AI-Driven Full-chain Management of Outstanding Claims Reserves—The "SkyNet" System Qorus-NTT DATA Innovation in Insurance Awards 2026

1.1 Innovation Objectives

This project aims to construct an "AI Decision Brain" to empower full-chain management of outstanding claims reserves, enabling intelligent and fine-grained control over outstanding cases. The AI Decision Brain is a decision engine centered on a case-specific outstanding reserve estimation model, integrating multi-functional modules such as intelligent case task allocation, loss adjustment strategy, and risk early warning.

Regarding case-specific outstanding reserve estimation, the project will develop a multidimensional evaluation model incorporating various factors including injury progression, litigation probability, payout escalation trends, and judicial environment. This enables real-time, dynamic, and precise calculation of outstanding amounts, addressing the traditional challenges of inaccurate manual loss estimation characterized by significant deviations, and striving toward the ambitious goal of achieving near-zero bias in outstanding loss estimation.

In terms of intelligent case task allocation, the system will automatically generate case follow-up tasks and handling strategies based on case typology. Frontline personnel will only be required to execute standardized procedures, significantly reducing subjective judgment pressure and operational workload.

With respect to damage assessment output, the system will establish a standardized and intelligent mechanism for generating and delivering assessment recommendations, aligned with negotiation and settlement strategies. This enhances the success rate of frontline negotiations, mitigates the loss of potential savings and cost leakage caused by the lack of standardized support in negotiation and assessment processes, and targets annual loss reduction of over 200 million RMB, maximizing professional contribution to loss optimization.

In terms of risk early alerting, the project will build a comprehensive risk identification and early warning system covering fraudulent incidents, fabricated losses, and high-risk entities. This addresses the difficulty in detecting fraud risks and significantly enhances the capability to identify fraudulent claims.

1.2 Innovation Background and Current Challenges

Outstanding case reserve management is fundamental to regulatory compliance, essential for the sound operation of insurance enterprises, and critical to enhancing customer experience.

First, outstanding reserve management constitutes the cornerstone of regulatory compliance. Regulatory mandates explicitly require insurers to prioritize and strengthen the quality management of reserving base data, ensuring authenticity, completeness, consistency, and validity. Second, it safeguards the accuracy of externally disclosed financial information. Precise outstanding loss reserve provisioning directly impacts the integrity of financial cost reporting, the reliability of solvency assessments, and the credibility of public disclosures. Third, it serves as a vital foundation for internal business decision-making. Outstanding claims data are key inputs into underwriting and pricing models, and their accuracy directly influences the rationality of product pricing and market competitiveness. Finally, effective reserve management is closely linked to improved customer experience. "Difficult claims settlement" is the leading cause of insurance-related complaints, accounting for over 50% of all complaints, with 60% of these stemming from delays or inefficiencies in handling outstanding cases.

Despite its critical importance in regulatory compliance, financial stability, operational decision-making, and customer experience, the entire insurance industry faces persistent challenges in achieving transparency and effective control in outstanding reserve management—driven by inherent unpredictability in claim development and inadequate monitoring of subjective adjustments.

The insurance industry widely faces the common challenge of "inaccurate outstanding loss estimation" and "significant estimation bias." According to data from the first half of 2025, for every CNY 10 billion in claims, Ping An's outstanding loss estimates exceeded final settlements by CNY 90 million, PICC Underwriters underestimated by CNY 200 million, and CPIC overestimated by CNY 160 million. Such deviations not only distort financial forecasts but have also triggered regulatory scrutiny, with more than ten insurers receiving regulatory penalties ranging from warnings to business suspension in 2025 due to manipulation of outstanding reserve data.

Claims decision-making confronts a core challenge of "inherent unpredictability," primarily attributable to prolonged claim durations and high dynamism. Key factors include: (1) Non-linear injury progression influenced by treatment trajectories and individual variability, rendering early-stage estimation of ultimate losses highly uncertain; (2) Uncertainty regarding whether a claimant will initiate litigation or accept mediation, which is affected by emotional state, financial condition, and social dynamics—factors lacking robust quantification, thereby complicating litigation risk prediction; and (3) Regional disparities in compensation standards for disability benefits, loss of earnings, and nursing care costs, coupled with inconsistent judicial interpretation and broad judicial discretion, leading to significant variability in award amounts for seemingly similar cases.

A systemic "latency" issue exists in reserve management. Due to the absence of a real-time monitoring and early warning indicator system during the claims lifecycle, enterprises are forced to rely on ex-post metrics—such as year-end loss development deviations—for assessment. Management mechanisms are thus heavily dependent on retroactive penalties, resulting in delayed problem identification, increased remediation costs, and a notable lack of foresight and proactivity in managerial decision-making.

1.3 Involved DepartmentThis project has established a diversified, interdisciplinary collaboration team spanning multiple business functions and professional domains, fully leveraging the synergistic advantages of business and technical expertise to ensure efficient progress throughout all phases of the project, including requirements analysis, solution design, technical implementation, and operational deployment. The core participating teams include:

(1) Actuarial Analytics Office, Claims Operations Center, Ping An Property & Casualty Insurance (Business Lead)  As the business lead of the project, this department is responsible for analyzing and defining core operational pain points in claims processing, defining business rules, ensuring the technical solutions are closely aligned with real-world business scenarios, and guaranteeing that the project outcomes are implementable and deliver tangible business value.

(2) Data Intelligence Team, Technology Development Center, Ping An Property & Casualty Insurance (Technical Development)  

As the technical backbone of the project, this team is responsible for data acquisition and processing, intelligent algorithm development, and system architecture design and implementation. Leveraging big data and artificial intelligence technologies, it constructs efficient and intelligent solutions to drive automation and intelligent upgrading of business processes.

Through the deep integration of business and technical expertise, the project has established a collaborative mechanism of "business-driven, technology-empowered" operations, providing a solid foundation for high-quality project delivery and continuous optimization.