Nan Fang Du Shi Bao·2026-01-15 05:24Core Insights - Alibaba's C-end AI application "Qianwen" has launched a life assistant feature that allows users to order food, book flights, and hotels, and find restaurants through simple voice commands, currently integrated only within Alibaba's ecosystem and not with external platforms like Meituan [1][4]. Group 1: Functionality and User Experience - The "Qianwen" app can automatically call Taobao's flash purchase service to recommend nearby coffee shops based on user location when given a command to order coffee [1]. - For hotel and flight bookings, "Qianwen" integrates with Alibaba's Fliggy, providing users with multiple options based on price and comfort, allowing them to select suitable options through card links [4]. - The app has seen a 300% month-on-month increase in user inquiries for product recommendations since its launch two months ago, indicating a growing reliance on AI for shopping decisions [10]. Group 2: Technical Implementation - "Qianwen" utilizes an API-based approach for its AI capabilities, allowing for a smooth user experience by recognizing and analyzing user commands, retrieving inventory from partner platforms, and generating pre-orders for user confirmation [14]. - The AI agent only executes sensitive actions up to the "pending confirmation" stage, requiring users to review and confirm order details before payment, ensuring user oversight in transactions [14]. Group 3: Competitive Landscape - The mobile AI agent is seen as a crucial future traffic entry point, with competition intensifying among major model vendors and mobile manufacturers [14]. - ByteDance's Doubao mobile assistant, which uses screen reading and simulated operations, has gained significant attention but faces challenges in execution smoothness and privacy concerns, leading to temporary restrictions on its capabilities [15]. - The API interface approach has its own challenges, including limited availability of third-party app APIs and the need for extensive negotiations to access these interfaces, making it difficult for mobile AI agents to meet diverse user needs [15].