三、公司AI智能体的构建进展如何？ 答：利欧数字在营销领域持续深化AI智能体布局，以利欧 数字全链条业务场景为基础，实现从策略制定、创意生产到用 户运营等全环节智能化升级，打造营销各环节专属智能体（包 括策略智能体、创意智能体、投放智能体、运营智能体等）。 目前，公司已为部分业务岗位构建多个智能体并应用于实际业 务。 证券代码：002131 证券简称：利欧股份 利欧集团股份有限公司投资者关系活动记录表 编号：2026-001 | | ■特定对象调研 | □分析师会议 | | --- | --- | --- | | 投资者关系活 | □媒体采访 | □业绩说明会 | | | □新闻发布会 | □路演活动 | | 动类别 | □现场参观 | □一对一沟通 | | | □其他 华创证券 姚蕾 | | | | 南华基金 王梦恺 | | | | 中信证券 朱欣頔 国金基金 杨凯雯 | | | | 广发基金 徐运铎 | | | | 永赢基金 陈永晟 | | | | 华安基金 张杰伟 | | | 参与单位名称 | 博道基金 高笑潇 | | | 及人员姓名 | 复胜资产 马晋隆 | | | | 竹润投资 邹伟 | | | | ...

想象一下，只需要一句话描述，AI 就能为你拍出一部完整的短剧？以后可能真的人人都能当导演了。不用学复杂的拍摄技巧，不用买昂贵设备，甚至不用找演 员。有个好想法，AI 就能帮你实现。 为了让这个想法变成现实，香港大学黄超教授团队开源了 ViMax 框架，并在 GitHub 获得 1.4k + 星标，专注于 Agentic Video Generation 的前沿探索。通过多智能体 协作，ViMax 实现了真正的 "自编自导自演"—— 从创意构思到成片输出的完整自动化，把传统影视制作的每个环节都搬进了 AI 世界。 ViMax 的 "一人剧组" 有多强？ 它就像一个数字化的全能团队 ——AI 编剧负责写剧本，AI 导演掌控节奏和镜头语言，AI 摄像师负责构图和视觉呈现，AI 剪辑师 精心打磨每个细节。这些 AI 小伙伴会自己讨论创意，分配任务，协调配合。你只需要输入一个想法，AI 就能独立完成整个制作流程，输出千赞级别的视频内容。 在 AI 视频制作领域，我们正在见证一场从 "片段生成" 到 "系统化制作" 的重要转变。这不仅仅是技术升级，更是创作方式的根本改变。 实验室地址：https://sites.goog ...

Core Insights - The three principles proposed by the founder of Yidu Technology, Ms. Gong Rujing, emphasize that AI should not replace doctors, should remain contextually relevant, and should not abandon inclusivity in healthcare [1][9] - The forum held in Kuala Lumpur gathered global investment leaders and industry experts to discuss the future of AI in healthcare [1] Group 1: AI in Healthcare - Ms. Gong Rujing highlighted that there is no "universal AI," but rather precise solutions that fit into workflows, emphasizing the need for high-quality medical data and seamless integration into the entire diagnostic and treatment process [3][4] - Yidu Technology's "AI Medical Brain" YiduCore has processed over 6 billion medical records, establishing a robust foundation for AI applications in healthcare [4] - The company has developed over 1,000 intelligent agents covering various clinical needs, significantly improving efficiency and standardization in areas like oncology [4][6] Group 2: Inclusivity in Healthcare - The ultimate mission of AI in healthcare is to provide accessible and affordable health services to everyone, as stated by Ms. Gong Rujing [6] - Yidu Technology has participated in the development of health insurance projects across multiple provinces, offering insurance products with annual premiums as low as 100 yuan, benefiting millions [6][7] - The company has served over 40 million insured users, demonstrating the social equity driven by technology [6] Group 3: Globalization and Localization - Ms. Gong Rujing emphasized that globalization in AI healthcare should not be a mere replication of solutions but should involve "technology generalization + local adaptation" to create value with local partners [8] - Yidu Technology has implemented localized projects, such as the BruHealth digital health platform in Brunei, covering over 60% of the population [8] - The company is also involved in Singapore's "MIC@Home" project, supporting remote monitoring and management of discharged patients [8]

