This global company asked Trility back to enhance its machine-learning (ML) based claims software platform with new functionality and scale capacity for new customers. These enhancements helped attract its largest customer and increased the number of documents processed without manual intervention by its engineering team.
Trility initially developed a Minimum Viable Product (MVP) to automate the client's manual claims process. This automation focused on attracting new customers by streamlining claim processing, reducing operational costs per claim.
As the claims software acquired more customers, the client needed to scale the performance capacity by 1,000 percent and reduce manual interventions to keep the system operational during errors. Another key enhancement was allowing multiple invoices to be submitted with a single upload to reduce effort for their customers.
Trility used a two-pronged approach to help the client manage an increasing number of documents: streamlining document processing and enhancing automation capabilities.
First, the client wanted to improve its machine-learning platform by upgrading its document processing capabilities. Trility installed a third-party Optical Character Recognition (OCR) application that allowed the platform to automatically split multiple invoices out of a single uploaded document. This time-consuming task previously required the client’s end-users to submit them one by one.
Secondly, Trility collaborated with the stakeholders to identify and address obstacles to automation. The team identified and fixed the underlying issue that affected six other projects, reducing error rates from 33 to 8 percent.
In addition, the team focused on understanding the system and identifying bottlenecks, providing the client with observations and recommendations that enabled its team members to continue to tackle more pressing issues based on scaling the solution or reducing costs.
The client reduced manual reviews by 50 percent and added functionality for more flexible matching, resulting in a 40 percent increase in matched invoice items. This led to a faster claim experience for their customers and reduced costs associated with third-party manual reviews. It also enabled the client’s engineers to focus on revenue-generating activities and avoid context-switching for manual intervention.
Trility’s expertise in optimizing flow and solving impactful problems led to a significant increase in the speed of the most-used API calls by over 3,000 percent with caching, indexing, and reducing unnecessary metadata usage. These performance improvements provided internal stakeholders with greater confidence as they onboarding new customers.
A testing strategy enabled small, frequent deployments without constant QA oversight and streamlined debugging, boosted confidence with each code change, and met client speed expectations while overdelivering on quality.
