By Fan Hongmin, China Economic Daily, Beijing
Insurance agents are accelerating their onboarding to the AI open platform to compete for AI consumer entry points.
Recently, Alibaba Qwen expanded its agent ecosystem with over a dozen financial service agents, including insurance providers such as Xiaoyusan and HuiZe Insurance.
Meanwhile, the insurance agent on Ant Aifu App is expanding. On 10 month 10, Fosun Prudential Life Insurance and Ant Health announced they will build a dedicated agent on the Ant Aifu App to integrate insurance with professional health management. Previously, several insurers including PICC Property & Casualty, CPIC Health Insurance, CPIC Property & Casualty, Zijin Property & Casualty, CITIC-Prudential Life Insurance, and others have already deployed agents on the Ant Aifu App.
According to reports from China Business Journal, some insurers have received regulatory inquiries after joining relevant AI platforms. Regulators are focusing on the types of large models used by agents on these platforms and the extent to which user privacy data is utilized.
Capture the first touchpoint in consumer decision-making
On 9 7, Alibaba Qwen announced the launch of its first batch of financial agents, covering securities investment, funds, futures, insurance, and more. Among the insurance agents are Xiaoyusan and Huize Insurance.
In the above agents, Xiaoyusan is operated by Xiaoyusan Insurance Brokerage Co., Ltd., an insurance intermediary under Huihui Group (02621.HK), a Hong Kong-listed insurtech company. Huize Insurance is operated by Huize Insurance Brokerage Co., Ltd., an insurance intermediary under HUIZ, a Nasdaq-listed insurtech company.
Based on the capabilities outlined above, the insurance agent provides services including product recommendations, comparative consultations, clause interpretation, rule inquiries, family coverage needs assessment, and insurance planning.
Before Xiaoyusan and Huijie Insurance joined Alibaba Qwen, multiple institutions had already deployed AI agents on the Ant Aifu app. Currently, the app features several insurance AI agents, including: CHINT Life's "CHINT MOVE," Pacific Health's "Blue Medical · Blue Little Bao," CPIC Property & Casualty's "Jian Little Bao," PICC Property & Casualty's "AI Kangbao," Zijing Property & Casualty's "Zijing Little Fu," and CITIC Property & Casualty's "Tai Little Bao."
All the aforementioned agents are equipped with identity verification authorization in their functional modules to query valid policies under a user's name. They also come preloaded with health service toolkits such as "AI Clinic," "Medicine Box Recognition," "Report Interpretation," and "Skin Analysis," effectively extending low-frequency post-underwriting contracts into daily, high-frequency health management companionship.
On 10/10, Fosun Prudential Life Insurance and Ant Health announced a partnership to create a next-generation "Family Protection + AI Health" solution. This initiative combines Ant Health's AI-powered and human-guided health management capabilities with Fosun Prudential's expertise in insurance product design and actuarial risk control.
Specifically, Fosun Prudential Life will launch a dedicated AI agent on the Ant Afu App to integrate insurance with professional health management. After purchasing coverage, customers not only gain protection but also access an interactive health hub for consultations, explanations, and reminders—enabling dynamic assessment and ongoing family health support. This transforms their approach from "reactive response" to "proactive prevention."
"Insurance companies are increasingly joining AI open platforms to build agents—a pragmatic choice aligned with the industry's digital transformation," said Guo Jinlong, Director of the Insurance and Economic Development Research Center at the Chinese Academy of Social Sciences. From a technical perspective, developing in-house large models involves high computing costs and steep technical barriers; integrating with mature AI platforms enables rapid deployment of scenarios such as consultations, policy inquiries, and claims pre-screening, significantly shortening development cycles. From an operational standpoint, this approach represents a low-capital pilot that requires minimal upfront investment, allowing for cost-effective experimentation and enhanced customer interaction experiences. Additionally, the public traffic on AI platforms serves as a new customer acquisition channel, helping insurers overcome traffic limitations within their own apps and enabling precise user operations to drive intelligent upgrades in insurance services.
"Users are increasingly accustomed to using AI platforms for conversational queries to assist with purchasing decisions. By joining these platforms, insurers can intervene at the earliest stage of decision-making to identify needs and guide users, securing the first touchpoint in the consumer journey. Additionally, insurers can leverage open platforms to access real-world conversation data and user feedback, gaining valuable insights into customer needs that inform product design and marketing messaging. The platform also enables low-cost testing of user intent and conversion paths," said Si Erxun, Industry Research Manager at IDC's Financial Services team.