Core Insights - The "Athena Cup" innovation and entrepreneurship competition showcased 19 teams out of 300, highlighting the importance of AI in bridging the gap between technology and practical applications in industries [2] - Liu Bo, Vice President of Dingjie Smart, emphasized the need for a collaborative approach to address the complexities and uncertainties in enterprise decision-making through AI and data synergy [2] - The challenge of applying general AI models in industrial settings is attributed to their inability to grasp specific, tacit knowledge unique to individual factories [2] Group 1: AI and Industrial Applications - The commercialization of AI models is accelerating across various industries, but the "last mile" application challenge remains prevalent in industrial contexts [2] - The focus on digitizing industrial knowledge involves capturing unstructured data through multimodal and fragmented approaches, which can lower the barriers to knowledge storage [3] - By accumulating sufficient data across different industries, a "process knowledge graph" can be constructed to enhance data quality and improve the effectiveness of AI model applications [3] Group 2: Multi-Agent Collaboration - Dingjie has updated its Indepth AI platform and launched the Manufacturing Multi-Agent Protocol (MACP) to facilitate efficient collaboration among AI agents [4] - The platform allows for dynamic sensitivity analysis and knowledge querying, enabling the generation of comprehensive operational plans based on various business metrics [4] - The practice of multi-agent collaboration requires understanding the enterprise's knowledge system and business processes to effectively manage and control resources [5] Group 3: Future Directions - The development of AI applications within enterprises is expected to follow a pattern similar to Moore's Law, potentially doubling every 18 months, which poses challenges for management and coordination [3] - Dingjie Smart aims to deepen technological research and ecosystem development, guided by an "Intelligent+" strategy to foster innovation and breakthroughs in AI applications [5]

Core Insights - The article discusses the importance of aligning cutting-edge technology with industry needs as enterprise-level AI approaches a critical year for large-scale implementation [1] - The "Athena Cup" innovation and entrepreneurship competition aims to bridge the gap between innovative projects with robust technology and the industrial ecosystem [1] Group 1: Company Overview - Dingjie Smart, established in 1982, has 43 years of experience in the digital intelligence field, focusing on the integration of "AI + Industrial Internet" technology [1] - The company has developed the Dingjie Athena Smart Native Base, which features a cloud-edge-end collaborative architecture and focuses on scenario-based intelligent algorithm development [1] Group 2: AI Application and Innovation - During the competition, Dingjie Smart's Executive Vice President Liu Bo demonstrated the workflow of the Athena Indepth AI multi-agent collaborative platform, showcasing its ability to assist in creating reliable business plans under various constraints [2] - Liu Bo emphasized that the future competitiveness of enterprises will significantly depend on the density of internal AI applications, which may follow Moore's Law, potentially doubling every 18 months [3] Group 3: Competition Details - The competition was structured in a "2+4" model, consisting of two groups (innovation and entrepreneurship) and four tracks (advanced manufacturing, digital future, health technology, green economy), attracting around 300 teams from both sides of the Taiwan Strait [5] - After multiple rounds of competition, 19 teams advanced to the finals, with awards given for outstanding projects, including two excellence awards, four distinguished projects, and thirteen potential projects [5]

