（原标题：李开复： AI落地企业，智能体是CEO最需关注的核心技术） Gartner曾在报告中预测，到2028年，33%的企业级软件应用将整合AI Agent，届时，15%的日常工作任 务决策可实现完全自主化。 李开复表示，AI Agent浪潮是中国产业升级、实现数智化转型必须抓住的机遇，将深刻改变组织的工作 方式和组织形态，成为推动下一轮商业智能变革的核心引擎。 （零一万物CEO李开复、联创沈鹏飞接受记者采访 董静怡/摄） 从"降本"到"增效" Agent已经不是一个新的概念，多家大厂和创业公司纷纷布局，但市场对于其实际应用效果一直有疑 虑。 "AI Agent如果只能做到降本，那企业客户的付费意愿其实还是偏低的。"零一万物联合创始人沈鹏飞向 21世纪经济报道记者表示，企业更为关注的是其能否真正带来"增量"价值。 而如今，Agent技术在经历重要演进。 21世纪经济报道记者 董静怡 上海报道 海内外科技企业Agent产品持续涌现，行业对于AI Agent在技术架构和产品形态层面正在逐步成型， 2025年正在成为的"AI Agent元年"。 "超级智能的AI和Agent将无处不在，我们相信它将带来有史以来最大 ...

Core Viewpoint - The article emphasizes that 2025 will be a pivotal year for AI Agents, highlighting the shift from traditional language models to more versatile AI Agents capable of performing complex tasks through simple natural language commands [4][6]. Group 1: AI Agent Development - The rise of AI Agents is driven by the increasing capabilities of mobile devices, with predictions indicating that by 2027, global AI mobile penetration will reach approximately 40%, with an expected shipment of 522 million units [9]. - Major tech companies, including Apple, are launching their own AI models, such as Apple Intelligence, while domestic manufacturers like Xiaomi and OPPO are also entering the market with their versions [9]. - The challenge lies in overcoming app isolation, as different applications typically prevent data sharing, necessitating either API agreements or the use of accessibility permissions to enable AI operations [11]. Group 2: Security and Privacy Concerns - The use of accessibility permissions raises significant privacy risks, as AI applications can potentially access sensitive information, including payment passwords and chat records [6][12]. - There are two main technical paths for AI Agent development: an interface model that requires cooperation between app developers and a non-interface visual solution that utilizes system-level permissions [11]. - The article notes that while the interface model is safer, it is also more complex and costly due to the need for adaptation across different devices [12]. Group 3: Market Potential and Growth - The AI Agent market is projected to grow from $5.1 billion in 2024 to $47.1 billion by 2030, with a compound annual growth rate of 44.8% [17]. - A survey indicated that over half of respondents have encountered data privacy and security issues, with 60.09% believing that AI could uncontrollably collect and process personal information [17]. Group 4: Regulatory and Industry Response - The article suggests that proactive measures are essential for managing AI risks, with companies needing to enhance their awareness of privacy issues [19]. - Recommendations include defining the minimum data required for specific functions and establishing data quality management standards to ensure data integrity and security [19][21]. - Regulatory bodies are encouraged to adopt agile governance strategies to address the rapid evolution of technology and its associated risks, ensuring a balance between user protection and innovation [21].

Group 1 - The core viewpoint is that 2025 is expected to be a pivotal year for AI Agents, with significant advancements in technology and market penetration anticipated [1] - AI Agents are capable of performing more complex tasks than traditional chatbots, utilizing special accessibility permissions in Android systems to execute commands on behalf of users [1][3] - The market for AI Agents is projected to grow significantly, with a forecasted increase from $5.1 billion in 2024 to $47.1 billion by 2030, reflecting a compound annual growth rate of 44.8% [9][10] Group 2 - The global penetration rate of AI mobile devices is expected to reach approximately 40% by 2027, with an estimated shipment of 522 million units [2] - Major companies, including Apple and various Chinese manufacturers, are launching their own AI models, indicating a competitive landscape for AI mobile assistants [2][3] - The challenge of enabling AI Agents to operate across different applications is highlighted, with two main technical paths identified: interface-based and non-interface visual solutions [3][4] Group 3 - The use of accessibility permissions raises significant privacy and security concerns, as these permissions can allow AI applications to access sensitive user information [4][7] - There are reports of fraudulent activities exploiting accessibility permissions, where scammers have manipulated users into granting these permissions for malicious purposes [8][9] - The industry is currently in a phase of exploration, with no perfect solutions available, but the interface-based approach is considered more sustainable for long-term development [11] Group 4 - Regulatory measures are being discussed, including the need for explicit user consent before enabling accessibility services for AI applications [13][14] - The importance of establishing clear data management standards and privacy protection mechanisms is emphasized to mitigate risks associated with AI Agents [12][14] - Collaboration among various stakeholders, including tech companies and regulatory bodies, is necessary to address the complexities of AI Agent deployment and ensure user safety [14]

