而且还能看到360技术优势和产品风格的延续： 选择本地部署和MCP工具集成，而不是通常的云端Host，大模型进行深度检索和社交平台 操作的时候，更方便绕过登陆墙、广告墙，使用方便的同时还不用担心安全问题。 白交 发自 凹非寺 量子位 | 公众号 QbitAI Agent赛道热闹非凡，周鸿祎力推的纳米AI搜索，体验上能有什么不一样？ 首先， 依然很"挤" ，一不小心就会挤爆服务器。 但更进一步实测了一波之后，我们还是发现纳米AI搜索，不管从打开方式还是产品特点，其实都很"老周"…… 直接说结论—— 首先，它把MCP的使用门槛给打下来了 。作为首个真正面向toC市场的MCP平台，普通人也能够真正体验到基于MCP的高阶智能体。以往 的MCP都是面向专业人士，在开发者中流行。但现在纳米AI上的4亿用户都能调用海量MCP工具来完成真实世界的复杂任务。 其次，真正意义上的MCP开放生态 。纳米AI目前已有超100个自研和优选MCP工具，并且更多第三方MCP工具正在进驻中。 具体的细节，咱展开来说。 实测纳米AI万能工具箱 使用纳米AI万能工具箱很简单，只需下载APP应用并注册登录即可，无需额外任何配置，入口就在左侧的「智 ...

Core Insights - The event "Global Tech Network" hosted by Plug and Play China focused on the theme of embracing intelligent agents, highlighting the pervasive impact of AI technology in various aspects of life [1] - The discussions emphasized the evolution of AI, particularly large language models (LLMs), and the importance of understanding key trends driving their development [3][5] Investment Opportunities - The presentation by Zhu Xiaowen outlined six core trends that will drive the future of large models, including refined training, breakthroughs in multimodal integration, and the collaboration between cloud and edge intelligence [3] - The AI maturity model presented by Qian Wenying identified five stages for enterprises transitioning to "super intelligent agents," emphasizing the need for strategic and talent transformation [5][6] Industry Trends - Wu Heng discussed three collaboration modes between humans and AI: embedding, intelligent assistance, and full autonomy, with a focus on the agent model that can execute tasks and close loops, indicating a potential industry transformation [8] - The current applications of AI in customer service, smart office, and smart manufacturing are creating significant value through efficiency and cost optimization [8] Startup Innovations - Six representative startups showcased their innovative applications of intelligent agents, including Wanfa AI, which aims to provide accessible and high-quality legal services through AI technology [12] - Other notable startups include AI Agent Foundry, which offers robust operational support for AI applications, and ALLO TECH, which focuses on applying large model technology to industrial design [14][22] Future Collaborations - Plug and Play China is facilitating opportunities for tech companies to participate in major innovation events and connect with potential partners [10] - The event highlighted the deep resonance between technology and business, illustrating the evolution from efficiency tools to intelligent systems capable of cross-scenario collaboration [24]

Group 1 - The conference "2025 China Digital Valley · West Lake Forum" focuses on AI applications and security governance, gathering 200 policymakers, tech enthusiasts, and industry leaders to discuss breakthroughs in application scenarios, challenges in trust and safety, and new paradigms in agile governance [1][17] - Three main themes are highlighted: breakthroughs in application scenarios, challenges in trust and safety, and agile governance [2] - The event aims to witness the birth of a new paradigm in AI security governance, emphasizing the importance of safety as a competitive edge in the AI landscape [1][17] Group 2 - The first theme addresses the evolution of intelligent agents, differences in application scenarios between China and abroad, and compatibility challenges in cross-industry collaboration [2] - The second theme focuses on constructing a full lifecycle protection system for intelligent agents, including the development of key standards for trustworthiness and risk prevention throughout the training, deployment, and operation phases [2] - The third theme discusses the need for flexible regulatory frameworks that can adapt to the rapid iteration of AI technology, comparing the EU's sandbox regulation with China's negative list model [2] Group 3 - The conference will unveil the "Safety Intelligent Agent Cube: Maturity Model Evaluation Research Report," which is the first domestic maturity assessment system for intelligent agent safety [3] - Practical case studies will be presented, showcasing AI's application potential in vertical fields and revealing governance paths through customized safety solutions and technical tools [3] - A signing ceremony for the first batch of intelligent agent ecological alliances will take place, promoting collaboration among industry leaders to co-create the future of intelligent agents [4] Group 4 - The "Joint Creation of Intelligent Agent Ecological Partners Program" aims to recruit 30 leading enterprises, tech developers, and ecological service providers nationwide to build an open, collaborative, and win-win intelligent agent ecosystem [6] - The program emphasizes shared resources, technical support, and the exploration of foundational large model capabilities for industry applications [7][8] - The initiative seeks to establish benchmark cases in key sectors such as finance, transportation, and healthcare, while also facilitating government technology project applications and industry standard formulation [12]

