AI-Driven 360° Comprehensive Insurance Underwriting Risk Assessment and Decision Framework Qorus-NTT DATA Innovation in Insurance Awards 2026

My doctoral research introduces an AI-Driven Property Insurance Underwriting Framework and prototype application (“Ryskore”) designed to modernize and streamline the end-to-end underwriting journey. The objective is to support underwriters with faster, more consistent, and more transparent risk assessments by combining document automation, hazard intelligence, explainable ML predictions, and structured human oversight.

The initiative was born out of a practical need: property underwriters spend significant time consolidating data from slip documents, engineering reports, loss records, and various hazard sources. This manual process is slow, inconsistent across teams, and vulnerable to data gaps. The research aims to standardize and augment underwriting decisions through AI while preserving human judgment.

The competitive landscape today consists mainly of partial solutions/OCR tools, exposure dashboards, rule-based engines, or specialized analytics vendors. Few platforms offer an end-to-end, unified workflow tailored to direct or facultative property underwriting insurance or reinsurance, and even fewer provide explainability layers suitable for auditability and regulatory expectations. This research fills that gap.

Sources of inspiration include leading reinsurers’ digital initiatives, academic research on AI-assisted decision systems, and discussions with senior underwriters across global markets. The framework aligns with enterprise requirements for transparency, governance, and hybrid human-AI collaboration.

The development required collaboration between IT, underwriting, risk management, and data analytics teams. It also involved partnerships with external AI services for document parsing and geospatial hazard sourcing.

To date, the project successfully produced a functional prototype capable of: - Automated extraction of key underwriting fields from slip documents - Retrieval of natural hazard and man-made risk indicators - Consistency checks and completeness validation - ML-assisted risk scoring across major COPE dimensions - AI-generated preliminary underwriting recommendations with human-in-the-loop approval - A unified digital review interface for decision rationale and audit trails

Initial internal feedback from underwriters and industry experts has been strongly positive, confirming the framework’s relevance, potential impact, and scalability.

The project is currently undergoing expert validation interviews and refinement for broader implementation with subject matter expertes at companies in Europe, Lebanon, and GCC.