On August 22, Taikang Insurance Group officially unveiled its "Taikang Health and Elderly Care Large Model 1.0", once again thrusting the artificial intelligence race in the insurance sector into the public spotlight. As AI technology continues to evolve, vertical large models are rapidly transitioning from concept to industrial application, positioning the insurance industry as a key participant in this technological wave.

Unlike tech companies that develop medical large models aimed at general capabilities and consumer-facing C-end markets, insurers entering the vertical large model space are primarily anchoring their efforts in core business operations. They seek breakthroughs through closed-loop scenarios that integrate insurance payments, medical services, and elderly care facilities. What's even more promising is that the open-source wave is continuously lowering the barriers to entry for large models, with "technology democratization" opening a window of opportunity for small and mid-sized insurers. As industry insiders note, the decisive factor in the future will not be model size, but the ability to rapidly translate AI into operational efficiency and customer value. By focusing on niche scenarios, strengthening data governance, and leveraging ecosystem partnerships, smaller institutions can also achieve a "curve overtaking" in the wave of intelligent transformation.

As large model technology moves from "proof of concept" to "large-scale deployment", the AI competition in the insurance industry has entered its "second half". Notably, insurers of varying sizes and resource endowments are charting differentiated paths on the large model track, each aligned with their strategic positioning and resource accumulation.

Leading insurers such as Ping An Insurance, PICC, and New China Life Insurance are primarily leveraging large model technology to build comprehensive intelligent application systems. For instance, Ping An Insurance is advancing its "AI in All" strategy, with its Kunpeng intelligent agent platform equipped with nearly a thousand professional skills. At Ping An Property & Casualty, artificial intelligence has been fully integrated across the entire value chain of insurance services, marketing, operations, management, and administration. Meanwhile, New China Life Insurance has launched 11 business intelligent agents covering customer service, agent enablement, policy processing, risk prevention, and office compliance across all scenarios.

In contrast, institutions like Taikang Insurance Group and Taiping Reinsurance are focusing on niche tracks to develop proprietary vertical large models, forming a differentiated competitive landscape. For example, Taiping Reinsurance (China) has introduced "Ruishu", an AI intelligent agent for the reinsurance vertical domain. This system follows a "technology generalization-capability specialization-scenario verticalization" approach, integrating large language models with reinsurance expertise to build a closed-loop framework of "intent understanding-knowledge reasoning-decision generation". On the other hand, Taikang Insurance Group's "Taikang Health and Elderly Care Large Model 1.0" is based on mainstream large model foundations, trained using data accumulated from its five medical centers and 32 operating elderly care communities, driving deep integration of AI with real-world scenarios in healthcare, elderly care, rehabilitation, nursing, and health management.

Zhou Jin, a consulting partner at Baker Tilly China's financial services division, noted that the development of large models relies on four key elements: algorithms, computing power, data, and scenarios. Algorithms and computing power require massive foundational investment and rapid iteration, making it unsuitable for the insurance industry to over-invest in these areas. Instead, insurers should adopt cooperative or leasing models to access these resources. "However, data and scenarios possess high industry specificity and are heavily dependent on insurers' specific operations in sales, service, risk control, and investment, making them critical areas for industry AI deployment," Zhou Jin explained.

Using Taikang Community as an example, Chang Cheng, Deputy General Manager of Taikang Insurance Group's Technology Center, believes that years of accumulated data from healthcare and elderly care services have formed large-scale, continuous real-world records of the longevity population, covering multiple dimensions such as health assessment, health management, chronic disease follow-up, long-term care, and rehabilitation intervention. These datasets are characterized by long time series and contextual continuity, representing premium vertical resources that generic public datasets cannot match.

Data from listed insurers indicates that with AI support, work processes and efficiency have simultaneously improved. For instance, China Life Insurance has driven the deployment of a series of large model applications, achieving an average claims processing time of just 0.36 days. Sunshine Insurance has launched its independently developed "AI Customer Operations Assistant", capable of responding within seconds and rapidly generating personalized customer operation plans. ZhongAn Online leverages AI technology to empower its health ecosystem, having served over 150 million health insurance users cumulatively.

The "White Paper on Deep Empowerment of the Insurance Industry by Large Model Technology (2025)" indicates that within a trustworthy, controllable, and auditable governance framework, large model technology will accelerate the insurance industry's shift from the traditional logic of "passive claims and post-hoc compensation" toward full-process management covering pre-event, in-event, and post-event stages, thereby nurturing a "new paradigm of intelligent agent-driven insurance".

Due to high investment costs, publicly disclosed self-development and increased AI investment currently come primarily from large and mid-sized insurance institutions, while smaller insurers more often adopt cooperative approaches to implement AI applications. "Large insurance companies hold significant advantages over smaller ones in data infrastructure, professional talent, customer base, application scenarios, and financial strength, leading to earlier AI deployment and greater investment," Zhou Jin stated. However, in terms of investment strategy, aside from the heavy foundational investments made by some companies before the advent of DeepSeek, the industry generally adopts a relatively "light model" approach, avoiding excessive spending on the "heavy" aspects of algorithms and computing power while focusing more on application scenarios with industry-specific vertical characteristics.

Nevertheless, subtle differences persist between large and small companies. Large companies possess better data governance foundations, making them more proactive in leveraging general-purpose large models for training, fine-tuning, and deploying applications tailored to their vertical business scenarios. "To some extent, current technological development represents a form of 'technology democratization'. The gap between large and small companies in infrastructure and general models is narrowing rather than widening, thus providing smaller companies with more opportunities to catch up and even surpass," Zhou Jin observed. He emphasized that how small companies can quickly realize AI's application value by aligning with their business, operational, and process characteristics is a critical strategic choice and implementation issue requiring swift decision-making and investment.

"The future key to AI competition is not who possesses the largest model, but who can better convert AI into operational efficiency and customer value. For small and mid-sized insurers, this may actually present an opportunity for differentiated development," said Zhu Junsheng, a postdoctoral fellow in applied economics at Peking University. He noted that opportunities for small and mid-sized insurers lie more in scenarios than in the models themselves. This involves four approaches: first, focusing on niche segments to create distinctive scenarios; second, strengthening data governance and knowledge base construction; third, leveraging technology companies and industry platforms to acquire AI capabilities; and fourth, promoting intelligent transformation across marketing, claims, customer service, and risk control processes.