The agentic smartphone emerged as one of the most compelling and contested narratives at the 2026 World Artificial Intelligence Conference (WAIC) in Shanghai. Several Chinese technology companies unveiled devices designed to serve as personal AI assistants, capable of executing complex tasks across multiple applications with simple voice commands, without the user needing to manually navigate individual apps.
Yet even as the technology was being demonstrated on the exhibition floor, the commercial reality was already proving more complicated: major app platforms are actively restricting the access that these AI agents require to function as advertised.
The Vision: A Phone That Acts, Not Just Responds
The appeal of the agentic smartphone is straightforward. Rather than requiring users to open Alipay to make a payment, then switch to DiDi to book a ride, then open a messaging app to confirm an appointment, the AI agent handles the entire sequence autonomously in response to a single natural-language instruction. This paradigm shift promises a fundamentally more intuitive and efficient user experience, reducing the cognitive overhead of managing a complex digital life.
StepFun introduced the STEPX Neo, which it describes as an “AI agent-native smartphone” built around its Step AOS operating system — the company’s “world-first” OS designed natively for AI agents. According to a Xinhua-sponsored report, StepFun has established deep partnerships with Alipay and DiDi, enabling the STEPX Neo to complete travel bookings, shopping, office tasks, and video editing through a single AI-powered interface.
Nubia presented the NaviX Ultra, powered by ByteDance’s Doubao chatbot, following the sold-out success of an earlier experimental “Doubao Phone” in December 2025. Honor also showcased its Robot Phone concept, featuring a movable camera mounted on a compact robotic arm that can recognize gestures and respond to music. As Firstpost reported, these devices represent a significant leap in mobile computing ambition.
The Platform Pushback
The commercial challenges facing agentic smartphones became apparent almost immediately after the first experimental devices were released. Shortly after the Doubao Phone’s limited launch, Alibaba, Tencent, and JD.com restricted the built-in AI assistant’s access to their platforms, effectively disabling many of the device’s key capabilities, particularly those involving financial transactions. As a result, ByteDance was forced to disable certain advanced functions.
The resistance from major platforms is commercially rational. Allowing a third-party AI agent to mediate the user experience threatens to disintermediate app developers, reducing their direct engagement with users and potentially impacting advertising revenue, data collection, and the commercial relationships that underpin their business models.
Kiranjeet Kaur, Research Director at IDC, noted that major digital platforms are reluctant to grant third-party AI systems unrestricted access because doing so would weaken their relationships with users. This dynamic is not unique to China, it mirrors the tensions that have emerged between AI assistant developers and platform operators globally. We covered the early signs of this conflict in our analysis of China’s AI agent phone market surge.
The Road Ahead for Agentic Mobile Computing
Despite the current friction, industry analysts remain broadly optimistic about the long-term trajectory of agentic smartphones. Marc Einstein of Counterpoint Research predicted that within the next five to ten years, smartphones will rely far less on traditional applications, with AI agents increasingly acting as the primary interface between users and digital services. The question is not whether this transition will happen, but how quickly the ecosystem conflicts can be resolved.
The path forward likely involves a combination of standardized APIs, negotiated access agreements, and potentially regulatory intervention to ensure that AI agents can function effectively across the digital economy. The companies that succeed in building agentic smartphones will be those that can navigate the complex ecosystem dynamics, building trust with platform operators, ensuring robust security protocols, and delivering a user experience that genuinely justifies the shift away from the familiar app-based interface.
AI Wearables and the Broader Device Ecosystem
The push toward AI-native devices at WAIC extended beyond smartphones. Alibaba demonstrated its Qwen S1 AI glasses, integrating AI assistance with navigation, payments, and first-person imaging. iFlytek showcased lightweight AI glasses focused on translation, meetings, and workplace productivity. Rokid and Mobvoi also highlighted their own AI operating systems for wearable devices, positioning these products as complementary interfaces for users who want AI assistance without the friction of interacting with a smartphone screen.
The broader pattern is clear: China’s technology industry is engaged in a comprehensive effort to make AI agents the primary interface between users and the digital economy, across every form factor from smartphones to glasses to earbuds. The competition is no longer centered solely on hardware specifications but on which ecosystem can deliver the most capable, most trusted, and most seamlessly integrated AI assistant.
Japan’s Brain Technologies launched its Natural AI Phone in partnership with SoftBank earlier this year, and Google is rapidly expanding AI capabilities across its mobile devices, evidence that the agentic device race is global, even if China is currently setting the pace. The outcome of this competition will have profound implications for the future of the digital economy and for the companies that currently dominate the app-based model of mobile interaction.