Honor’s next generation terminal operating system! AgenticOS released, the industry’s first system-level Agent architecture

Kuai Technology reported on July 19 that recently, at the 2026 World Artificial Intelligence Conference, Honor officially released AgenticOS, a new generation of partner-based multi-modal intelligent operating system created by Honor. At the 2026 Shanghai Mobile World Congress held in June this year, Fang Fei, President of Honor Product Line, announced that Honor will launch AgenticOS, the next-generation human-centered terminal operating system, and plans to officially release it in July. According to the product plan announced by Honor, the Honor Robot Phone will be equipped with the AgenticOS kernel for the first time, and the Honor Magic9 series released in the fourth quarter of this year will be equipped with an early adopter version of AgenticOS. AgenticOS adopts the industry's first system-level Agent architecture to support global multi-agent collaboration, full-scenario adaptive precipitation, and self-circulation capabilities for all-scenario agents. It is understood that AgenticOS has four core features: intent-driven, natural interaction, active intelligence, and native cross-device. Among them, intention-driven means that the system will shift from "application-centered" to "user-intent-centered". Users do not need to repeatedly switch between multiple apps, and the system can independently understand and complete tasks according to needs. In terms of natural interaction, AgenticOS uses human senses as the entrance. Voice, gestures, eyes and movements can all be used as interaction methods, making the communication between users and devices more natural. In terms of active intelligence, the system takes Agent as the core and has active planning, active service and active execution capabilities. It can dismantle tasks and complete operations according to user needs. Tiansheng Cross-Terminal uses a unified intelligent brain to schedule multiple devices to achieve cross-device collaboration and multi-Agent collaboration.