Core Insights - The forum "Intelligent Agent Driven Supply Chain Transformation" was held in Shanghai, showcasing the launch of ONEAIX's autonomous planning AI Agent platform, marking a new phase in supply chain intelligence [1][2] - The event gathered nearly 200 international experts, entrepreneurs, investors, and government officials to discuss innovations in the supply chain logistics sector [1][5] - The CEO of ONEAIX highlighted the significant role of AI in reshaping global trade, which is currently valued at $1.7 trillion, accounting for approximately 20% of total service trade [1][2] Company Developments - ONEAIX upgraded its brand from "1Data" to "ONEAIX" and introduced the 运小沓数字员工4.0, the first AI Agent in the supply chain sector with autonomous planning capabilities [2][3] - The company has evolved its 运小沓 product line from 1.0 to 4.0, enhancing automation and efficiency across various supply chain scenarios [2][3] - The CEO emphasized a commitment to customer-centric product development and the integration of AI as a core technological driver for future innovations [2] Industry Trends - The "Global Port and Shipping Information Technology Development Report (2024)" was released, summarizing technological innovations and predicting future trends in the port and shipping sector [4][5] - The report highlighted the successful implementation of AI technologies, such as the "智行问价Agent" digital employee, which significantly improved service levels at Qingdao Port [4] - The forum featured discussions on the necessity of deploying AI digital employees in leading supply chain logistics companies to unlock efficiency benefits [6][7] Expert Contributions - The forum included presentations from various industry leaders and academics, focusing on the latest achievements and practices in supply chain AI applications [5][6] - Notable speakers included representatives from top universities, investment firms, and shipping companies, providing diverse perspectives on the transformation driven by intelligent agents [5][6] - The event facilitated high-quality discussions on the opportunities and challenges of AI in supply chain management, contributing to a comprehensive understanding of the industry's future [6][7]

Core Insights - The 2025 World Artificial Intelligence Conference (WAIC) held in Shanghai focused on the theme "Collaborating in the Intelligent Era," showcasing innovations in AI applications within the human resources sector, particularly through the upgraded AI recruitment assistant "Ailin" by Zhilian Recruitment [1][3]. Company Developments - Zhilian Recruitment's CTO, Wang Hao, highlighted the significant time savings for HR professionals by automating repetitive tasks such as screening, inviting, and initial interviews through AI technology [1][3]. - The AI recruitment assistant "Ailin" has evolved through several key stages, starting with internal efficiency improvements, followed by the introduction of features like job posting and automated communication in September 2024, and further enhancements in February 2025 with the integration of DeepSeek-R1 for improved talent matching [3]. AI Technology Advancements - The upgraded "Ailin" can now provide "intelligent full management" services, capable of screening 1,000 resumes in just 5 seconds, conducting multi-modal analysis during interviews, and automatically responding to candidates to enhance conversion rates [3]. - The AI recruitment assistant is designed to cover the entire recruitment process and can be customized to meet the specific needs of different industries, allowing for personalized recruitment experiences [3]. Future Outlook - Wang Hao envisions a future where resumes may become obsolete, as information could be gathered through multi-round dialogues between candidates and AI agents, allowing for deeper insights beyond traditional text-based resumes [4][5].

Group 1 - The computer ETF (159998) experienced a slight decline of 0.97% with a trading volume exceeding 31 million yuan, maintaining the top position among similar products [1] - The ETF saw a net inflow of over 18 million yuan yesterday, with four out of the last five trading days showing net inflows, totaling over 230 million yuan [1] - As of July 21, the latest scale of the computer ETF reached 3.557 billion yuan, marking a historical high in both scale and shares at 3.854 billion [1] Group 2 - The cloud computing ETF (517390) fell by 1.09% and closely tracks the CSI Cloud Computing Industry Index, which includes companies involved in internet services and computing hardware [2] - The AI and office software industry is at a transformative turning point, driven by breakthroughs in large model technology, enhancing productivity and reshaping market dynamics [3] - The computer and software development sector is expected to see growth in various sub-sectors, including AI computing and software outsourcing, with improvements anticipated in the second half of the year due to new technologies and policy support [3]

Core Insights - The AI + office software industry is at a transformative turning point, shifting from tool intelligence to workflow reconstruction, driven by breakthroughs in large model technology [1] - Major international players like Microsoft dominate the high-end market, while domestic companies are rapidly rising due to localized data advantages, policy support, and deep engagement in vertical scenarios [1] - The industry growth engine is fueled by the normalization of online work, deepening digital transformation of enterprises, and the demand for domestic alternatives [1] Technology Developments - Multi-modal interaction, agent technology, and cross-platform integration capabilities are accelerating product innovation in the office software sector [1] - AI Agent technology is evolving office software from efficiency tools to intelligent enterprise hubs, achieving end-to-end operational loops [1] Market Dynamics - The current trend in the industry is towards vertical specialization and ecological integration, with leading companies promoting "AI + tools + business" full-link intelligent collaboration through cross-platform interface encapsulation and business flow integration [1] - Domestic vendors with comprehensive scene integration capabilities and accumulated know-how in vertical fields are expected to continue expanding their competitive advantages under policy empowerment and substitution opportunities [1] Investment Insights - The software ETF tracks a software index compiled by China Securities Index Co., which selects listed companies involved in software development and IT services from the Shanghai and Shenzhen markets [1] - This index exhibits high growth potential and technological orientation, making it suitable for investors focusing on technological innovation and information development [1]

