Alibaba Qianwen is about to launch Personal Agent “Qianwen Go”, which upgrades the scope of services based on the Agent architecture. HardAI has exclusive information that the first scenarios for Qianwen Go include flight delay claims, e-commerce price protection, car maintenance inquiry and reservation, sports venue booking, as well as calculating the actual price of home appliances by combining subsidies, old machine recycling, and installation costs.
In the case, the user only asked to “help me communicate with customer service on the app regarding compensation,” without providing any flight information. Qianwen Go logged into the cloud-based booking platform and checked each order from four airlines individually. Based on customer service feedback, it was determined that East Air MU2445 was the delayed flight. After the external call was blocked, the user returned to the order page to find the manual customer service entry and switched to conversation communication. When the user refused a voucher, the agent wrote “only accepting cash compensation.” After multiple rounds of negotiation, 70 yuan in cash compensation was obtained and transferred to the Alipay account. In the Taobao price protection scenario, the Agent found the price protection entry by itself, traversed all orders, and identified a reduction in price for the “Sanhé 0-coated real titanium frying pan non-stick” product, resulting in a refund of 52 yuan for the price difference.
Qianwen Go has extended its short-link operations to tasks without a fixed process. It is necessary to switch entry points, platforms, and communication methods based on the execution results, and the completion rate depends on this switching ability. The cloud environment provides the execution conditions, and Meta’s Muse has an independent cloud computer and browser, allowing it to continue working even after the user closes the App.
The institution has re-evaluated the commercial and computing power increments of individual Agents. A research report by Bank of America on October 7 suggests that Agents are expected to expand overall online activities. Morgan Stanley estimates that the market for consumer AI targeting individuals in China could reach approximately 300 billion yuan in 2030, and potentially 1.6 trillion yuan by 2040. As tasks are carried out, computing power demands increase. Citibank raised its CPU market forecast on October 6, expecting the global scale to rise from 29 billion US dollars in 2025 to 300 billion US dollars in 2030. Alibaba, which owns both Agent entry points and trading platforms, may also enhance conversions.