Currently, the wave of AI intelligent agents, represented by so-called "AI agents," is sweeping across various industries, driving a technology-led transformation. The traditional operational systems dominated by human labor are being gradually restructured, heralding a new era of human-machine collaboration and multiplied efficiency. "Artificial intelligence is reshaping the operational paradigm of the insurance industry. With the rise of AI agents like these, the industry's original operational models may undergo fundamental changes," a senior executive from an insurance institution recently stated. The insurance industry's traditional operations often rely on multi-person collaboration and division of labor. As agent technology matures, a new framework will emerge where humans lead and AI handles most standardized processes. This deep restructuring of industry production relations is accelerating implementation in core business scenarios such as underwriting, claims assessment, compliance, and risk control.
The "Insurance AI Agent Skill Insights Report" indicates that the market supply of intelligent agent skills in the insurance sector is noticeably accelerating, with a strong focus on rule-intensive, document-heavy, high-frequency review, and strict professional calibration scenarios like underwriting, claims, and risk compliance. Currently, large model technology is driving the insurance industry's deep transformation from "dialog interaction" to "business empowerment." The application model centered on "AI agents + agent skills" is flourishing in the insurance field, becoming the core engine for the industry's intelligent upgrade.
Simply put, a Skill is a pre-installed "professional skill package" for AI agents, packaging business knowledge, operational processes, and compliance requirements into directly callable modules. This allows AI to efficiently complete similar tasks according to standard procedures without learning from scratch each time. For the insurance industry, which is inherently highly professional, process-oriented, and compliance-driven, the capabilities of general-purpose models alone are often insufficient to stably adapt to real business scenarios. Therefore, the significance of Skills is particularly pronounced—they can encapsulate vertical domain expertise such as policy clause comprehension, regulatory requirements, underwriting logic, claims rules, and review standards into callable capabilities, thereby significantly enhancing the agent's performance in professional depth, output stability, and scenario adaptability.
A recent report from Professor Xu Xian's research team at Fudan University's School of Economics, based on an evaluation of 539 public Skill samples, shows that post-2026, insurance-related capability components are rapidly entering a phase of concentrated release and continuous updates. From a business function distribution perspective, underwriting/claims and risk/compliance are the primary business scenarios. Among these, claims-related agent skills account for 13.65%, regulatory and compliance-related skills for 13.54%, and risk management skills for 13.00%, significantly higher than supportive business functions like finance, strategy, and human resources.
In terms of insurance types, there are 193 Skills dedicated to non-life insurance, only 52 for life insurance, and 298 generic Skills, constituting an absolute majority. This indicates that a large portion of the supply remains broadly designed for "general scenarios in the insurance industry."
Industry insiders believe AI has richer application scenarios in the non-life insurance sector. The aforementioned insurance executive cited auto insurance operations as an example. Under constraints like premium limits, speed limits, and discount restrictions, scientifically setting budget targets and rationally allocating auto insurance resources have become critical industry challenges. Budget formulation requires comprehensive consideration of multiple factors, including the sales capabilities of branch offices, business structure, policy costs, and regional new car sales volumes. "Currently, many branch offices rely on subjective, experience-based judgments and trend-following when preparing budgets, lacking refined calculations. They often lack quantitative basis for choices regarding business types, team division of labor, and the focus on passenger versus commercial vehicles or new versus used cars." In their view, solving such multi-dimensional, highly complex optimal budget problems is extremely difficult through manual deduction. In the future, leveraging AI to integrate various influencing factors could enable precise budget calculations and optimal resource allocation, significantly enhancing the operational efficiency of property and casualty insurers.
Addressing the insurance industry's complex, high-value, and trust-dependent nature, AI is empowering the agent system, covering the entire process from customer acquisition, outreach, to communication and conversion, serving as a professional sales assistant for agents. "Regarding AI empowerment for agents, our ultimate goal is to use machines to directly interact with and sell to customers, instead of people. In this process, we are experimenting and exploring across the entire agent lifecycle: onboarding, training, and management," the insurance executive stated. AI democratization has a clear role in raising the average level of agents—by standardizing and replicating professional capabilities and intelligent resources at scale, even average sales personnel can access the sales techniques and professional skills of top performers, providing strong support for business growth.
On the operations side, underwriting/claims and risk/compliance are the primary areas where AI applications are concentrated. The report suggests that current Skills are relatively concentrated in business segments where agent applications are more mature and structured knowledge extraction is more thorough within the insurance value chain. These two types of scenarios inherently rely on policy verification, document organization, issue identification, and report generation, making them most suitable for being encapsulated into reusable templates via Skills. "In contrast, functions like finance, strategy, and human resources, while also able to benefit from agents, often have more fragmented demands and are more deeply coupled with internal organizational processes. Therefore, they have not yet formed a mainstream supply in the public market," the report stated.
This data conclusion aligns with the general consensus among insurance industry leaders regarding the path of AI-driven industry transformation. Wang Xiaohang, Chief Technology Officer of Ping An Group and General Manager of Ping An Technology, stated that as intelligent agents evolve from providing advisory services to executing tasks and proactive management, AI is rapidly empowering and enhancing the efficiency of numerous operational roles and processes, accelerating the industry's shift towards being computation-intensive. Particularly in core areas like risk management, loss reduction, and insurance "underwriting/claims," new technologies enable more accurate product matching and customer selection, moving full-link risk control forward into the sales and operations环节, constituting a significant industry transformation.
From the practices of leading insurers, intelligent claims processing and intelligent risk control have helped companies achieve significant cost reduction and efficiency gains. For instance, in health insurance claims scenarios, China Pacific Insurance redesigned the human-machine collaborative workflow, establishing an intelligent claims processing system. In the first nine months of 2025, the intelligent claims project reviewed an average of over 80,000 claims cases monthly, with a review consistency rate of 96%, cumulatively generating gains and reducing losses by over 37.98 million yuan, significantly lowering operational and claims payout costs. Regarding AI risk control effectiveness, in 2025, Ping An P&C's intelligent anti-fraud claims interception reduced losses by 10.51 billion yuan, marking the third consecutive year of loss reduction exceeding 100 billion yuan.
The report also shows that when broken down into sub-functions, compliance review and internal control, along with data analysis and business intelligence, are significantly领先, followed by risk assessment and underwriting decisions, and claims review and adjustment. This indicates that the core value of insurance Skills, in the short term, lies more in "helping professionals complete preparatory judgments and document processing faster," rather than directly replacing complex business decision-making itself.
Insurance industry insiders noted that in the implementation of intelligent claims processing, the deep involvement of business teams is indispensable. Beyond assisting in refining review logic and annotating key scenarios, business teams actively participate in validating the output of large models and analyzing exceptional cases, forming a closed-loop mechanism of "application-feedback-optimization."