在AI 2.0时代初期，大模型更多以单一工具的形态存在，帮助企业员工处理文档生成、数据统计等基础 工作，本质上是对传统办公流程的效率优化。 多智能体进阶 但随着技术的演进与应用的深化，这种"一人一工具"的模式已难以承载AI的价值。麻省理工学院 （MIT）所发布的报告《The GenAI Divide: State of AI in Business 2025》甚至指出，"95%的企业AI试点 项目都以失败告终"。 当大模型技术的军备竞赛从参数规模的比拼逐渐转向应用落地的"深水区"，企业的组织形态正迎来一场 前所未有的"奇点式"变革。从单一的AI工具辅助，到智能体（Agent）的独立作业，再到如今多智能体 （Multi-Agent）协同，AI技术不再是游离于企业核心架构之外的辅助手段，而是开始深度嵌入组织管 理、生产运营、商业决策的每一个环节。 如今，多智能体技术的崛起正在打破这一局限，推动企业从"工具辅助"向"团队协同"的进阶，进而重构 企业的组织结构与价值生产链条。 2026年1月5日，北京零一万物科技有限公司（以下简称"零一万物"）正式对外发布了《中国企业智能体 2026六大预判》，并对外展示了"万智2. ...

智东西12月19日报道，由三名00后武汉大学校友创办的大模型领域科技创企模态跃迁（MercAllure），已完成两轮累计数千万元融资，投资方 包括深圳高新投、力合科创、楚天凤鸣天使基金、武汉基金、奇绩创坛等机构。 自2023年8月在武汉注册成立以来，这支年轻的创业团队聚焦一件事：打造一套"智能体的3D打印机"——跃迁基座。 跃迁基座是一套面向大规模AI应用的综合技术中台，整合了多模态模型、智能体框架、RAG框架、合成数据引擎、本地推理加速技术等中间 组件。 今年4月，模态跃迁CEO邴龙志为创新创业青年企业家代表，在武汉市民营经济发展大会第一排就座并发言，引起广泛关注；11月，模态跃迁 先后成功入选"光谷AI新势力八强"榜单。 做Agent Infra的创企并不鲜见，这支从高校孵出的团队，如何趟出一条让企业级智能体真正落地的新路来？ 近日，模态跃迁CEO邴龙志、COO常元和、CTO丁震宇三位联合创始人与智东西进行了深入交流。 据透露，模态跃迁团队规模今年快速壮大，已发展至约40人，其中80%以上为硕博生。该公司文化强调"年轻、开放、扁平"，并把开源社区当 作关键的招人渠道，来自北京、深圳、成都、西安、杭州等地的 ...

责任编辑：山上 关键词阅读：智能体 大模型 中关村科金 12月9日，"超级连接·智见未来"EVOLVE 2025峰会在北京召开。中关村科金与华为云、阿里云、百度智 能云、火山引擎、亚马逊云科技、超聚变、软通动力等产业领军企业，共同发布"超级连接"全球生态伙 伴计划，携手打造开放、连接、可持续的"人工智能+"产业协同生态圈。 会上，中关村科金公开企业级智能体落地路线图，并发布"3+2+2"智能体产品矩阵，包括得助大模型平 台5.0、得助智能客户平台5.0、得助智能工作应用平台及得助金融智能体平台、得助工业智能体平台等 一系列产品和解决方案，覆盖营销服、办公、研发生产等企业核心场景。 目前，中关村科金相关产品已服务2000余家行业头部客户，业务遍及180多个国家和地区。 ...

Core Insights - The article discusses the launch of the "3+2+2" full-stack intelligent agent product matrix by Zhongguancun KJ at the EVOLVE 2025 summit, aimed at addressing challenges in enterprise AI implementation [1][14] - The product matrix is designed to provide comprehensive solutions across technology, application, and industry, enhancing the usability of AI technology [1][14] Group 1: Technical Foundations - The three foundational platforms support the intelligent agent's deployment by ensuring a full-cycle guarantee across model, capability, and data dimensions [2] - The upgraded Dazhu Model Platform 5.0 serves as the core engine for enterprise-level intelligent agents, integrating over 300 ready-to-use agents across six industries, achieving a success rate of over 95% in deployment [2] - The AI Capability Platform offers high-precision recognition tools tailored for vertical industries, while the AI Data Platform focuses on activating data value for informed decision-making [2] Group 2: General Scenario Platforms - Two general scenario platforms enhance core business processes, focusing on customer operations and office collaboration [3] - The Dazhu Intelligent Customer Platform 5.0 sets a new standard for human-machine collaboration in marketing, customer service, sales, and overseas operations, significantly improving performance and efficiency [3][6] Group 3: Industry-Specific Solutions - The Dazhu Financial Intelligent Agent Platform addresses specific needs in the financial sector, supporting over 500 leading financial institutions and enabling product and service innovation [9][10] - The Dazhu Industrial Intelligent Agent Platform collaborates with industry partners to optimize core business processes, achieving significant efficiency improvements in production and energy management [12] Group 4: Open Ecosystem Empowerment - Zhongguancun KJ has launched the "Super Connection" global ecosystem partner program, collaborating with major cloud service providers to enhance the adaptability of the "3+2+2" product matrix across various industries [13][14] - The product system has already served over 2,000 leading clients across more than 180 countries, establishing itself as a preferred solution for enterprise-level intelligent agent deployment [14]

