Tencent WeBao's AI Claim Reporting Completes 100,000 Cases and Speeds Submission by Nearly 40%

Tencent WeBao, Tencent's official insurance agency platform, says that as of April 2026 its AI claim reporting tool, Anxinpei AI Smart Claim Reporting, had completed 100,000 claim reports, and that nearly half of daily claim users tried it. According to a WeBao representative, the time users need to submit claim materials improved by nearly 40%, and by about 55% for key hospitalization products. The representative said: "Hospitalization and critical illness products used to require 40 to 50, or even more than 100, documents, which made the operation complicated for users. Now AI automatically classifies them and extracts the information in an average of just 4 minutes, and the efficiency gain is especially clear for complex products." Claims had long suffered from heavy paperwork, tedious form filling, delayed review and a poor user experience, with receipts often numbering dozens or hundreds, and materials that were missing, blurry or in the wrong format being sent back. The service launched in January 2026 and now covers the 180 products currently on sale, including medical, accident and cancer insurance; users entering the claim page can choose whether to use AI assistance. It runs on WeBao's self-developed model and Tencent's multimodal vision model, in three stages. AI material classification lets users upload 20 medical images in one batch, identifies invoices, medical records, expense lists and discharge summaries, and checks clarity, completeness and creases. AI pre-review guides users to fix missing, wrong, blurry or cropped materials in real time, with medical document recognition accuracy above 98%. AI automated form entry fills key personal, policy and medical fields within 5 seconds, and the user confirms with one click. The representative said all models used are self-developed or deployed locally. Simple cases such as outpatient claims are handled entirely by AI, while disability, large-amount, rare-disease and disputed cases still need deep human involvement.