ExaApp launched Hong Kong’s first AI-assisted document triage application at Hong Kong Baptist Hospital. The application supports admission teams by triaging documents received through eFax and email within hospital-controlled infrastructure.

The deployment marks an important step in bringing private AI into a sensitive, high-volume healthcare workflow where security, reliability, and human oversight are essential.

Admission documents can arrive through different channels and require staff to review and route each item appropriately. The new application provides a consistent AI-assisted first pass across documents received from eFax and email, helping teams organize incoming information for human review.

Integrated with the hospital’s appointment system, the application turns document content into actionable insights within the existing admission workflow. It simplifies document assignment and helps staff connect incoming information with the relevant appointment while retaining control over final handling decisions.

By supporting these repetitive steps, the system allows admission staff to focus their attention on exceptions, incomplete information, and decisions that require operational context.

The application combines three complementary AI technologies:

• vLLM-based language intelligence supports the understanding and triage of document content.
• Self-trained document segmentation AI identifies and separates relevant sections within mixed-format admission documents.
• Handwritten character recognition AI processes handwritten information that conventional text-only workflows may miss.

Together, these models structure incoming information and produce actionable insights for the connected appointment workflow before documents are presented for staff review. The AI pipeline runs inside the hospital’s controlled environment instead of relying on a public AI service.

The application was designed around healthcare security and industry compliance requirements. Its on-premise architecture keeps document processing within the hospital environment and reduces external data exposure, while a human-controlled workflow keeps admission staff responsible for final handling decisions.

This launch demonstrates how private, multi-model AI can support healthcare operations without separating the technology from the security, governance, and workflow requirements of the hospital environment.

For More Information

To explore how on-premise AI could support a document-heavy healthcare workflow, contact ExaApp to discuss the operational challenge and deployment requirements.