stethoscope on desk with a laptop near it and papers

AI-Powered Healthcare Benefits Platform Slashes Validation Time From Weeks to Seconds

A medical plan benefits provider leverages AI-powered document processing to slash benefit validation time from weeks to seconds.

Industry

Healthcare

Teams & Services

/DevOps /Backend /Dependent Life Event Verification (DLEV)

Tech & Tools

/Bedrock /Claude 3 /Parameter Store /Step Functions /Lambda /API Gateway /ECR /Python

Key Data Points

Reduction in processing time from 2+ weeks to nearly instantaneous
Drove consolidation of identification points to ease identification efforts
Generalized processing of document types to scale the addition of new information

The Vision

Leverage advanced AI capabilities to increase accuracy, scalability, and efficiency in document processing to drive faster turnaround for customers.

The Goal

As the platform grows, the amount of time spent validating documents by hand is increasing exponentially. This cumbersome and error prone process will be dramatically optimized by leveraging AI to automate the identification process and speed up turn around time.

The Challenge

The current identification process focused on easing the needs of the person doing the validation. While this process worked well for some time, the business’s rapid growth introduces bottlenecks to critical customer facing processes. The current manual evaluation of documents does not scale and will be no longer sufficient in the near future.

With this challenge in mind, Bswift set out to add automation to their document pipeline to help build out a more scalable process. By leveraging Intelligent Document Processing (IDP) the goal is to reduce time to feedback for their clients and more quickly identify those documents that need to actually be manually validated.

The Solution

Understanding this will save a considerable amount of time and effort, bswift partnered with Protagona to establish a pipeline to automatically identify pieces of information for specific documents that made up a majority of their processing. Leveraging Amazon Bedrock in conjunction with Claude3, an idea was tested to validate outcomes of a common document.

This idea was then expanded, and further generalized so that as bswift’s client base grows, and potential uses of the pipeline scale out, the process can be applied across additional business units internally.

OUTCOMES

Your data is trying to tell you something

Contact us

... are you listening?