Document processing for insurance is inundated with multiple operational challenges. Insurance operations teams deal with various types of documents: 100+ types of ACORD forms, custom forms, SOVs, loss run reports, quotation slips, claims images, etc. These documents could again be in a whole range of formats: PDFs, Excel, Word, and image formats (jpg/tiff/png, etc.). These documents range from structured to semi-structured and unstructured. In addition, the documents corresponding to the same application or claim may arrive at different points in time, often days apart.
THE ISSUES WITH TRADITIONAL PROCESSING
Dedicated teams, like underwriting support teams, handle these documents manually, combing inboxes or shared drives, putting relevant documents together, checking for completeness, reconciling data between the forms and with the insurance systems, and manually typing the details into insurance systems like Guidewire, Majesco, Insurity, or the like so that underwriters can look at a submission for risk assessment. This tedious process is highly time-consuming, error-prone, and affects team morale.
Agility is crucial for the insurance industry since 56% of the time, it is the first responder that wins the business. Underwriters and assistants also spend up to 15% of their time validating and preparing submission data, whereas their core focus should be evaluating risk and building relationships with brokers, as per the SMA report.
Fraud is another issue facing the insurance industry. Because these documents are manually processed, it becomes challenging to identify fraud. It is estimated that insurance fraud costs up to $40B per year, per the FBI. Insurance fraud thus costs, on average, between $400 and $700 per family per year in the form of increased premiums.
Customer experience becomes another casualty in traditional processes when customers are put on hold or have to repeat their information for the umpteenth time when their call is transferred internally while trying to get a quote or have their claim filed or processed.
LOOKING TOWARDS AI FOR HELP
Rapid technology advances, evolving customer experience, complex software maintenance, monitoring protocols, and a changing competitive environment are a few change agents compelling insurers to transform their existing technology stacks. As per a McKinsey report, by 2030, the whole underwriting process will be reduced to seconds as the majority of underwriting will be automated.
Artificial intelligence (AI) and the latest advancements in generative AI have a considerable role to play in enabling the upcoming paradigm shift. As per a study by Genpact, 87% of surveyed insurers already see their companies invest $5 million or more in AI technology each year, and 74% of insurance executives plan to increase their investment in AI.
HOW DOES AI HELP?
- Preparation And Checking For Completeness: AI can be used to combine these incoming forms in different formats into a submission or claims package, check for completeness, and handle quick declines based on OFAC, license, clearance, and dedupe checks.
- Extracting Data From Various Documents: AI can help extract data trapped in different forms and formats—whether unstructured or semi-structured—and transform them into structured data with very high accuracy of up to 99.5% or above.
- Faster Quote Generation: Once the data is extracted, it can be validated against business rules and submitted automatically into insurance systems. The underwriters no longer have to spend time on administrative tasks and can focus on risk assessment, classification, and pricing instead. This can speed up the quote generation process by up to 80%.
- Expedited Claims Processing: Similarly, claims requests that also involve images of the damages along with forms have to be prepared, and pertinent information has to be extracted and filed for claim adjudication that can be expedited by orders of magnitude.
- Identification Of Fraud: While AI is extracting and validating the data, it can automatically and tirelessly 24/7 check the data against existing systems data to detect anomalies in the submission or claim. Suspicious applications or claims can be flagged based on metadata extracted from the claims images or past history from loss run reports.
- Customer Experience: AI-powered automation can enable customized, empathetic, agile resolution of tasks at hand. This is where much progress has been made with conversation bots, but generative AI will be a game changer with meaningful human-like conversations. Coupling this with routing and task resolution automation can help customers get a much better experience that they truly deserve.
THE PATH FORWARD
AI is already transforming insurance document processing. While selecting a partner to expedite your transformation journey, don’t forget to take the following into consideration:
- AI-Powered, Not “AI-Washed”: Given that AI is the mot du jour, claims by vendors that their product also has AI when actually it does not—a.k.a, “AI-washing”—is rampant. Ask your vendor what technologies are being used to power their product. Another worthwhile question to ask is, given the rapid development in the space, how much ongoing investment are they making in R&D to keep their tech stack current?
- Cold Start Problem: While analyzing IDP vendors, an important consideration is how quickly you will be able to get your ROI. An important criteria to consider here is how out of the box the solution is. An out-of-the-box AI-powered solution provider for intelligent document processing should not require tons of data to train their AI engine, and should not take months on end to bring about the business transformation you envisage and deserve. This should not devolve into a costly, long-drawn service engagement.
- End-To-End And No Halfway AI-Powered Automation: Many IDP vendors only provide extraction capability and necessitate investment in an RPA tool from a partner vendor to complete the offering. Such a composite solution may be made to work, but because of moving parts is bound to complicate the deployment and support of such a solution. Look for a vendor that has a one-stop solution and a single throat-to-choke support model.
AI-powered automation is ready to transform the insurance industry, and intelligent document processing is the stepping stone on this journey. While embarking on the journey, judicious, well-researched vendor choice is key to success.
Seasoned executive with roles at LexisNexis, HP, Dell, VMware & scaled startups to successful exits. Chief Product Officer at Kanverse.ai