中关村科金正式发布得助大模型平台5.0，它是高效构建企业级智能体的全能基座，帮助企业更快、更好落地AI创 新。全面升级的智能体集市，集成了金融、工业、汽车、零售、交通、政务六大行业超过300个智能体，即取即 用，让企业可以快速进行场景验证，更加聚焦创新而非基础建设。此外，线索分析数字员工帮助企业智能洞察客 户需求，精准连接销售机会，在汽车客户实践中可实现超过55%的到店线索量提升；智能写作可实现超过10万字 的专业报告撰写，通过灵活接入企业内部知识和网络信息，支持精准溯源和交叉验证；智能审核是企业全面审查 关键场景合规风险的利器，通过融合使用大小模型与规则引擎，进行风险等级评估并自动生成可视化报告等。 12月9日，中关村科金在2025大模型与智能体产业创新峰会上，公开企业级智能体落地路线图，并发布"3+2+2"智 能体产品矩阵，包括大模型平台、AI能力平台、AI数据平台三大基础平台，智能客户平台、智能工作应用平台两 大通用场景应用平台，以及金融和工业两大行业智能体平台，助力企业快速开发及使用智能体。 中关村科金总裁喻友平表示，从互联网、移动互联网到人工智能的演进，本质是连接的进化。智能体作为超级连 接器，通过 ...

Core Insights - Oracle aims to redefine its position in the enterprise AI market with a product suite centered around "AI + Data" as it seeks to reshape its image in the AI era [2] - A report from MIT indicates that 95% of generative AI investments have not yielded substantial benefits for enterprises, highlighting the importance of data infrastructure and architecture [2] Group 1: Strategic Focus - Oracle's strategic focus has shifted from cloud infrastructure to overall AI capabilities, emphasizing openness in technology architecture, multi-cloud support, and external ecosystem collaboration [4] - The introduction of Oracle AI Database 26ai is positioned as a "next-generation AI-native database" that supports the creation and deployment of intelligent applications [4][5] Group 2: Product Offerings - Oracle AI Data Platform (AIDP) aims to integrate databases, data lakes, and large model services under a unified architecture, providing tools for data visualization and workflow management [5] - The new OCI Zettascale10 Cluster supercomputing cluster supports up to 800,000 GPUs and is designed for high throughput and low latency, serving as a performance engine for OCI [5] Group 3: Partnerships and Collaborations - Oracle has partnered with Dify, an open-source LLM application development platform, to embed AI into the data layer and enhance enterprise-level intelligent applications [4][6] - The collaboration with Dify allows for the integration of structured and vector data into Dify's workflows, enhancing the capabilities of Oracle's AI Database 26ai [7] Group 4: Market Positioning - Oracle's collaboration with Dify is seen as a strategy to tap into more enterprise-level scenarios amid increasing competition in AI infrastructure [8] - The company is also expanding its data center deployments in the Middle East and Africa to cater to the outbound needs of Chinese enterprises [8]

Core Insights - The 2025 World Internet Conference in Wuzhen highlighted significant innovations in AI, with Ant Group's Ant Financial Technology receiving two major awards for its contributions in the field of large models and AI integration into industries [1][2] - Ant Group's Agentar platform has been instrumental in providing a comprehensive foundation for building intelligent agents, addressing challenges in data integration, model stability, and compliance risks [1][2] Group 1 - Ant Group's Agentar platform has enabled the launch of over 200 intelligent agent applications in key financial sectors such as wealth management, risk control, and smart marketing [2] - These applications have demonstrated effectiveness by reducing fraud losses by 30% and increasing marketing conversion rates by 25%, thereby enhancing business growth and user experience for financial institutions [2] - The Energy TS model, focused on the energy sector, offers integrated solutions for power generation and management, showcasing the potential of AI in transforming traditional industries [2] Group 2 - Ant Group is committed to advancing the application of large model intelligent agents in critical industries like finance and energy, aiming to facilitate a broader intelligent transformation across various sectors [2] - The company emphasizes the importance of creating a safe, efficient, and compliant AI ecosystem as it continues to deepen its efforts in enterprise-level intelligent agent services [2]

