The "AI Plus" initiative became a major topic of discussion during the 2026 National People's Congress sessions. The government work report explicitly called for deepening and expanding "AI Plus," promoting the accelerated adoption of next-generation intelligent terminals and AI agents, and driving the commercial and large-scale application of artificial intelligence in key industrial sectors.
The Minister of Industry and Information Technology, Li Lecheng, stated that for 2026, the ministry will vigorously promote a mutual advancement between AI and manufacturing, as required by the government work report. Efforts will focus on driving AI PCs, AI phones, and smart home devices to better meet public demand for an improved quality of life.
Consequently, intelligent terminal forms, represented by mobile phone AI assistants, are encountering new development opportunities. The GUI (Graphical User Interface) approach, with its strengths in versatility and low implementation barriers, appears to be the optimal solution for the current large-scale deployment of AI assistants. It has also become a significant pathway for implementing the congress's directives and promoting widespread AI accessibility, moving AI from theory to practical application.
The core demand for bringing AI technology to the masses is solving high-frequency pain points in daily life. Explorations of different technical routes aim to better achieve this goal. As a key vehicle for popularizing AI, the core value of mobile phone AI assistants lies in "inclusivity," allowing average users to enjoy the convenience of AI without needing to master complex operations. Currently, the technical implementation of AI phone assistants is primarily divided into two routes: GUI and API (Application Programming Interface), representing two sides of innovation in AI deployment.
The API route requires third-party applications to open standardized interfaces to enable deep integration between the AI assistant and the apps. However, progress has been slow due to constraints like protocol barriers, commercial negotiations, and partnership models. In the short term, this approach might even exacerbate the issue of "ecosystem isolation," a practical challenge in deployment where a multitude of small and medium-sized applications lack standardized interfaces, limiting their functional potential.
In contrast, the GUI approach allows the phone's AI to directly simulate user interactions with application interfaces, eliminating the need for pre-opened APIs. Theoretically, any interface that can be interacted with can be perceived and operated. This characteristic enables it to break through the closed nature of application ecosystems, allowing AI assistants to perform complex tasks across different apps, thereby addressing users' frequent pain points.
Since last year, assistants like the Doubao mobile assistant, developed by ByteDance, have utilized the GUI route. Through cooperation with phone manufacturers at the operating system level, they achieve system-level operational permissions. Coupled with the comprehension and reasoning capabilities of the Doubao large model, they can execute complex cross-application operations. Users can activate them via voice or side buttons at any time, enabling features like screen information extraction and multi-app coordination without interrupting the current task flow, even offering personalized services. This practice not only validates the feasibility of the GUI route but also provides a reference mobile solution for the deployment of various AI execution tools, bringing AI's problem-solving capabilities closer to everyday life.
During the congress, delegates and industry experts generally agreed that the development of next-generation AI technology should balance innovation with regulation, encouraging diverse technical explorations while maintaining safety standards. The GUI route aligns well with this direction, and its undervalued status in the market is changing.
An associate professor from the Department of Computer Science and Engineering at the Hong Kong University of Science and Technology emphasized that the value of the GUI route is underestimated by the market. "The API route sounds elegant, but its fundamental flaw is the need for the other party to open an interface. A vast number of small and medium-sized applications and long-tail services lack standardized APIs and have little incentive to create them. The GUI route, in theory, allows for perception and operation wherever an interactive interface exists, potentially alleviating the 'ecosystem isolation' problem more quickly."
A professor from the Artificial Intelligence Research Institute at Beihang University also pointed out that the API route is constrained by factors like protocol integration, commercial cooperation, and closed ecosystems, leading to a relatively slower implementation pace. In the short term, it still carries the possibility of worsening ecosystem isolation.
Further examining industry practices, the exploration by AI assistants like Doubao confirms the practical value of the GUI route. By not requiring third-party apps to open interfaces, they achieve cross-application cooperative operations, leading to more convenient and inclusive deployment and providing a viable model for overcoming implementation challenges for AI assistants.
Within the global AI competitive landscape, international tech giants like OpenAI and Microsoft have also begun investing in GUI-related technologies. Several domestic phone manufacturers are advancing the implementation of the GUI route. It is evident that the value of the GUI route is being redefined, becoming a core force driving the large-scale deployment of AI assistants and solving industry challenges, aligning with the congress's call to "promote the commercial and large-scale application of artificial intelligence."
Legal experts have clarified that AI phone assistants utilizing the GUI route, or AI agents themselves, do not inherently constitute a privacy or security threat. The primary risks stem from how permissions are used and whether data processing complies with standards.
One expert indicated that the GUI route itself does not necessarily pose a privacy threat; risks mainly arise from non-standard permission usage and data handling. "Under strict authorization, data minimization, and controllable operations, associated risks can be entirely managed within reasonable bounds. These are not inherent risks of the technical route itself."
He also stated that within China's current legal framework, the viewpoint requiring AI assistants to obtain "dual authorization" from both third-party apps and users lacks legal logic and would hinder technological innovation and application普及. User authorization to the phone system for simulating clicks and proxy operations is sufficient;强行 imposing additional authorization requirements from third-party applications lacks legal basis and contradicts technological development trends.
Another expert believes that addressing challenges and seizing opportunities requires leveraging the advantages of the national system and the vast application scenarios of the large domestic market. Support should be given to exploring various technical routes like GUI, API, and A2A, fostering a technological innovation and product application ecosystem based on safety that allows for parallel exploration and complementary paths.
Experts generally agree that the implementation practices of the GUI technical route, exemplified by assistants like Doubao, align with the national industrial development direction that encourages innovation, safety, and controllability. Moving forward, diverse technical explorations should be supported within a compliant framework to promote the coordinated development of GUI and API.