Core Insights - The AICon Global AI Development and Application Conference highlighted the transformative impact of AI on the financial industry, emphasizing a shift from efficiency tools to value engines [1][3] - 2025 is projected to be a pivotal year for AI technology to fundamentally alter core business scenarios in finance, driven by advancements in foundational models and AI Agent technology [3][4] Financial Industry Developments - The continuous evolution of foundational models like DeepSeek and breakthroughs in AI Agent technology have enabled the financial sector to reach a critical point in its intelligent transformation [3] - Qifu Technology has developed an end-to-end risk control capability system based on rigorous innovation in core risk control modules, improving accuracy and efficiency in risk assessment [3] - The company has enhanced its data mining capabilities through multi-modal video capabilities, achieving over 95% accuracy in user feature consistency verification [3] Applications in Financial Services - The intelligent research report system for the securities industry utilizes large models to instantly analyze vast amounts of financial report data, significantly improving analyst efficiency [3] - In the insurance sector, an intelligent underwriting system employs knowledge graphs and multi-turn dialogue technology to drastically reduce the processing time for complex policies [3] - The consumer finance sector benefits from dynamic risk pricing optimization through 360-degree user profiling and cross-validation with macroeconomic data [3] Future Outlook - The integration of AI and finance is expected to lead to a significant transformation in the industry, marking the beginning of a century-long change in financial value creation [4] - The ongoing technological innovations centered around large models are anticipated to break traditional constraints and pave the way for a more efficient, precise, and creative development path in the financial sector [4]

Core Insights - The company reported a revenue of HKD 4.114 billion for the fiscal year 2024, with the sports knowledge subscription platform generating HKD 2.59 billion, a year-on-year increase of 13.68%, accounting for 62.9% of total revenue [1] - The company's loss narrowed to HKD 14.7 million, a 32.2% reduction compared to the HKD 21.7 million loss in 2023, primarily due to increased other income and reduced selling expenses [1] Revenue and Growth - The sports knowledge subscription platform's revenue increased by 13.7% compared to the previous year [2] - The total user base of the platform rose by 21.4% to 106 million, with average monthly active users growing by 17.4% year-on-year [1] Strategic Initiatives - The company aims to diversify its digital sports entertainment business while addressing the competitive landscape and evolving consumer demands [1] - Significant resources have been allocated to R&D, leading to the launch of targeted products in the sports knowledge subscription sector, enhancing competitiveness and user retention [2] Market Expansion - The company is focusing on optimizing its business model in the sports and leisure gaming sector, reducing investment in lower-yield games, and leveraging existing IP for sports game distribution [2] - International expansion efforts have progressed, with successful localized testing in Southeast Asia receiving positive feedback from users [2] Brand and Community Engagement - The company has successfully hosted major events, including the WBC Professional Boxing Championship in Foshan, which garnered over 300 million views, significantly enhancing brand influence [3] - The company is committed to social responsibility by promoting fitness through various sports events, contributing to the development of a strong sports culture in China [1]

Core Viewpoint - The article highlights the strategic investment in Zhipu AI, marking a significant move in the generative AI sector in Hangzhou, with over 1 billion yuan raised from local investment funds, indicating a strong governmental push towards AI development [1]. Group 1: Zhipu AI's Strategic Positioning - Zhipu AI's open-source model is reshaping the global AI innovation landscape, showcasing two main paths for domestic AI breakthroughs: algorithm innovation reducing computational power dependency and building an open-source ecosystem that attracts global developers [2]. - The DeepSeek-R1 model demonstrates a cost-effective approach, achieving performance comparable to billion-dollar models at a cost of $5.6 million, challenging the traditional paradigm of "computational power equals competitiveness" [2]. - Zhipu AI's ChatGLM series has gained significant traction, with over 50,000 stars on GitHub and 30 million downloads, reflecting the value of open-source models as a technological foundation [3]. Group 2: Technological Advancements and Market Trends - The shift towards multi-modal interaction and physical world manipulation is evident, with Zhipu's GLM series models capable of understanding various inputs and executing complex tasks, enhancing efficiency in sectors like finance and education [6]. - The anticipated "open-source week" by DeepSeek and Zhipu's upcoming AutoGLM framework signify a transition from single-model to toolchain open-sourcing, potentially transforming development paradigms and innovation focus [3]. - The AI Agent technology is expected to see significant advancements by 2025, with applications in both enterprise efficiency and personalized consumer services, indicating a structural shift in organizational roles and decision-making processes [5]. Group 3: Regional Economic Impact - Hangzhou's digital economy now accounts for over 28.8% of its GDP, transitioning from e-commerce to hard technology, with strategic goals to become a hub for computational power and AI industry development [7]. - The collaboration between Zhipu AI and local industries is set to enhance the integration of large model technology into various sectors, driving intelligent upgrades across manufacturing, healthcare, finance, and government [8].