Core Insights - The popularity of educational large models is declining, with chat-based apps losing their prominence in the market [1][2] - The industry is becoming more rational, recognizing the limited short-term commercialization potential and focusing on long-term development of higher-quality intelligent agents [2] Market Trends - In October, only three educational apps made it to the top 30 in monthly active users, each with fewer than 3 million users, indicating a significant drop in engagement for chat-based educational products [3] - The head of general large models has surpassed 100 million monthly active users, which has squeezed the survival space for vertical educational products [3] Product Development - Educational companies are shifting from developing standalone chat-based apps to integrating large models into traditional products, focusing on multi-agent collaboration [5][6] - Multi-agent systems enhance learning interaction diversity and depth, moving beyond the limitations of single-agent models [5] Commercialization Challenges - Despite ongoing technological upgrades, the commercialization of educational intelligent agents has not shown significant improvement, with some apps generating only tens of millions in revenue [8][9] - The focus is shifting towards the usage frequency and quality of educational intelligent agents rather than immediate revenue generation [10] Investment in R&D - Companies like Huatu Education have significantly increased R&D spending, with a 160.41% year-on-year rise to 145 million yuan, primarily to expand their research teams [11] - The investment in high-quality data and user engagement is seen as a necessary step to build future barriers in the educational intelligent agent market [12][13] Future Outlook - Huatu Education plans to develop a digital companion for exam training, aiming to create a comprehensive ecosystem of intelligent agents [14]

Core Viewpoint - OPPO and Ant Group have signed a strategic cooperation agreement to collaborate in various fields including AI, service ecosystems, near-field interaction, healthcare, and insurance [1] Group 1: AI and Technology Collaboration - The partnership aims to advance the multi-end intelligent agent collaboration solution, Agent Hub Access (AHA), which was showcased at the 2025 OPPO Developer Conference [1] - The AHA solution is designed to enable efficient collaboration between OPPO's system-level AI and Alipay's agents, covering smart service scenarios such as transportation, government services, healthcare, logistics, food delivery, and utility payments [1] Group 2: New Features and Services - A new feature, "Alipay Tap to Send Red Envelopes," will be launched by the end of October, marking the first implementation of this function in domestic smartphones [1] - OPPO Wallet will collaborate with Ant Insurance to cover various insurance types, including auto insurance and accident insurance [1] Group 3: Healthcare Initiatives - The two companies will enhance their collaboration in healthcare services, promoting the popularization and service upgrade of AI health applications [1] Group 4: Strategic Importance - OPPO's Senior Vice President and Chief Product Officer, Liu Zuohua, stated that the cooperation aims to explore multi-agent collaboration in vertical fields and to jointly create industry-leading AI technology solutions [1] - This partnership is viewed as a significant practice in AI ecosystem development between terminal manufacturers and internet service platforms [1]

Core Insights - The automotive industry is undergoing a wave of smart technology, with the evolution of intelligent cockpits becoming a key insight into the future of the industry [1] - The China Society of Automotive Engineers (CSAE) released the "2025 Automotive Intelligent Cockpit Technology Trends" report during the 2025 International Automotive Intelligent Cockpit Conference, outlining key innovation trends and development paths for the next three years [1][25] Group 1: Technological Innovations - Hardware and algorithm innovations are accelerating the optimization of large model architectures, model compression, and acceleration technologies for cockpit applications, enhancing immersive and secure user experiences [3] - The transition of electronic and electrical architecture from distributed to centralized computing platforms will break down data barriers between vehicle domains, gradually enhancing the intelligent cockpit [10] - The integration of HUD technology will evolve through optical innovations and algorithm advancements, increasing driving safety and user trust in intelligent driving systems [14] Group 2: User Experience Enhancements - Future intelligent cockpits will go beyond simple functionality stacking, providing immersive and integrated user experiences through multi-agent collaboration [7][9] - The development of multifunctional interiors is shifting from providing a single comfort experience to integrating visual, tactile, and biometric sensing technologies for a more enjoyable environment [16][18] - Modular design and ecological interactivity will transform cockpits from single-function spaces to customizable environments, leveraging expanded interfaces and software ecosystems [19][22] Group 3: Industry Trends and Future Outlook - The rapid development of intelligent cockpits is just a glimpse of the overall smart upgrade in the automotive industry, driven by systematic collaboration in vehicle electronic architecture, software platforms, and intelligent technologies [25] - The CSAE has been conducting ongoing research since 2021 on the overall direction of automotive technology, with the latest report being a deep exploration into intelligent cockpits [25] - The upcoming release of the "2026 Annual China Automotive Technology Trends" report is anticipated during the 32nd CSAE Annual Conference, highlighting continued advancements in the sector [26]

