According to two people familiar with the matter cited by the Financial Times on June 2, Tencent is testing a prototype of an AI agent embedded in WeChat, and plans to start the compliance processes required before public launch as early as this month. After that, it will first open testing to a small number of external users, followed by a phased full rollout. Since the time needed for compliance remains uncertain, the official release window is still unclear. According to one person who saw an early demo, users will be able to swipe right on the WeChat home screen to enter the agent’s chat interface, where they can use natural language commands to automatically invoke millions of mini-programs within WeChat to complete tasks such as finding cafés by taste and price preference, or ordering drinks. After the news was announced, Tencent’s share price rose about 4%; a Tencent spokesperson declined to comment.
The people familiar with the matter said Tencent has made the WeChat agent its highest strategic priority, but senior executives are highly cautious about the launch details, striving to get every step right, which could prolong the testing cycle and involve multiple rounds of revisions. At present, insufficient computing power remains a major constraint: early internal estimates suggest that fully deploying the agent across WeChat’s 1.4 billion monthly active users would be extremely costly, and it remains unclear whether commercial revenue generation can cover this investment in the short term. On the competitive front, Alibaba has integrated its e-commerce, travel and mapping services into its Tongyi Qianwen AI app, while ByteDance’s Doubao has also introduced agent features such as online shopping. Although neither has nearly as many monthly active users as WeChat, both are growing rapidly. WeChat already has a built-in chatbot with search capabilities called ‘Yuanbao’. If this agent feature is successfully rolled out, it will significantly expand its automated execution capabilities.