China's "life insurance leader" achieves major breakthrough, starts using "AI Agent"

```

Amid the global financial industry's accelerated embrace of artificial intelligence, a new trend is emerging.

AI is no longer just a backend tool. It is now appearing as an “Agent” directly involved in customer service, decision support, and even business execution.

Banks are trying, brokerages are trying, and now it’s the insurance companies’ turn.

For a long time, buying a life insurance product—when it comes to checking policy details or asking about surrendering the policy—usually meant waiting for a manual reply, a slow and lengthy process.

Especially for questions like “How much have I paid?” and “How much would I get back if I surrender now?”—these seem simple, but involve personal data and rule-based calculations. Customer service doesn’t dare reply casually, while customers are eager to know.

This gap between “long-term commitment” and “instant service” has always been a challenge for the life insurance industry.

Zishitang found that China’s largest life insurance company—China Life—is using its overseas platform to experiment with a new type of AI (called AI Agent) to address these frequent and sensitive questions.

China Life Takes the Initiative

Zishitang learned that China Life (Overseas) has recently partnered with NetEase’s Cloud Business unit, deploying AI Agent applications in the after-sales service process.

According to official disclosures: the system’s answer accuracy rate in “policy maturity and surrender queries” exceeds 90%.

For example, when clients ask “How much premium have I paid?” or “How much can I get if I surrender?”, previously it required a manual login to check and reply, taking time. Now, AI Agent can automatically retrieve data, calculate results, and generate compliant responses, significantly shortening response time.

What are the pain points of the life insurance giants?

For customers who buy life insurance products, “after-sales service” means everything they need to interact with the insurance company after buying the insurance, such as checking the policy, asking about payments, changing details, or finding out “how much will I get when the policy matures” or “what is the loss if I surrender now.”

These questions seem simple, but often require customer service to manually check the system one by one, with longer waiting times during peak periods.

Because each policy has different payment records and additional benefits, any oversight during manual searches can lead to inconsistent replies, leaving customers confused.

For the insurance company, such queries hide complex “computational work” behind them.

For example: a life insurance policy may have been paid for 5, 10, or even 20 years. Has there been any payment interruption? Are there bonus or universal accounts attached? How to calculate cash value on surrender?

These can’t rely on memory or estimates. Multiple systems must be logged into, historical records retrieved, and calculations made step by step following actuarial rules.

Previously, it was all manual, which not only took a long time but also easily led to mistakes from reading the wrong line or entering wrong numbers, resulting in errors. If the customer gets an incorrect amount, it could lead to complaints or even disputes.

What is an AI Agent?

The Chinese term for AI Agent is “artificial intelligence agent,” which is different from traditional chatbots.

Ordinary AI can only answer questions passively, while AI Agent (artificial intelligence agent) can actively understand goals, plan steps, use tools, and complete tasks.

For example, when a customer asks “I want to know about surrender,” the Agent will automatically identify the intention, retrieve policy information, calculate cash value, and explain the process, all without human intervention.

This is not easy in the financial context.

Because insurance, banks, and other services involve lots of personal sensitive data and strict compliance, AI cannot “perform freely,” and every step must operate within preset rules. For example, it cannot disclose someone else’s policy information at will, nor make decisions for the client—it can only provide factual calculations and guidance.

Because of this, AI Agents in the financial field are more like “professionally trained digital employees”: they know what can be done, what needs to be reported, and what must be judged by humans.

Currently, in the financial industry’s practice cases, when sensitive operations or complex judgments are involved, the system will still transfer to real customer service staff, ensuring safety and experience.

Not the First Attempt

In fact, this is not the first time China Life (Overseas) has publicly disclosed its AI transformation initiatives.

Zishitang found that in early February this year, this insurance giant launched two AI applications in Hong Kong and Macau:

First is a 7×24-hour AI assistant for customers, available in the OneService APP, handling common inquiries about policy status, value, maturity documents, renewal methods, etc.

Second is an AI knowledge search tool for financial advisors, to quickly acquire product and policy information.

Both tools are developed based on the DeepSeek V3 large model, emphasizing natural language understanding and accommodating the regulatory requirements for Hong Kong and Macau.

The principle is: simple questions get instant AI responses, complex questions are handed over to humans during office hours, achieving human-machine collaboration.

How Big is the “Policy Capacity” Behind This?

Public data shows: In 2025, China Life Overseas performed well, with profits reaching new highs.

Specifically, last year total premium income was HKD 49 billion, new value-type business premium (converted) was HKD 10.5 billion, up 63% year-on-year.

Total investment return for the year was HKD 22.15 billion, with a comprehensive annualized investment yield of about 6%, a year-on-year increase of 0.91 percentage points.

From this scale of operations, with more policies comes more client queries—relying solely on manual replies is slow, costly, and prone to mistakes.

Naturally, this “leading life insurer” has launched countermeasures.

From another angle, steady operational performance provides space for exploring technological applications.

Risk disclosure and disclaimerThe market comes with risks, and investment requires caution. This article does not constitute individual investment advice, nor does it take into account the specific investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article suit their specific circumstances. Investments based on this article are made at the user's own risk. ```