Resource Intelligence Hub Qorus-NTT DATA Innovation in Insurance Awards 2026

I. Concept and Objectives

The Resource Intelligence Hub is a strategic resource scheduling and value-creation platform developed by Ping An Property & Casualty Insurance to lead the industry's digital and intelligent transformation. At its core, it implements the traditional model by establishing a data- and algorithm-driven smart business model that exchanges repair resources for premium income. The platform is designed to holistically reshape and intelligently upgrade the fragmented and inefficient repair resource dispatch process in auto claims handling.

Our overarching goal is to build a multi-win ecosystem: for customers, we deliver a hyper-personalized service experience defined by complete transparency and precise matching; for over 110,000 partner repair shops, we establish a fair, stable, and highly efficient resource supply and capacity conversion mechanism that helps reduce costs and increase revenue; and for Ping An itself, we drive a paradigm shift from human-expertise-driven to system-and-algorithm-driven operations—maximizing comprehensive resource efficiency and optimizing overall cost structure. In doing so, we are forging a sustainable path for value growth in an intensely competitive, zero-sum market.

II. Project Background

Externally, the deepening of comprehensive motor insurance reform has pushed the industry into a fiercely competitive, zero-sum market where cost reduction, efficiency improvement, and quality enhancement have become the lifelines of insurers. Premium growth is under pressure, and controlling claims costs while elevating service experience is now a make-or-break imperative. Internally, Ping An P&C faced mounting operational pain points that rendered its traditional claims resource management model unsustainable. These systemic bottlenecks manifested across three dimensions: First, from the perspective of partner repair shops, resource allocation relied heavily on offline manual coordination and rudimentary rotation rules, lacking transparency and fairness. This led to the ice and fire paradox—some shops suffered from idle capacity while others were overwhelmed with backlog—severely dampening partner enthusiasm and compromising service consistency. Second, from the customer service perspective, policyholders involved in accidents were often given generic, distance-based recommendations that failed to accommodate their multidimensional preferences—such as repair quality, brand affinity, expected wait time, or even expertise in servicing specific vehicle models. Fragmented service touchpoints, such as missing information about third-party vehicles or the absence of alternative recommendations after rejection, resulted in a disjointed experience and high decision costs. Third, from the company's operational standpoint, internal dispatch processes long depended on manual spreadsheets and non-intelligent systems, leading to slow response times, lengthy rule-adjustment cycles, and significant execution deviations. This caused misallocation of valuable repair resources and loss of revenue potential. Moreover, the absence of end-to-end data traceability and attribution capabilities made it difficult for management to accurately assess the effectiveness of push-repair strategies, severely limiting scientific decision-making and continuous improvement. These interwoven challenges formed the fundamental drivers behind the launch of the Resource Intelligence Hub—a project designed to tackle deep-seated problems through top-level planning and technological innovation.

III. State of competition

Currently, most insurers remain in the early stages of digital transformation in claims resource dispatching. Peer systems for repair shop recommendation are largely rule-based (e.g., round-robin, fixed priority) or rely on simplistic static models such as pure distance calculation. These approaches are inherently static and experience-locked, incapable of adapting to real-time fluctuations in shop capacity, understanding complex customer preferences, or achieving dynamic, multi-objective equilibrium among customer satisfaction, shop utilization, and company cost. As a result, resource allocation efficiency remains low, and push-repair acceptance rates along with customer satisfaction have hit persistent ceilings.

The design of our solution draws inspiration from the deconstruction and creative synthesis of best-in-class digital practices across industries. We borrowed the real-time dynamic dispatch logic employed by super-platforms like Meituan and Didi, and built a data middle office powered by Flink stream processing to enable minute-level sensing of workbay availability, parts inventory, and technician status across a vast network of partner shops. We also adopted the hyper-personalization ethos of leading e-commerce and content platforms, constructing a 360-degree customer profiling system that integrates in-app behavioral data, claims history, and other multi-source data, while leveraging large language models (LLMs) and NLP techniques to deepen customer understanding.

Yet we did not stop at imitation. Confronted with the unique two-sided market dynamics and complex constraints of the insurance-specific "resources-for-premium" model, we pursued deep, original innovation. We pioneered the application of reinforcement learning—specifically the Upper Confidence Bound (UCB) algorithm in the Multi-Armed Bandit framework—to resolve the fundamental "exploration vs. exploitation" dilemma in resource allocation: how to nurture promising new shops while leveraging proven high-performers. We further integrated NSGA-II, a multi-objective optimization algorithm from operations research, to scientifically navigate the impossible triangle of customer experience, partner shop performance, and corporate cost. Through these breakthroughs, we have built a formidable technological moat and achieved a generational leap beyond conventional industry solutions.

IV. Departments involved

The project was spearheaded by the motor insurance agency channel as the primary driver of business value, with deep integration of expertise from individual agency, motor insurance, claims, and other core front-, mid-, and back-office functions. The corporate technology team provided end-to-end technical support, spanning data middle office development, algorithmic modeling, and systems integration.

The cornerstone of our success lies in an agile business-technology fusion collaboration and decision-making mechanism embedded throughout the project lifecycle. Business units do not merely submit requirements; they lead the entire process—defining business scenarios, designing commercial rules, closing the value loop, and driving frontline adoption. Technology teams, in turn, are embedded from the outset, acting as co-creators who translate complex business logic into iterative, computable intelligent systems using cutting-edge capabilities.

Concretely, we established a tripartite joint decision-making mechanism involving motor agency, motor insurance, and claims, alongside a data-sharing protocol and a continuous feedback collection system. We ensured that this strategic platform was not only well-built but also deeply utilized and continuously refined—transforming technological prowess into enduring business competitiveness.

V. Main results so far

Since its launch, the platform has delivered significant, quantifiable economic and operational benefits nationwide, robustly validating both the foresight of its business model and the superiority of its technological approach.

On the customer experience front, by incorporating over 40 factors—such as frequently used addresses and historical behavior—to enable precision matching and personalized recommendations, the push-repair acceptance rate has soared. Average customer decision and wait times have been reduced by over 30%, service satisfaction has notably improved, and the customer churn rate directly attributable to service dissatisfaction has dropped by 5 percentage points.

On the partner ecosystem side, the intelligent dispatching model—which holistically considers shop capacity, location, service capabilities, and more—has enabled fair, transparent, and highly efficient allocation of repair resources. Overall resource utilization and throughput at partner shops have increased by 15–20%. For underutilized or newly opened shops, monthly production target achievement rates have improved by an average of over 25%, greatly enhancing channel stickiness and ecosystem vitality.

Internally, the achievements have been the most fundamental and systemic: First, we have institutionalized a full-cycle management framework—inventory resources, formulate strategy, monitor execution, conduct operations, and support review—that enables automated, intelligent dispatch for over 800,000 claims per month, dramatically reducing manual intervention and management complexity. Second, through precision resource allocation and transparent process oversight, we have contributed meaningfully to optimizing overall claims costs. Third, and most importantly, we have successfully validated a complete value conversion loop: from intelligent dispatch, resource exchange, to channel empowerment and premium growth. This has established a replicable and scalable new operational paradigm for Ping An P&C as it navigates the deep waters of motor insurance digital transformation, laying a solid foundation for sustained competitiveness in the years ahead.