Many enterprises in insurance vertical have transformed the underwriting process. Underwriting is about buying an insurance with policy. Some of the insurers might not have digital payment and have a lengthy process to deliver the policy to customer. The goal of these enterprises in insurance sector is to transform the legacy, be customer focus, add new rules and policies, regulator engagement, and create a value for the customers.
Underwriting engine consists of rules and policies. Underwriting engine works with an insurance platform which has workflow, API, dashboards, reporting, and third party integration features. Enterprises are trying to move the platforms to cloud and use microservices to redesign the engines. The goal is to reach 90% efficiency and automation. There will be minimal manual involvement to 5 %. These platforms are utilizing AI/ML techniques such as Simulation, NLP, and text mining.
![AI Simulation For Insurance Underwriting Engine Testing]()
Agile process methodology is used for the legacy transformation. This process methodology helps in delivering the modern platform in sprints with good quality. The challenge in the quality assurance area is testing the modern platform of legacy rules and policies. Typically the old insurance rules are in an excel sheet. Manually, the policies are created and checked for various scenarios of the customer depending on the type of insurance like auto, health, property & casualty, commercial, life, and travel. Testers need to check the generated policy quote and the policy for the various scenarios. This process is associated with huge cost implications.
No code platforms have made an impact to improve the legacy transformation by adding features for adding new policy modules with rules. Testing process is a challenge even in no code platforms where manual effort is involved. Underwriting transformation projects are using simulation for checking the rules and policy against the data base of legacy rules and customer profiles. AI is used in other areas like customer’s biometric analysis and customer profile test data generation. Test data is generated based on the legacy rules and policies. Test data includes the customer profiles and his interests in buying different types of insurance. After the test data is generated, simulation process is started by randomly selecting the customer profiles to start with, the selection process changes based on the output and the result of the tests. If the tests pass, the simulator moves to the next scenario. If the tests are failing, in the same scenario, various features are randomized to check for failure scenarios.
![AI Simulation For Insurance Underwriting Engine Testing]()
The breadth of scenarios are covered first and if all the tests pass, then the depth of each scenario is analyzed. Root cause of the failure is presented and the success is also explained using explainable AI technique. Explainable AI is about predicting the outcome with explanation. The test results are presented and explained why they pass and fail. Quality assurance team uses functional test automation tools for checking the application features. Similar to the functional test automation, simulator is built to run against the test suite which consists of test cases.
Simulator will have different methods to specify the test scenarios and verify them. It can optimize the execution process of the test cases. It can support multiple data sets for execution of the tests. Simulator can be run on different environments - QA, UAT, and Stage environments on the cloud and in-house. No Code simulator helps in improving the efficiency and quality.
The usage of simulator for legacy transformation of underwriting process is very common. Once the underwriting process is rearchitected, there will be a need for changing the test scenarios based on new policies and rules. No code simulator helps in cutting down the down time in the environment for introduction of new policies and rules.