In the view of Zhu Junsheng, a professor and postdoctoral researcher in applied economics at Peking University, the AI era will bring profound transformation to the insurance sales industry.
1. Customer acquisition channels are shifting from users actively searching for insurance products to AI-driven needs description and coverage recommendations.
Second, the sales model has shifted from product promotion to demand identification, solution matching, and professional explanation.
Third, deepen human-AI collaboration: AI handles standardized inquiries and preliminary analysis, while professionals focus on complex planning, underwriting explanations, claims disputes, and high-trust services.
Fourth, competition is escalating beyond channels and products to include professional service capabilities, data governance, and customer trust.
"AI is unlikely to fully replace insurance sales professionals, but it may transform their roles, customer acquisition methods, and value creation models," said Zhu Junsheng.
Trust and compliance are non-negotiable.
Notably, as a highly regulated industry, insurers remain hesitant to integrate with AI open platforms due to risks such as "AI hallucinations" (where AI-generated content diverges from actual data or user instructions) and compliance red lines.
On 9 15, results from the national crowdsourced security testing of large AI models were released during the National Cybersecurity Awareness Week. The initiative mobilized 2467 white-hat hackers to conduct vulnerability assessments on 54 large model and agent application products from 25 domestic AI vendors. A total of 873 security vulnerabilities were identified, including 608 unique to large models and agents, and 265 traditional vulnerabilities.
In 4 this year, Professor Xu Xian's research team at the School of Economics, Fudan University, released the "Insurance 'Lobster' Skill Insights Report" based on 539 public skill samples. A key finding of the report is concentrated security risks: a total of 1937 risk items were identified, with data privacy leaks (569), compliance and regulation (358), and external dependencies (281) being the core concerns.
Tian Hongzhen, Party Secretary of Fosun Prudential Life Insurance, noted that while AI can produce hallucinations, the financial and insurance sector demands high information accuracy. In high-risk scenarios such as underwriting and claims processing, AI applications must be used cautiously, with key decisions still requiring human confirmation.
Separately, the company received regulatory inquiries after onboarding to relevant AI platforms. Regulators are focusing on the types of large models used by the platform's agents and the extent to which user privacy data is utilized.
"Insurance is a highly regulated industry. Financial agents involved in product recommendations and policy interpretation must strictly comply with regulatory requirements on suitability of sales, disclosure of information, and prohibitions against misleading statements," said Si Erxun.
"The general-purpose public large model platform has inherent limitations in adapting to the insurance industry's stringent regulatory requirements and high privacy protection standards," Guo Jinlong also told reporters. In the absence of well-defined AI compliance regulations for the industry, insurers face compliance risks such as data breaches and misleading scripts when joining AI open platforms.
However, the regulatory policy framework is gradually being refined. On 2026 6, 18, the National Financial Regulatory Administration issued the "Guidance on Safe Development and Application of Artificial Intelligence in Banking and Insurance," requiring financial institutions to assume primary responsibility for AI applications, improve human-machine collaboration management processes, and implement classified and graded management of application scenarios. The document also specifies that generative AI applications involving underwriting and claims handling, risk management, or those directly affecting customer interests and the conclusion of financial contracts are classified as high-risk scenarios requiring focused oversight.
In 2026, the Shanghai Financial Regulatory Bureau issued "Several Measures to Promote AI Applications in Shanghai's Banking and Insurance Sectors," requiring that generative AI insurance applications for public use must be reported to regulators prior to launch. This requirement applies regardless of whether models are self-developed, API-based, or privately fine-tuned, and covers both in-house and purchased models.
"In the long run, regulation will extend from traditional oversight of products and personnel to penetrative oversight of algorithms, models, and AI-generated scripts. Model filing and algorithmic audits will become industry standards," said Guo Jinlong.
"Insurance is a highly regulated, trust-driven financial sector. AI adoption will be gradual and likely evolve into a 'core self-operated, peripheral open, human-AI collaborative' model," said Si Erxun. In the future, insurance marketing will exhibit four characteristics: intelligent front-end, professional mid-end, automated back-end, and full-process regulatory oversight. The winners won't be institutions that simply possess AI, but those that effectively integrate AI, expert advisors, compliant data, and consumer trust. Insurance sales will shift from "channel-driven" to "trust + service + data-driven." AI acts as an accelerator, but trust and compliance remain the bottom line.
