SLVRCLD | Automated content claim quantifications and settlements

Having claimants manually type in their claimed items in a text field during the First Notification of Loss (FNOL) gets rid of the old paper claim form, but is not structured enough for an insurer to automatically quantify the claim. This results in claims agent still needing to physically work through the item list to obtain prices from suppliers. Besides being costly to process it delays the claim finalisation, which is one of the largest customer complaints. The current manual process also prevents insurers settling claims through robotic process automation (RPA) if required.

We are able to do this as we have build-up a large dataset by analysing millions of content claims to accurately identify the claimed item, through machine learning determine the correct replacements and then quantify them from normalised supplier quotes. The claim can then be settled via the most appropriate method such as a voucher, pre-paid filtered debit card, purchase order or cash-in-lieu.

Insurers empower their clients by building the functionality directly into their digital FNOL process or breaking-out to a self-help portal.

Insurer's can also have policyholder's build-up a household content catalogue ensuring that their clients are sufficiently insured, speed up the claim process even further and prevent fraud during the claim process.

There are three main aspects to our solution stack: 

• the Data Engine (managing our product and pricing datasets)
• the Core Back-End (accessed via API and used to manage process-flow and data transfer) 
• the Management Portal (providing a visual interface with the APIs in a web-hosted application)

The heart of our system is the Data Engine. It is used to identify a claimed item, determine its best possible replacements, and to price each of these from a list of quotes that is constantly and automatically sourced. This requires the ingestion, transformation and analysis of millions of products and quotes spanning a multitude of suppliers in numerous countries. To do this, the Data Engine leverages several cloud-based technologies. It also employs a machine learning engine to help determine aspects of the product landscape, such as stock availability, regionality, product end-of-life, etc, without the need to integrate into each supplier’s inventory management system.

We chose a technology stack that would provide us the ability to rapidly develop and deploy changes, while maintaining auditability and security. We accomplish this by running a cloud native, mostly serverless architecture, utilizing Amazon Web Services. This is both scalable and cost effective.