Insurance companies face significant challenges in managing unstructured data, a predicament that has plagued the industry for years. This data, often ensnared in various formats—from paper documents and emails to attachments—comprises approximately 80 percent of all the data utilized within the sector. The reliance on manual processing of this data leads to inefficiencies, incomplete analysis, and enormous financial losses, as highlighted by recent reports from AM Best and Accenture. These manual processes are painstakingly slow and prone to errors, which can have severe ramifications on underwriting and claims decisions.
The Scale of the Problem
A survey conducted by IDC uncovered that 40 percent of mid-market and enterprise businesses still rely on “mostly manual” document processing methods. In the insurance industry, the effects of these manual processes are even more pronounced due to the enormous volumes of data processed daily. The inability to effectively convert unstructured data into structured formats significantly impacts underwriting and claims decisions, resulting in extensive financial losses. Accenture’s estimates reveal that inefficiencies attributed to unstructured data in underwriting could lead to potential losses ranging from $85 billion to $160 billion by 2027.
These inefficiencies are further exacerbated by the reliance on human experts to perform administrative tasks such as data entry and document reworking. The time and resources spent on these non-essential tasks result in wasted value and undermine the overall productivity of insurance operations. AM Best reported a staggering $21 billion loss in Property and Casualty underwriting in 2023 directly attributable to the failure to capitalize on unstructured data. Insurers are thus coerced to seek more innovative and effective solutions to combat these losses, with the focus shifting towards maximizing efficiency and effectiveness in their operations.
Traditional Automation Tools and Their Limitations
Over the past decade, the insurance industry has employed legacy automation technologies, such as Robotic Process Automation (RPA), Intelligent Document Processing (IDP), and Intelligent Process Automation (IPA), to mitigate unstructured data problems. However, these traditional tools have consistently failed to deliver the desired outcomes, largely due to their inherent limitations. RPA tools, for instance, have been criticized for being brittle, slow to maintain, and overly dependent on structured data to function effectively.
These tools were not designed to handle the complexities of unstructured data, resulting in continued inefficiencies and errors. Consequently, many insurers have observed that the expected improvements in processing speed and accuracy have not materialized. This persistent struggle has necessitated the exploration of more advanced solutions capable of understanding and managing unstructured data in a more effective and sophisticated manner.
Enter Generative AI
Chaz Perera, co-founder and CEO of Roots Automation, asserts that Generative AI offers a promising solution to the unstructured data dilemma faced by the insurance industry. Generative AI models excel in natural language tasks, possessing the ability to understand context, extract relevant information, and generate human-like responses. These capabilities make them ideally suited for automating insurance workflows and reducing the incidence of errors. Accenture’s analysis suggests that Generative AI could automate up to 62 percent of insurance underwriting and claims processes, boosting operational efficiency by effectively managing unstructured data.
This automation would allow human experts to redirect their focus to higher-value tasks, enhancing productivity and overall operational effectiveness. Moreover, Swiss Re estimates that leveraging AI to extract insights from unstructured data could result in a 12-25 percent improvement in claim loss adjustment. The potential for such efficiencies and productivity gains makes Generative AI an attractive proposition for insurers seeking to navigate the complexities of unstructured data and improve their operational outcomes.
Industry-Specific Solutions
Despite the promising capabilities of Generative AI, public horizontal solutions provided by tech giants like Google, OpenAI, and Meta are not tailored to the unique needs of the insurance industry. These solutions lack the training required to handle insurance data accurately and to interface effectively with insurers’ systems in a contextual manner. Consequently, they are impractical for complex and regulated environments such as insurance, where precision and compliance are paramount.
Building a robust in-house Generative AI solution demands significant resources and expertise, a feat beyond the reach of most insurers, except a select few large multinationals. A poll conducted by Oliver Wyman and Celent revealed that while over 20 percent of US insurers are attempting to develop internal vertical Generative AI solutions, less than half have advanced beyond the proof-of-concept stage. The reliance on specialized internal knowledge, sophisticated infrastructure, and extensive data resources to develop and manage models with insurance domain expertise presents considerable challenges for many firms.
Advantages of Tailored Solutions
Industry-specific Generative AI solutions offer substantial advantages over their generic counterparts. These tailored solutions can more effectively address the distinctive needs of underwriting and claims management processes. For instance, an insurance-focused Generative AI system can automate document indexing, analysis, and clearance during the underwriting process, extracting up to 90 percent more information from unstructured data. This level of automation can lead to a 10 percent reduction in underwriting costs, resulting in faster and more accurate quotes and improved customer satisfaction.
On the claims side, Generative AI-powered solutions can drastically reduce document handling times, transforming processes that once took days to mere minutes, with over 95 percent accuracy. By decreasing manual claim setup by 90 percent, these solutions help insurers significantly reduce claims overpayments and increase their operational capacity to manage surges in demand, such as those following natural disasters. These improvements underscore the potential for Generative AI to revolutionize claims management, leading to substantial operational benefits and enhanced profitability.
Impact on Claims Management
Insurance companies grapple with significant challenges when managing unstructured data, a problem that has plagued the industry for years. This type of data, which includes paper documents, emails, and attachments, makes up about 80 percent of all the data used within the sector. The industry’s heavy reliance on manual processing to handle this unstructured data leads to considerable inefficiencies, incomplete analysis, and substantial financial losses. Recent reports from AM Best and Accenture underscore these points. These manual methods are not only time-consuming but also highly error-prone, which can severely impact critical processes like underwriting and claims decisions. The slow and error-laden nature of manual data processing can lead to misguided risk assessments and poorly informed decision-making, further exacerbating financial losses and operational inefficiencies. As technology in other sectors advances rapidly, the insurance industry must find innovative solutions to handle and process unstructured data more effectively to stay competitive and mitigate these issues.