NEOS is redefining customer service in the life insurance industry with generative AI
Cevo has partnered with leading life insurance brand NEOS to use generative AI to radically improve the efficiency of managing large volumes of email correspondence. The proof-of-concept exceeded expectations, demonstrating the significant potential of generative AI in streamlining email processing, reducing response times to customer service requests, enhancing service quality and boosting team productivity.
Generative AI
Insurance
Industry
NEOS is a specialist financial services distribution, administration and technology business that partners with leading brands, advisers, insurers, trustees and reinsurers to deliver insurance solutions to Australians. Since launching in 2018, NEOS has been redefining industry standards, winning numerous awards for service excellence, including the 2024 Adviser Ratings Most Consistent Retail Insurance Provider, 2024 Australian Wealth Management Awards Best Retail Insurance Provider, the 2023 Money Magazine Life Insurance Product Provider of the Year, and the 2022 Association of Financial Advisers Overall Life Insurance Company of the Year.
Business challenge
Managing and scaling a rapidly growing operation is challenging for all businesses, including NEOS.
NEOS identified an opportunity to streamline the management of the huge volumes of emails received from customers, financial planners and medical providers required to arrange and manage life insurance policies. This workflow allocation task was being done manually by a team of people. There had to be a better way.
So, it sought a strategic partner with specialised expertise for this task – Cevo.
Solution
Cevo assembled a multidisciplinary team of cloud and data engineers and AI specialists to analyse and solve the challenge of reading, categorising and allocating emails to the correct departments.
Working closely with NEOS’ customer engagement team, we successfully developed a pilot solution leveraging AWS Bedrock and several other AWS services.
Classification of structured emails using regular expression
We began by developing a solution to automate the classification of structured emails, such as claim forms and policy inquiries. This involved identifying common patterns within the emails and creating robust regular expressions to match these patterns. We then implemented basic rules for email classification, enabling NEOS to efficiently and accurately classify emails without manual intervention.
Classification of unstructured emails using a Claude 3 model
For the more complex task of handling unstructured emails, such as free form replies, general inquiries, policy information updates and complaints, we deployed AWS Bedrock’s advanced natural language processing capabilities. Utilising a pre-trained Claude 3 model, we enabled NEOS to process email content by normalising text, allowing for comprehensive analysis and classification of emails into pre-selected categories. This accurate classification is critical as it serves as the starting point for many downstream users, including underwriting new policies and handling documentation.
Implementation steps
We developed an event-driven architecture to operationalise this solution. The process began with the automation of email extraction and upload to Amazon S3 using AWS Lambda, which connected directly to a third-party mailbox provider. Orchestration and integration were managed through AWS Step Functions, where each new email triggered an event via Amazon EventBridge. This detailed process involved identifying, cleansing, and transforming email content into a structured format such as JSON, with customised rules to ensure accurate classification. During the data enrichment process, new data required for downstream processes was added by querying NEOS’ internal data platform.
This event-driven architecture can also connect other third-party providers, downloading health reports that contribute to the underwriting process analysis and will be used by downstream users. Finally, the email responses generated by AWS Bedrock were added to the event and made available for integration into another third-party provider.
This architecture supports NEOS in achieving faster and automated response times, improving service delivery, enhancing overall customer engagement and laying a foundation for future scalability and innovation.
Outcomes
This proof-of-value demonstrated to NEOS the significant potential of generative AI technologies to transform its customer engagement operations and set the stage for future innovation.
The pilot successfully streamlined the handling of both structured and unstructured emails, freeing up NEOS’ customer service team to focus on higher-value activities and improve operational efficiency. As a result, response times to customers were reduced, while maintaining and improving their service level agreement (SLA) response times. The comparison between the manual and automated email classification processes showed a positive match in approximately 75% of cases for structured emails and 50% for unstructured emails. The automated classification process significantly increases accuracy and demonstrates the value of this project to NEOS.
In turn, the automation of manual tasks during this pilot project has the potential to boost employee satisfaction and productivity. With routine tasks automated, employees can focus on higher cognitive tasks and engage in more rewarding activities. We also identified opportunities for improving data quality and email classification by addressing previously misclassified emails. This proof of concept lays the groundwork for integrating automation into customer service workflows and extending it to other back-office applications, paving the way for broader organisational benefits.
These outcomes demonstrate a clear path towards a faster time-to-market, enhanced productivity, reduced costs and a competitive advantage. As a result, NEOS is well-positioned to navigate a new era of growth and innovation, leveraging Cevo’s expertise to implement more automation and improved systems and processes.
Enjoyed this customer story?
Share it with your network!
You may also like
