State Auto Improves Processes across the Life Cycle Using AWS Machine Learning, Computer Vision, and Serverless Architecture

State Automobile Mutual Insurance Company (State Auto) wanted to better understand and anticipate the requirements of its customers to provide them with the information they needed before they even knew to ask for it. The company began using technology to help its customer service representatives (CSRs) meet quality and customer satisfaction score goals when it built its SA360 solution on Amazon Web Services (AWS). By using data-fueled insights and making these insights available to customers and CSRs, State Auto was able to build a better service experience, redirecting typical customer calls to self-service channels so that CSRs were able to focus on those customers with more complex needs. Following the success of this project, State Auto began using more AWS services, including machine learning (ML), computer vision, and serverless services, to help further its goals in other areas such as the automation of the underwriting processes and early detection of fraudulent claims to expedite case review.

By using managed services for computer vision from AWS, State Auto has been able to automate processes that it previously performed manually. To simplify the property inspection process, the company uses Amazon Rekognition, which automates image and video analysis with ML and offers pretrained and customizable computer vision capabilities to extract information and insights. State Auto is able to use Amazon Rekognition to tag non-industry-specific risk factors in photos and videos from property inspections, giving risk engineers more time to focus on identifying industry-specific aspects in their risk assessments. To train State Auto’s ML models, the company extensively used Amazon SageMaker Ground Truth, a data labeling service. After the model went live, the company also used Amazon Augmented AI (Amazon A2I)—an ML service that makes it simple to build the workflows required for human review—for active user validation.

State Auto also uses Amazon Textract—which automatically extracts printed text, handwriting, and data from scanned documents—to index documents using a wide variety of business optimization use cases. “Our use of Amazon Textract touches almost every step in the insurance life cycle,” says Ramanujam.

Beyond managed services, State Auto is also using serverless architecture to power its antifraud solution, which combines the expertise of the company’s business users with data-driven insights. The solution was built on AWS Lambda—a serverless, event-driven compute service that lets companies run code for an application or backend service without provisioning or managing servers—and uses AWS Step Functions. The whole solution was seamlessly brought together using AWS Glue—a serverless data integration service that makes it easy to discover, prepare, and combine data for analytics, ML, and application development. After a hassle-free deployment of the fraud detection service, State Auto can now assess 83 percent more total claims for potential fraud than before, identify suspicious claims 3 days earlier in the claims process, and catch the 20 percent of claims that would previously have gone unflagged.