Core Viewpoint - The rise of industry-specific large models is bridging the gap between general large models and specific industry needs, facilitating intelligent transformation in enterprises and enhancing operational efficiency [1][2]. Group 1: Industry Large Models - Industry large models are trained or optimized based on general large models to cater to specific industry data and tasks, offering high cost-effectiveness and customization [2][3]. - These models can be privately deployed, ensuring data security and allowing organizations to utilize private data effectively [2][3]. - Applications of large models are penetrating various sectors, with digital-native industries like the internet and gaming leading the way due to their high digitalization and data richness [2][3]. Group 2: Company Overview - The company, S&P Zhiyuan, aims to build the most intelligent enterprise management brain, focusing on achieving "autonomous driving" in business operations [3][5]. - S&P Zhiyuan has launched multiple industry large models covering finance, insurance, manufacturing, government, and renewable energy sectors [3][5]. Group 3: Product and Technology - The core product, BPai large model, is based on a trillion-parameter Transformer architecture, integrating advanced technologies for strong language understanding and generation capabilities [5][6]. - The company has developed intelligent agents that act as digital employees, efficiently handling tasks such as intelligent customer service and document recognition [6][8]. Group 4: Market Strategy and Global Expansion - S&P Zhiyuan is actively exploring international markets and has announced brand upgrades to enhance its industry influence and competitiveness [12][13]. - The company’s global strategy is based on thorough market research and aims to meet diverse digital transformation needs in sectors like finance, energy, and manufacturing [13][14]. - The company has established three global R&D centers in Shenzhen, Beijing, and London, serving over 100,000 enterprises and institutions [14].

机器之心发布 机器之心编辑部 《大规模语言模型：从理论到实践（第 2版）》 是一本理论与实践并重的专业 技术书 ，更是 AI时代不可或缺的知识工具书。 任何人 都能在本 书中找到属于自己的成长路径。 在人工智能浪潮席卷全球的今天，大语言模型正以前所未有的速度推动着科技进步和产业变革。从 ChatGPT 到各类行业应用，LLM 不仅重塑 了人机交互的方式，更成为推动学术研究与产业创新的关键技术。 面对这一飞速演进的技术体系，如何系统理解其理论基础、掌握核心算法与工程实践，已成为每一位 AI 从业者、研究者、高校学子的必修课。 2023 年 9 月，复旦大学张奇、桂韬、郑锐、黄萱菁研究团队面向全球学术界与产业界正式发布了《大规模语言模型：从理论到实践》。短短 两年，大语言模型在理论研究、预训练方法、后训练技术及解释性等方面取得了重要进展。业界对大语言模型的研究更加深入，逐渐揭示出许多 与传统深度学习和自然语言处理范式不同的特点。例如， 大语言模型仅需 60 条数据就能学习并展现出强大的问题回答能力，显示了其惊人的 泛化性 。然而，本书作者们也发现大语言模型存在一定的脆弱性。例如， 在一个拥有 130 亿个参数的模 ...

Group 1: Duty-Free Policy - The Ministry of Commerce and five other departments have announced a notification to optimize the departure tax refund policy, lowering the refund threshold and enhancing the supply of refundable goods [1] - The long-term sales volume of departure tax refunds in China is expected to reach 109 billion yuan, driven by policy relaxation, technological penetration, and consumption upgrades [1] - By 2030, the market size for departure tax refunds may exceed 1 trillion yuan, forming a multi-tiered structure led by Hainan Free Trade Port and supported by first-tier cities [1] Group 2: Nuclear Power Development - The State Council has approved five nuclear power projects, including ten new units, marking the first approval of domestic nuclear power projects since 2025 [2] - The total investment for the newly approved nuclear units is estimated to exceed 200 billion yuan, with each unit requiring approximately 20 billion yuan [2] - China's nuclear power development strategy has shifted to "actively and safely develop," with a significant increase in the number of approved units expected in 2024 [2] Group 3: Intelligent Agents - The International Telecommunication Union has released a new international standard for intelligent agents based on large-scale pre-trained models, led by the China Academy of Information and Communications Technology [3] - The market for AI agents is projected to grow from 5.1 billion USD in 2024 to 47.1 billion USD by 2030, with a compound annual growth rate of 44.8% [3] - AI agents are becoming a key driver of transformation across various industries, enhancing efficiency and expanding human capabilities [3] Group 4: Virtual Power Plants - The Ministry of Transport and nine other departments have issued guidelines to promote the integration of transportation and energy, encouraging the development of virtual power plants [4] - Virtual power plants can effectively aggregate distributed energy resources and enhance power supply capabilities while promoting renewable energy consumption [4] - The rise of distributed energy resources necessitates the development of virtual power plants to facilitate grid regulation and market transactions [4] Group 5: AI in Government Affairs - The Central Political Bureau has emphasized the need for breakthroughs in AI development and regulation, focusing on foundational research and core technologies [5] - Shenzhen has successfully deployed an AI government assistant, improving overall work efficiency by over 20% across 240 real-world scenarios [5] - The AI-enabled government information system market is expected to grow rapidly, with significant potential for efficiency improvements and cost reductions [5]

