On September 22, CaoCao Inc. (“CaoCao” or the “Company”) announced a partnership with Qwen AI Glasses at the 2026 Apsara Conference. The two parties will jointly launch an AI ride-hailing service featuring hands-free, near-eye interaction. Following integrations with Doubao Mobile Assistant, HONOR YOYO and OPPO Breeno, CaoCao is further expanding its on-device AI hardware service ecosystem, extending its ride-hailing services to smart wearable devices.
The partnership creates a complete closed loop of “device sensing + model understanding + service fulfillment”: Qwen AI Glasses sense user needs and provide trip-related prompts, the Qwen foundation model breaks down and matches those needs, and CaoCao ultimately fulfills the mobility service in the physical world.
To use the ride-hailing service, users simply put on Qwen AI Glasses and state their destination. The system can then automatically identify the pickup and drop-off locations, select the best route and get an estimate. During the trip, users can modify their destination or confirm the vehicle type for pickup at any time using voice commands. If they need to communicate with the driver, they can also initiate voice contact directly through the glasses. Key information, including the pickup vehicle and trip progress, is displayed as visual cards in the user’s field of view, keeping users informed throughout the journey.
On the technical side, the partnership adopts a two-layer architecture that combines MCP and Skill. CaoCao’s MCP Server is registered as a plugin on the Qwen Open Platform, exposing its full suite of mobility service capabilities to the platform. Interaction logic is handled by a Qwen AI Glasses Skill, which can flexibly orchestrate service calls, voice output and on-screen presentation for different use cases. This two-layer design allows CaoCao to focus on updating its mobility service capabilities, while the Qwen Open Platform can independently optimize the interaction experience. Each side can iterate independently, delivering a smoother user experience.