Abao connects phones and cars, unlocking 10,000+ services

At an airport exit, travelers may no longer need to carry their bags while opening mini-programs and hailing a car step by step. Li Jun, president of Alipay’s business group, described the use case this way: tell Abao, “Help me hail a car to the office, and while you’re at it, order a cup of coffee from the Starbucks downstairs,” then add another request the next day to have it send the electronic invoice for the previous day’s trip to an email address. Behind the scenes, the task is not completed by a chatbot acting alone, but by multiple agents working in sequence. In Hangzhou, Alipay released the AHA (Agent Hub Access) multi-agent cross-device interoperability protocol framework, bringing this kind of collaboration to devices such as phones and cars.

Abao has now completed intelligent adaptations for more than 10,000 services across eight major scenarios, including transportation, food and dining, culture and tourism, and government and public services. It has established cross-device collaboration with five major phone makers, whose combined market share exceeds 70%; on the automotive side, it covers 16 major automaker brands and has reached cooperation intentions with more than 60 automakers. At the event, Chinese tech media outlet ifanr saw users hail cars from their phones and order KFC by voice from inside a car. Alipay also said that connected intelligent services will be made available through Qwen, with more than 50 services included in the first batch.

AHA is not intended to make models better at chatting, but to enable services to find one another and hand off tasks without crossing permission boundaries. In the example involving OPPO Xiaobu and Abao, Xiaobu interprets the user’s intent and then passes the task to Abao; Abao completes service fulfillment and payment on its own side. In this example, user privacy and preferences remain with Xiaobu, while service fulfillment and payment are handled by Abao, separating the responsibilities and authorities of the two sides. Alipay said the protocol uses domain authorization and data isolation to address security issues in collaboration across manufacturers.

For merchants, Alipay has also introduced a full-stack commercial foundation for agents. A restaurant without an AI team can hand its existing pages, products and service processes over to Alipay’s AI Open Platform for hosting, converting them into standard capabilities such as Skills that agents can call. Merchants with development capabilities can create their own agents to handle recommendations and membership operations. Once adapted by a merchant, a service can potentially reach phones, vehicle infotainment systems, AI glasses, watches and large-model applications. Since Alipay’s AI Open Platform went live in July, some merchants’ development requests have already been scheduled through next year.

In practical terms, users get cross-service workflows that require opening fewer apps and making fewer confirmations: from a phone or car, they can state what they need, and the system can assign transportation, dining, ticket purchases and other steps to different specialized agents; merchants, meanwhile, have an opportunity to bring the same service to more types of devices. Alipay CEO Han Xinyi expects the next 6–12 months to be a critical window and said agent-based commerce could reach a genuine breakout point in the second half of next year, though this remains a forecast rather than an outcome that has already occurred. Payment security, the scope of permissions, responsibility when something goes wrong and how revenue is divided among the parties will determine whether this network can continue to operate.