On September 26, Didi Chuxing announced the public beta test of its AI travel assistant "Xiaodi Beta v0.8 version." This new feature intelligently understands user needs and provides personalized transportation solutions. Users just need to update the Didi App to the latest version, search for "AI ride" in the destination bar, and enter a code to experience this service. Xiaodi supports voice and text input, with simple interaction operations that can organize up to three matching vehicle options based on the user's detailed needs, combined with real-time information such as time and traffic conditions for the user to choose from.
The launch of Xiaodi has changed traditional ride-hailing methods. It first "understands" user needs and then finds more accurate and tailored solutions based on Didi's rich vehicle and driver information. During the beta test period, Didi encourages users to express their needs as detailed as possible so that Xiaodi can continuously learn and improve. In addition to personalized ride-hailing, Xiaodi can also provide travel solutions for more specific and diverse scenarios, such as checking the weather, planning departure times, and recommending vehicles that are convenient for loading luggage.
In addition, to promote innovation in large model applications and support the construction of an AI developer ecosystem, Didi has also launched the MCP service simultaneously. After simple configuration, AI developers can quickly access Didi's MCP, allowing their custom intelligent agents to independently plan travel itineraries, hail rides, check orders in real time, and make automatic payments, easily customizing a dedicated smart travel assistant. Developers can also combine Didi's MCP with other domain-specific MCP services to explore more application scenarios. Currently, Didi's MCP service supports calling express, economy, regular, comfort, private, and luxury cars. Didi will continue to explore the application possibilities of MCP technology in more fields under the premise of ensuring safety and trust, and work with developers to build an open AI ecosystem together.