Alibaba Cloud X Earth Property & Casualty Insurance: The country's first intelligent car insurance claim reporting system goes live

Going forward, in Anbang Property & Casualty Insurance’s claims reporting hotline, with just one phone call, you can complete a complex auto insurance claims reporting process.

For example, when a user calls: “Hi, I’ve rear-ended these two vehicles. I want to file a claim; I want to make a report of coverage,” the intelligent attendant will, by combining the voice link with large-model intent understanding, accurately collect and respond—within 2–4 minutes—with more than a dozen key fields such as license plates, accident time, location, and liability attribution.

This is Alibaba Cloud working jointly with Anbang Property & Casualty Insurance, based on Tongyi Xiaomi, to build the first domestic “intelligent auto insurance claims intake attendant.” Users can complete highly complex claims reporting through natural conversation. As of the end of February, Anbang’s auto insurance “intelligent claims intake attendant” business had a near 91% completion rate, and more than 43% of cases had already been handled by this intelligent robot.

Auto insurance claims intake and reporting scenarios require accurately collecting more than a dozen key fields within 2–4 minutes—such as vehicle license plate numbers, accident time, location, liability attribution, whether there are personal injuries or property damage—while also involving a large amount of unstructured spoken language (e.g., dialects, vague descriptions, emotionally charged statements, etc.). This places extremely high demands on the robustness of speech recognition, intent understanding, and dialogue logic, and is regarded in the industry as the “deep-water zone” of intelligent customer service.

Therefore, Anbang Property & Casualty Insurance, together with Alibaba Cloud, focuses on the business scenario of auto insurance claims reporting. Based on Tongyi Xiaomi’s full-duplex real-time voice interaction technology and an effects evaluation system, it has built a low-latency, highly human-like dialogue agent, creating the first domestic “intelligent auto insurance claims intake attendant.”

Tongyi Xiaomi is a low-latency, highly human-like dialogue agent built on the Qianwen large model. During the implementation of this auto insurance claims reporting project, the project team followed the principles of “usable, controllable, and measurable.” For the insurance claims reporting scenario, it customized a multi-turn dialogue engine and risk-control validation mechanisms. While ensuring millisecond-level responsiveness, it has significantly improved the accuracy of complex intent recognition and the human-likeness of interactions.

A relevant spokesperson from Anbang Property & Casualty Insurance said: “Auto insurance claims reporting is the first touchpoint of claims services and also a key part of the customer experience. We hope to use AI to address highly repetitive processes where rules are clear but labor is intensive, so that human attendants can focus more on complex cases and emotionally oriented services.”

An Alibaba Cloud spokesperson for the new financial industry said: “A new generation of agent architecture systems represented by general-purpose agents is reshaping applications in the financial industry. Alibaba Cloud’s general-purpose financial agent framework, formed from industry learnings, has evolved from the early ‘think-act’ pattern into a complete architecture with capabilities for deep reasoning, self-correction, multimodal interaction, and long-range planning. It has already achieved production-level deployment in Anbang’s core claims scenarios. The successful validation of its business value provides a reference for launching more scenarios with general-purpose agents in finance.”

In the future, Alibaba Cloud and Anbang Property & Casualty Insurance will continue to deepen their cooperation, carry out practical exploration around the main line of “Artificial Intelligence +,” and focus on enabling and improving efficiency and effectiveness in underwriting, claims, marketing, and customer service.