HDC2025丨华为发布鸿蒙智能体框架，全面迈入Agent时代

Core Insights - Huawei's Developer Conference 2025 introduced the Harmony Agent Framework (HMAF), marking a significant step towards integrating applications and intelligent agents within the Harmony ecosystem [1][3] - The framework aims to establish a new interaction paradigm between the operating system, Harmony applications, and intelligent agents, focusing on autonomous decision-making and collaborative capabilities [1][3] Group 1: Harmony Agent Framework - The Harmony Agent Framework is designed to support the intelligent upgrade of applications across various sectors, providing developers with a robust architecture for intelligent agent development [3] - Key components of the framework include upgraded interaction protocols, efficient development processes, and enhanced security measures, facilitating deep integration and collaboration among systems, applications, and agents [3][4] Group 2: Xiaoyi Intelligent Agent - Xiaoyi serves as the central intelligent agent within the Harmony system, enabling multi-modal interaction through voice dialogue, navigation, and suggestions, thereby enhancing user experience [4][11] - The Xiaoyi Intelligent Agent Open Platform offers a comprehensive development solution for developers, featuring various development modes such as LLM and workflow, tailored to different business logics [5][7] Group 3: Developer Engagement and Ecosystem - Over 50 pioneering Harmony intelligent agents are in development, with applications like Huawei Weather and various third-party services set to leverage intelligent agent capabilities for enhanced user interaction [8][10] - The platform supports over 4000 applications with 11 types of AI widgets, integrating AI capabilities into the Harmony ecosystem, significantly improving the intelligence of applications [11] Group 4: Future Prospects - The integration of the Harmony Agent Framework and Xiaoyi Intelligent Agent Open Platform is expected to create a new paradigm for human-machine interaction, enhancing consumer experiences across various scenarios [13]