AI Briefing
KO

How can claim payments be made as soon as an accident is reported?

·2024.12.24 19:00

Key point

This covers how Kakaopay Insurance is automating its claim payment process through legacy system decomposition and the introduction of Document AI.

Details

Kakaopay Insurance is exploring technical solutions to relieve customer inconvenience, with the goal of accurate and fast claim payments.

First, to address a massive Monolithic legacy system spanning 70 million lines, the company introduced Scavenger, a runtime dead code analysis tool. Through this, unused code was visualized and efficiently removed, and the claims service was split into claim-channel, claim-core, and margarine, securing the system's scalability and maintainability.

In addition, to automate the document review process that had previously been performed manually, Document AI technology was introduced. By leveraging TrOCR (Transformer-based OCR), a systematic pipeline consisting of Edge Detection, Layout Analysis, OCR, and Parsing was built, creating an environment capable of quickly and accurately recognizing various forms of receipts and documents.

Throughout this process, technology stacks such as Spring Boot, Spring Kafka, and Spring Cloud were actively utilized to simultaneously improve fault isolation and development productivity.

This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.

Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.