The Challenge
Silverbrook Insurance is the client behind this project. A San Francisco insurance provider managed policyholder claims across multiple categories. It wanted to reduce manual assessment work and move claims through the process more efficiently.
Complex documents, image-based damage assessment and varied policy rules made automation challenging. The platform also needed to identify potentially suspicious claims and support third-party integrations.
What ApexCore Digital built
ApexCore Digital combined OCR-based extraction, AI claim classification and computer-vision damage analysis. Fraud-risk scoring and automated policy-rule validation support assessment and routing.
Claim registration connects with document extraction, severity assessment and an adjuster dashboard. Automated routing helps send claims for appropriate review. The solution uses Angular and Spring Boot alongside Python-based AI components, with Kafka, Redis, PostgreSQL and AWS S3 supporting processing and storage.
How the workflow supports the business
The claims workflow begins with registration and document information, then brings policy checks, damage analysis and risk scoring into the assessment process. Automated routing sends claims for the appropriate review, while the adjuster dashboard organizes the information needed for handling the case. Extraction quality, fraud detection and processing time measure different parts of this workflow. The platform is designed to assist claims professionals by reducing repetitive handling and bringing assessment inputs together; the supplied scope does not establish automatic final settlement decisions.
Reported Results
- 94% document extraction accuracy.
- 91% fraud detection accuracy.
- 60% reduction in claim processing time.
- 65% reduction in manual processing.
Technology: Angular, Java, Spring Boot, Python, NLP, XGBoost / TensorFlow, Kafka, Redis, PostgreSQL, AWS S3.