Core Viewpoint - Minglue Technology, founded by Wu Minghui, has successfully listed on the Hong Kong Stock Exchange, achieving a market capitalization of HKD 42 billion and a stock price increase of 106% on its debut, positioning itself as the "first stock of Agentic AI" [2][27]. Company Overview - Minglue Technology focuses on "enterprise-level intelligent agents," providing marketing and operational tools to a range of industries, including consumer goods and automotive [2][5]. - The company has secured significant clients, including 135 Fortune Global 500 companies, and reported revenue of RMB 640 million for the first half of the year [2][5]. Financial Performance - The company has faced operational profit losses from 2022 to 2024, with a projected return to profitability in the first half of 2025 [2][27]. - As of June 30, 2023, the company reported a net cash flow used in operating activities of RMB -117.415 million, indicating ongoing financial challenges despite improvements [26]. Investment and Shareholding - Tencent is the largest shareholder, owning 25.96% of the company, while Wu Minghui retains 10.28% of the shares but holds 53% of the voting rights through an AB share structure [3][5]. - The recent IPO raised HKD 1 billion, which will alleviate some financial pressures and support further research and development [27]. Product Development - The company has developed several marketing products, including the "Miaozhen System," which generated RMB 320 million in revenue in the first half of the year, accounting for a significant portion of total income [7][8]. - Minglue has also launched a proprietary large model product line called DeepMiner, aimed at ensuring data accuracy and traceability in enterprise environments [10][18]. Market Expansion - The company aims to expand its services beyond large enterprises to the 60 million small and medium-sized enterprises in China, significantly increasing its potential customer base [22][24]. - By leveraging standardized AI products, Minglue seeks to create a powerful virtual marketing department that can be replicated at a low cost, thus democratizing access to marketing capabilities [24].

Core Insights - The 2025 China Mobile Global Partner Conference highlighted the theme "Carbon-Silicon Symbiosis, Co-creating the AI+ Era," where CaiXun Co., Ltd. was invited as a long-term partner of China Mobile to participate in the launch of the "Lingxi Intelligent Body Empowerment Ecosystem Plan" [1][2] - CaiXun's chairman, Yang Liangzhi, emphasized that AI is evolving from a simple algorithmic tool to a "digital life form" capable of perception, understanding, decision-making, and action, creating a self-optimizing intelligent loop within enterprise operations [1][2] Company Strategy - CaiXun's AI strategy focuses on enterprise-level applications, starting with its self-developed platform Rich AIBox 1.0/2.0, which includes modules such as enterprise knowledge base, intelligent customer service, smart marketing, and AI BI, aiming to deeply integrate AI into business processes [1][2] - The company aims to transition from "single-point innovation" to "ecological collaboration" in the AI industry, positioning itself as a "technology foundation + ecological connector" to rapidly implement AI across various industry scenarios [2][3] Future Outlook - CaiXun has proposed an "AI 2.0" strategy to develop the next generation of enterprise-level intelligent systems with self-learning, self-reflection, and self-collaboration capabilities, focusing on multi-modal understanding, complex task decomposition, and emotional interaction [2][3] - The competitive landscape in AI is shifting from "model competition" to "implementation competition," with CaiXun emphasizing the importance of industry depth and application complexity in the Chinese market [3]

Core Insights - Lenovo Group reported a 22% year-on-year revenue growth for Q1 of the 2025/2026 fiscal year, reaching 136.2 billion yuan, marking a historical high for the same period [1] - The company's net profit under non-Hong Kong Financial Reporting Standards also increased by 22% to 2.816 billion yuan [1] Business Segments Performance - The Intelligent Devices Group (IDG) generated revenue of 97.3 billion yuan, a 17.8% increase, with the PC business achieving its fastest growth in 15 quarters [1] - The Infrastructure Solutions Group (ISG) saw a revenue growth of 35.8% [1] - The Solutions and Services Group (SSG) reported a revenue increase of 19.8%, with an operating profit margin of 22.2% [1] AI and Market Trends - Non-PC business revenue now accounts for 47% of total revenue, indicating a more balanced growth structure and supporting the company's mixed AI strategy [1] - Lenovo's AI PC shipments exceeded 30% of total PC shipments, with AI PC shipments in the Chinese market reaching 27% of total notebook shipments [2] - The "Tianxi" personal super intelligent agent has seen a significant increase in user engagement, with a weekly active user ratio averaging 40% [2] AI Infrastructure and Product Development - Lenovo is building a comprehensive AI product system centered around a hybrid cloud platform, intelligent agent management platform, and industry-specific solution library [2] - The AI infrastructure business experienced a remarkable 155% year-on-year revenue growth [2] - The company is developing an AI model factory and intelligent agent platform to enhance its mixed AI capabilities [2] Financial Reporting and Market Focus - Under Hong Kong Financial Reporting Standards, Lenovo's net profit increased by 108% to 3.66 billion yuan, primarily due to non-cash gains from stock option valuation changes [2] - The management emphasizes that non-Hong Kong Financial Reporting Standards provide a clearer reflection of the company's core operational performance and quality [2]