AI Digital Underwriter: An End-to-End Intelligent Decision System Powered by Reinforcement Learning Qorus-NTT DATA Innovation in Insurance Awards 2026

1.1 Core Concept & Vision

In an era of increasingly fragmented and complex risks, this project aims to build the industry's first "Full-Stack AI Digital Underwriter." Beyond being an automation tool, it is a digital entity equipped with "human-like logical reasoning" capabilities. By integrating Large Language Models (LLMs) and Reinforcement Learning (RL), we have achieved full-chain digitization from business development, quotation, risk assessment to underwriting decisions. The vision is to break the productivity bottleneck of "experience-driven manual underwriting," establishing a digital underwriting hub with second-level response speed, precise cost control, and comprehensive coverage.

Vision: Transition from "human control" to "intelligent digital control," freeing underwriting experts from tedious compliance work to focus on high-value risk model iteration and complex catastrophe risk management.

1.2 Innovation Drivers & Industry Context

Traditional underwriting faces three key challenges:

1. Workload Overload: Conflict between surging business volumes and scarce underwriting experts.

2. Risk Penetration Difficulty: Emerging risks (e.g., new energy, drones, environmental liability) lack historical data, making manual identification extremely challenging.

3. Low Feedback Timeliness: Frontline sales agents often wait hours or days for quotes, leading to client attrition.

The Digital Underwriter addresses these through three core values:

• Intelligent Substitution for Efficiency: Free senior underwriters to focus on market strategy research and major client acquisition.

• Risk Quality Precision Control: Standardize risk dimension judgments to ensure consistent underwriting quality and eliminate human errors.

• Scenario-Based Efficiency Boost: Reduce quotation time from hours/days to seconds for simple cases.

1.3 Competitive Landscape

While AI adoption accelerates across industries, Ping An Insurance distinguishes itself as a leader by focusing on complex group property insurance scenarios. Our solution dissects the professional quotation process into five core modules: underwriting Q&A, policy pre-verification, risk analysis, document pre-check, and underwriting strategy formulation.

Vertical Expertise:

• Legal interpretation for environmental liability insurance.

• Special risk identification and new energy industry chain risk assessment.

Future Roadmap: The Digital Underwriter will learn full underwriting decision workflows, generating actionable underwriting recommendations—a pioneering advantage in the insurance industry.

1.4 Inspiration

During R&D, AI developers observed underwriters cross-referencing standard underwriting policies to determine compliance. This inspired the creation of a policy-reading AI system, giving birth to the Digital Underwriter concept.

1.5 Involved Departments

Led by the Group Business Unit of Ping An Property & Casualty Headquarters, this project involves:

• Technology Center.

• Group Technology Division.

• Huazhong University of Science and Technology.

• 42 branches (Guangzhou, Hubei, Anhui, Shanghai, etc.) with senior actuaries having 15+ years of experience.

1.6 Key Achievements

Since launch:

• 66.7% adoption rate (↑19.5pt from initial phase).

• first-quote time (40% efficiency gain).

• Covers 66% of group mid-to-large client business.