Core Viewpoint - The article emphasizes the practical application of generative AI, showcasing vivo's advancements in AI technology that enhance user experience and privacy through localized processing and personalized intelligence [6][30]. Group 1: AI Capabilities and Innovations - vivo introduced the "One Model" concept, a lightweight 3B end-side multimodal reasoning model that aims to provide a sustainable AI experience focused on user personalization rather than just parameter competition [8][9]. - The new AI capabilities include a 30 billion parameter model that can run smoothly on flagship mobile SoCs, achieving performance comparable to industry-leading 4B language models with a 60% reduction in parameters [9][11]. - The Blue Heart 3B model supports both language and multimodal tasks, allowing for complex reasoning to be performed locally on devices, thus enhancing efficiency and privacy [13][20]. Group 2: User Experience and Personalization - The integration of AI into the mobile operating system allows for a seamless user experience, where AI acts as a personal assistant capable of understanding and executing tasks without relying on cloud services [15][18]. - The AIOS framework is designed to mimic human cognitive processes, enabling real-time perception, memory, execution, and autonomous planning, which significantly improves task efficiency [20][21]. - vivo's approach to AI emphasizes the importance of personal data integration, creating a personalized AI experience that is both efficient and secure [18][30]. Group 3: Ecosystem and Collaboration - vivo aims to build an open ecosystem by collaborating with developers and partners, allowing for the rapid deployment of new AI capabilities and applications [23][26]. - The company has established partnerships to enhance its AI offerings, such as collaborating with Ant Group's AI health application AQ, which provides comprehensive medical services [28][29]. - vivo's vision includes equipping over 300 million devices with robust local AI capabilities within the next three to five years, indicating a strong commitment to advancing AI technology [31].

Core Insights - The article discusses the evolving landscape of digital asset security, highlighting the emergence of state-sponsored hacking groups, particularly North Korea's Lazarus Group, which has stolen over $6 billion in cryptocurrency since 2017, with $2 billion taken in 2025 alone [2][11] - It emphasizes the need for a paradigm shift in security measures, moving from traditional static defenses to AI-driven dynamic and proactive strategies to combat advanced persistent threats (APTs) in the digital asset space [4][10] Group 1: Evolving Threat Landscape - The digital asset security environment has fundamentally changed, with threats now involving state-sponsored professional hacker organizations rather than just individual criminal groups [2][3] - The Lazarus Group's activities are strategically aimed at funding North Korea's military programs, particularly nuclear weapons and missile development [2] - The characteristics of APTs in the digital asset realm include direct financial stakes, short attack chains, and highly customized attack methods targeting high-net-worth individuals and corporate executives [3] Group 2: AI-Driven Security Transformation - AI and intelligent agent technologies are essential for evolving security paradigms, as they align well with the transparent and data-rich nature of the digital asset world [4][5] - The shift from rule-based to behavior-driven defenses allows for the detection of previously unseen and highly disguised attack methods [4] - AI's ability to analyze vast amounts of on-chain data enables proactive threat prediction and real-time monitoring, crucial for countering state-level APTs [5][9] Group 3: Implementation of Intelligent Defense Systems - The concept of a "smart agent army" is introduced, where AI technologies create a multi-layered defense system for digital assets [6][8] - On a personal level, AI agents act as "digital bodyguards," monitoring wallet activities and intervening in real-time during suspicious transactions [7] - At the enterprise level, AI systems function as risk control officers, analyzing transaction patterns and freezing suspicious accounts before money laundering occurs [7] Group 4: Future of Digital Asset Security - The future security framework will rely on a collaborative ecosystem of multiple intelligent agents, enhancing the overall security capabilities [8] - The "on-chain firewall" concept is proposed, which utilizes AI for proactive defense, real-time monitoring, and rapid response to threats [9][10] - This AI-driven firewall represents a shift from passive vulnerability management to active risk intervention, establishing a comprehensive security lifecycle for digital assets [10]