--- title: "AI Smartphones Move Beyond Feature Wars as Honor Begins to Reconstruct Its Operating System" type: "News" locale: "en" url: "https://longbridge.com/en/news/293107724.md" description: "On July 18, Honor unveiled its Agentic OS technical framework at the 2026 World Artificial Intelligence Conference, aiming to reconstruct an operating system centered on \"intent\" and \"tasks.\" This framework shifts interaction logic from an app-centric model to one driven by user intent, with the system automatically breaking down and executing multi-step tasks. Meanwhile, Honor announced a partnership with Alibaba's Qwen to develop on-device large language models, emphasizing that future competition will hinge on integrating in-house and external models with hardware capabilities, rather than merely stacking features or relying solely on proprietary development" datetime: "2026-07-19T00:33:35.000Z" locales: - [zh-CN](https://longbridge.com/zh-CN/news/293107724.md) - [en](https://longbridge.com/en/news/293107724.md) - [zh-HK](https://longbridge.com/zh-HK/news/293107724.md) --- # AI Smartphones Move Beyond Feature Wars as Honor Begins to Reconstruct Its Operating System Competition in the AI smartphone sector is moving beyond the stage of simply stacking features. On July 18, during the 2026 World Artificial Intelligence Conference, Honor released its Agentic OS technical framework. According to Honor’s vision, the next generation of mobile operating systems will shift from being app-centric to focusing on user “intent” and “tasks.” Over the past two years, smartphone manufacturers have primarily competed on features such as AI object removal, call summaries, and document summarization. Although large language models (LLMs) have been integrated into smartphones, the way users interact with their devices has not fundamentally changed: they still need to open specific apps and sequentially complete steps like searching, selecting, and confirming. Honor aims to further transform this long-standing interaction logic. Huang Fei, Chief AI Scientist at Honor, stated that the essence of Agentic OS is not merely “adding an AI assistant to the system,” but rather reconstructing a new type of operating system centered on “intent” and “tasks.” Users no longer need to decide which app to open first; they only need to express their final goal. The system is then responsible for understanding the intent, breaking down the task, planning the execution path, and invoking different apps and services to complete the action. For example, if a user requests to order a cake, hail a ride to a restaurant, and book a KTV room, the phone must recognize that the user is preparing for a birthday party, break this goal down into multiple steps, and continuously execute operations across different apps. As a result, apps gradually recede into the background, while user intent becomes the new entry point. The smartphone is transforming from an app container waiting for item-by-item user operations into an execution platform capable of organizing models, apps, and hardware to accomplish tasks. However, moving from “understanding a sentence” to truly “getting things done” involves more than just a change in interaction methods; it also requires sufficient model support behind the operating system. In addition to continuing the development of its proprietary large models tailored for on-device scenarios, Honor also announced a collaboration with Alibaba’s Qwen to jointly develop on-device LLM solutions deeply optimized for smartphone use cases. This partnership reflects the realistic choices facing smartphone manufacturers in the era of large models. It is difficult for a single device manufacturer to independently train and maintain all types of models, nor can it easily cover every task users might request. Future competition in AI smartphones may not depend on whether manufacturers fully develop their own large models, but rather on their ability to organize proprietary models, external models, app services, and hardware capabilities into a stable system. If Agentic OS addresses how to orchestrate models and apps in the digital world, then Honor’s Robot Phone, showcased at the event, further extends this capability into the physical world. At the WAIC venue, Honor demonstrated some capabilities of Agentic OS through the Robot Phone. Users can initiate cross-app tasks using natural language, with the system automatically handling task decomposition, path planning, and operation execution. When deviations occur during task execution, the agent can backtrack to previous steps and readjust the operational path. For risky operations involving payments, the system pauses automatic execution, returns control to the user, and protects sensitive information through measures such as screen recording blackout. These capabilities primarily verify whether mobile agents can continuously complete tasks like humans and autonomously correct errors when they occur. The Robot Phone builds on this foundation by adding a four-degree-of-freedom gimbal system. It can identify and follow user movements, as well as sway to the rhythm of music. From cross-app operations to controlling the gimbal, the Robot Phone serves more as a technological testbed: verifying whether smartphones can evolve from digital tools into intelligent terminals with perception and action capabilities once the operating system can uniformly orchestrate models, apps, sensors, and actuators. Honor’s judgment on the future of smartphones is built on this logic. Honor believes that while smartphones will not be completely replaced by new AI hardware such as smart glasses or earphones in the next five years, their positioning will gradually shift from being the sole hub of personal devices to becoming core nodes connecting various terminals and agents. As standalone features like AI object removal and summarization become standard configurations for smartphones, differentiation in AI phones is beginning to sink deeper into the operating system layer. What the final product form of AI smartphones will look like remains uncertain. However, moving from feature stacking to system reconstruction has become an important path for smartphone manufacturers seeking the next-generation interaction interface. Risk Warning and Disclaimer The market carries risks, and investment should be approached with caution. This article does not constitute personal investment advice, nor does it take into account the specific investment objectives, financial status, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article align with their specific circumstances. Investors assume full responsibility for any decisions made based on this content. ### Related Stocks - [BABA.US](https://longbridge.com/en/quote/BABA.US.md) - [09988.HK](https://longbridge.com/en/quote/09988.HK.md) - [BABX.US](https://longbridge.com/en/quote/BABX.US.md) - [KBAB.US](https://longbridge.com/en/quote/KBAB.US.md) - [BABO.US](https://longbridge.com/en/quote/BABO.US.md) - [BABU.US](https://longbridge.com/en/quote/BABU.US.md) - [BABW.US](https://longbridge.com/en/quote/BABW.US.md) - [BBYY.US](https://longbridge.com/en/quote/BBYY.US.md) - [89988.HK](https://longbridge.com/en/quote/89988.HK.md) - [HBBD.SG](https://longbridge.com/en/quote/HBBD.SG.md) ## Related News & Research - [China considers tighter export controls on AI models and chips, FT reports](https://longbridge.com/en/news/293287358.md) - [Canva's cofounder explains why the company isn't forcing employees to use a specific AI tool](https://longbridge.com/en/news/292708451.md) - [10:30 ETLeadde AI Launches World's First Business Video AI Agent to Transform Corporate Communication](https://longbridge.com/en/news/293228605.md) - [13:30 ETONERugged Launches AI Rugged Windows Tablets Powered by Intel Lunar Lake Platform, Bringing Up to 115 TOPS for Edge AI Computing](https://longbridge.com/en/news/293371602.md) - [GoodVision AI to Build 100 MW AI Factory in Japan via Strategic Partnership with AI Storm | ALIS Stock News](https://longbridge.com/en/news/292929474.md)