policies to appropriate underwriting team resources for self-audit, phone, or onsite audits. An AI assistant can ingest historical audit data, claims data, and state regulatory guidelines and be trained to analyze current policies with similar characteristics efficiently. The assistant can then use this information to direct audits toward the most suitable path for processing.
Premium audits also require auditors to examine state guidelines and documents provided by policyholders. If they are missing information, the auditor has to reach out to get it. AI assistants can learn the state guidelines, review the submitted information, and rapidly determine if there is any incomplete or missing information, thus reducing the back-and-forth with the insurance agent or insured.
Autonomous AI assistants can be supportive companions to underwriting teams, helping to maximize workflows, improve risk-evaluation accuracy, and scale portfolio management and capacity. While implementing new technology can seem daunting, insurance organizations can start small, evaluate their workflows, and pinpoint particularly labor-intensive and repetitive processes that could benefit from deploying dependable, always-available AI assistants.
Sathish Kumar Manimuthu serves as the CTO at NeuralMetrics, an insurtech offering AI-powered risk-assessment intelligence for P&C carriers, brokers, and agents. With experience in launching innovative technology and products for startups and Fortune 500 companies, Sathish is responsible for the company’s suite of data-delivery engines and AI models.
See also: