以下文章来源于AGI接口 ，作者宋思杭 "AI持续落地，每个企业正在成为AI公司，每个人也将成为AI加持的'超级个体'。"汤道生开门见 山地表示。而在加大AI投入力度方面，腾讯的做法可以说是相当激进，其打通了各业务条线与AI 之间的壁垒，极大权限地对各业务开放了AI能力。 而"全面拥抱AI"，既是本场峰会的主题，同时也贯穿了腾讯2025年的全年战略重要。它正在影响 着这家科技大厂的方方面面。在演讲中，汤道生还提出了大模型、智能体、知识库和基础设施"四 个加速"的战略。其中，智能体则是极为重要的一种呈现形式。它所代表的是当大模型在国内走过 三年时间，应用范式的迭代。 然而，究竟什么是智能体？这个定义在行业里似乎还不够清晰。 在大部分人的眼中，智能体是这样的——当你想打开软件订机票，或者查找、整理信息，统计一 个excel表，不再需要打开不同软件app自己完成，而是告诉智能助手，它就能够自动帮你完成，最 终帮你订好机票。就好像在媒体群访间，腾讯云副总裁吴运声对媒体打趣道，"我现在就很希望有 一个智能体来替我回答你们的问题。" 从这个层面来理解，智能体的概念并不模糊。然而，目前之所以概念模糊，正是因为行业内存在 诸 ...

Core Viewpoint - Tencent is aggressively embracing AI across its business lines, aiming to integrate AI capabilities into various applications and services, which is a central theme of its 2025 strategy [1][2][4]. Group 1: AI Strategy and Implementation - Tencent's strategy includes a focus on "four accelerations": large models, intelligent agents, knowledge bases, and infrastructure [1]. - The company emphasizes that every enterprise is becoming an AI company, and individuals will become "super individuals" enhanced by AI [1]. - The intelligent agent concept is highlighted as a key area of development, representing a shift in application paradigms over the past three years [1][2]. Group 2: Understanding Intelligent Agents - The definition of intelligent agents remains unclear in the industry, with varying capabilities and interaction methods leading to user expectation gaps [2][4]. - Intelligent agents are expected to perform tasks autonomously, such as booking flights or organizing information, without requiring users to navigate multiple applications [2]. - Current limitations of intelligent agents include their ability boundaries, which are constrained by factors like permissions, API interfaces, and data silos [3][4]. Group 3: Technological Advancements and Customer Needs - The rapid development of AI technologies is driven by customer demands, leading to an acceleration in the release of visual and multimodal models [8][9]. - Tencent's intelligent agent platform integrates various AI capabilities to create systematic solutions that address real business needs [9][14]. - The company has developed significant knowledge management and workflow capabilities to support the deployment of intelligent agents [10]. Group 4: Challenges and Future Outlook - Challenges include the evolving technology of agent frameworks and the need for customers to understand how to effectively utilize these agents [11]. - The relationship between intelligent agents and large models is reciprocal, with advancements in one area promoting progress in the other [13]. - As model capabilities improve and customer understanding increases, intelligent agents are expected to deliver value across more industry scenarios [12].