Tencent Holdings Ltd. (TCEHY) has started testing Xiaowei, a native AI agent inside WeChat, its platform used by 1.4 billion people, to help users complete everyday tasks across an ecosystem of millions of apps. The early version, which began testing with select users in late June, allows typing or speaking prompts, searching online, summarizing chats and Moments, drafting replies, and triggering mini programs for tasks like booking flights, taxis, or meals. Tencent developed its own WeLM large language model and also uses DeepSeek's latest model for parts of the product, though the current experience appears limited, with some transactions taking minutes and often requiring user intervention. JPMorgan's Alex Yao stated that the debate over the next six to 12 months is whether WeChat's June beta testing gives investors enough evidence to stop assigning zero value to Tencent's AI roadmap, and he believes it does. If Tencent improves Xiaowei's reliability, investors may view the tool as a way to deepen user engagement and defend its platform against Alibaba Group Holding Ltd. (BABA) and ByteDance Ltd.