MediConCen automates medical insurance claim processing with Gemini and Vertex AI

Use case typeFraud detectionUpdated Jun 13, 2026

MediConCen is a Hong Kong-based InsurTech company that automates medical insurance claim processing for insurers and medical institutions. The company identified that OCR alone still left claim decisions dependent on manual review because extracted text needed better understanding and summarization. Using Google Cloud, MediConCen launched an AI-enabled claims processing platform in April 2025 to improve extraction accuracy, speed, security, and multilingual processing.

Organization
MediConCen
Industry
Insurance
Location
Hong Kong
Published
April 2025

Reported outcomes

+100%

quantified impactAutomation & deflection

+98%accuracy

Strategic outcomes

New product / capabilityLaunched AI-enabled claims processing platformSpeed & agilityReduced claim processing to hoursNew product / capabilityEnabled multilingual claim understandingCustomer experience & trustFewer customer complaints

Primary read

Use case focus

Showing 3 of 3

  • 1Claims processing automation
  • 2Document understanding
  • 3Fraud detection
  • OCR-generated text still required manual review before insurance claim decisions could be made.
  • The claims workflow needed higher extraction accuracy, higher straight-through processing, faster turnaround, and strong security controls.
  • The company also wanted to support multilingual claim understanding and future fraud detection.
  • MediConCen built an AI manager using agentic AI that breaks claims handling into hundreds of subtasks and orchestrates nearly 100 Gemini instances.
  • The team used Gemini for medical-content understanding and Google Cloud for enterprise-grade security.
  • Vertex AI was used to fine-tune the generative AI models for better domain-specific comprehension.
  • Cloud Run was used to deploy the claims processing system for scalability and CI/CD support.
  • Cloud Key Management was used with customer-managed encrypted keys to strengthen data protection.
  • The company plans to use BigQuery for fraud detection based on similarity in claim content.
  • Field-level data extraction accuracy increased to 98%.
  • Claim processing time was reduced from days to hours.
  • Straight-through processing rate increased by 100%.
  • Fine-tuning typically completed in less than one day.
  • Customers reported fewer complaints due to improved efficiency and consistency.
Architecture

The claims workflow uses an agentic AI manager that decomposes insurance claim handling into hundreds of smaller tasks and orchestrates nearly 100 Gemini instances. The platform runs on Cloud Run, uses Vertex AI for model fine-tuning, and protects data with Cloud Key Management customer-managed encrypted keys. BigQuery is planned for fraud detection.

Sources & evidence2
Groundedness: 4/5

AI-generated summary. Verify important details with the linked sources before relying on this case.

Explore related AI use cases

Was this useful?

Similar cases