How Bestow Is Revolutionizing Post-Issue Audits With AI-Powered UW Assist

Bestow's innovation team, made up of individuals from the engineering, product, and insurance teams, is spearheading the GenAI transformation at Bestow with the creation of Underwriter Assist (UW Assist). UW Assist is a Generative AI-powered solution to transform how we carry out Post-Issue Audits (PIAs). This will enable carriers to operate more efficiently while upholding high standards of accuracy and compliance. 

The Challenge of Post-Issue Audits

Carriers have long been obligated to audit a portion of their instant underwritten business. Traditionally, these audits have been manual and time-consuming, often relying on skilled labor to review and assess the quality of underwriting decisions. As the industry faces a growing demand for efficiency, the existing processes present several challenges:

• Resource-Intensive: Manual audits require significant manpower and expertise, consuming valuable resources.
• Cost of Errors: While false negatives may not severely impact outcomes, they can lead to oversight of critical risks. Conversely, false positives necessitate time-consuming reviews before any actions, such as rescissions, are taken.
• Scalability: It is crucial to efficiently scale audits as business volume grows, yet traditional methods often fall short.

Introducing UW Assist: A Solution Built on AI Innovation

At Bestow, we recognized these challenges and embarked on a mission to develop a solution that leverages the power of Generative Artificial Intelligence (GenAI). In August 2023, we initiated a GenAI workshop with Google Cloud Platform (GCP), leading to the conceptualization of UW Assist with our underwriting team.

Key Features of UW Assist

1. Automation of Audits: UW Assist automates the initial stages of the auditing process, significantly reducing the time required to conduct thorough reviews. This enables carriers to focus their human resources on more complex cases.
2. Data-Driven Insights: UW Assist utilizes a sophisticated Retrieval Augmented Generation (RAG) architecture that collects and tokenizes documents before processing them through a large language model (LLM). This offers actionable insights based on real-time data analysis.
3. Scalable Architecture: Designed to handle both bulk uploads and individual document assessments, UW Assist allows for scalability while maintaining consistent performance.
4. Ongoing Learning: UW Assist is set to improve its accuracy and efficiency as the technology evolves. Our latest version of the LLM shows promising signs of enhanced performance, indicating that the tool will only get better over time.

Recent Progress and Milestones

Since the initial development phases, we have made significant strides in bringing UW Assist to life. In the first quarter of 2024, we hardened the prototype and refactored it into an API endpoint to allow batch processing and seamless integration with existing systems. By midyear, we successfully processed third-party PIAs using UW Assist and compared the results with traditional methods. The initial results indicate that UW Assist can lighten underwriting’s workload by capturing straightforward cases with high-impact declines. 

Looking Ahead: The Future of UW Assist

Our journey with UW Assist is just beginning. As we gather more data and refine the tool's capabilities, we anticipate even higher accuracy rates and broader applications across the underwriting process. The feedback from initial partnerships will be invaluable in shaping future iterations of UW Assist, ensuring that it meets the industry's evolving needs.

At Bestow, we believe that the future of life insurance lies in harnessing the power of technology to drive efficiency and accuracy. UW Assist is a testament to our commitment to innovation in underwriting, offering a glimpse into how GenAI can transform traditional processes.

Stay tuned for more updates as we continue to develop and refine UW Assist. For further information about UW Assist and its capabilities, please feel free to contact our Enterprise team.