WeChat's Internal Test Introduces Built-in AI: Right Swipe to Switch Mini Programs and Order Coffee, But Insufficient Chip Inventory Becomes a Constraint on Computing Power

According to TechInsight Beating Monitor, Tencent is testing an embedded AI chatbot prototype in WeChat, with plans to launch regulatory compliance reporting as early as this month. In the early demo, users can swipe right on the WeChat main interface to enter the chatbot's chat box. After entering a command, the chatbot can automatically call on a large number of mini-programs within the WeChat platform to complete tasks such as "finding a coffee shop and placing an order based on taste and price preferences."

In the large-scale model field, Tencent's progress is lagging behind domestic competitors. Alibaba has integrated e-commerce, travel, and map services into the Tongyi Qianwen App, while ByteDance has added functions such as smart shopping to the Douyin App. Although the active user base of Douyin and Tongyi Qianwen is still far below WeChat's 1.4 billion, the rapid growth of the two apps has forced Tencent to accelerate the release pace of the WeChat chatbot. Tencent President Liu Chiping stated at last month's earnings conference that in addition to basic large models, chatbot AI is a breakthrough application scenario, and the WeChat platform has a natural advantage in hosting AI chatbots.

While the WeChat chatbot prototype can successfully complete tasks, mass promotion still faces a computational bottleneck. Prior to the NVIDIA chip export ban, Tencent was overly conservative in hoarding chips, and currently, China's domestic semiconductor supply remains relatively tight. Tencent's internal preliminary assessment shows that the cost of fully promoting the WeChat chatbot is extremely high, and it is difficult to generate enough revenue from related businesses in the short term to cover the computational and operational expenses.