智能体，正在吹皱AI江湖的一池春水。 从最初的简单算法模型到如今能够理解语言，进而模拟思维并做出决策，智能体在短短几年的时间里显示出令人瞠目结舌的进化能力，这也使得100年前 艾尔伯特·哈伯德的名言——"机器应该工作，人类应该思考"，越来越接近于现实。 智能体的每一次判断、推理、规划、优化和学习，都需要进行复杂的计算来实现。如果我们想让智能体真正能够像人类一样灵活理解语义、处理图像、学 习经验，就需要巨大的计算资源来支撑更大的模型和更多的数据。 智能体打破了人类能力的边界，但是与人类一样，复杂任务的完成也需要多个智能体的共同协作，这对算力的异构化、分布式和调度能力提出了更高的要 求……显然，在这个专业化日渐增长的领域里，谁能融汇贯通、让"数智化"实打实的应用落地，谁才可谓抓住了产业发展的主线。 4月16日，以"共建智能体时代"为主题的超聚变探索者大会2025，在河南郑州成功举办。 超聚变成立于2021年11月，距今仅仅三年有余。成立之初适逢全球算力产业格局的重构，业界对于这家新生企业的入局并不太为在意，孰料三年后的当 下，超聚变已经成为算力领域的"顶流"。 三年以来，超聚变从0到1、快速发展，以销售收入为例， ...

Core Viewpoint - The article emphasizes the importance of AI in transforming industries and highlights Lenovo's initiatives to facilitate a seamless transition to intelligent operations through innovative solutions and partnerships [2][14]. Group 1: AI Technology and Industry Transformation - By 2025, AI technology is expected to penetrate key industry scenarios, becoming a core engine for businesses to restructure production efficiency and business models [2]. - Lenovo aims to assist enterprises in achieving intelligent transformation with a focus on "scene-driven, security-first, and ease of use" through its AI workstation solution pre-installed with the DeepSeek model [2][14]. - The "36Kr AI Partner Conference" held in Shanghai gathered industry leaders and investors to discuss innovation paths and ecosystem collaboration in the era of AI [2][3]. Group 2: Lenovo's AI Solutions and Ecosystem - Lenovo's AI workstation supports AI inference applications from 7B to 70B models, enabling rapid deployment within 2 hours using NVIDIA's Ada series GPUs [6]. - The company offers three collaboration models for ecosystem partners: leveraging Lenovo's computing foundation, joint branding for marketing support, and integrating products into Lenovo's core sales system [6]. - Lenovo's AI integrated platform combines hardware and software solutions, ensuring a good user experience and addressing various enterprise needs [10]. Group 3: NVIDIA's Contributions and Technological Advancements - NVIDIA's next-generation Blackwell GPU technology introduces significant upgrades, including support for 8K and 96GB memory, enhancing performance for local deployment of inference models [8]. - The collaboration between Lenovo and NVIDIA focuses on optimizing performance, power consumption, and thermal management for AI applications [8]. Group 4: Investment Opportunities in AI - Lenovo Capital emphasizes the changing investment landscape in the AI era, identifying "silicon-code-data-intelligent agents" as core production materials [12]. - The shift from traditional software customization to AI-driven efficiency enhancement presents new opportunities in the B2B market [12]. - Lenovo Capital is particularly interested in AI-first projects and is willing to support teams developing vertical intelligent agents based on Lenovo workstations [12]. Group 5: Future Directions and Ecosystem Growth - Lenovo plans to strengthen its AI innovation ecosystem by collaborating with partners to build a comprehensive "computing-model-scenario" collaborative system [14]. - Future initiatives include launching upgraded solutions supporting the next-generation Blackwell architecture and introducing 500 ecosystem enterprises through joint development [14][16]. - Lenovo aims to empower businesses of all sizes and drive industry upgrades through AI, fostering a new ecosystem of mutual empowerment [16].

Group 1: Industry Overview - The domestic AI industry is experiencing a historic development opportunity amid escalating global trade tensions, with AI technology being a key driver for breaking technology barriers and enhancing self-sufficiency [1] - The evolution of large model capabilities is expected to significantly broaden AI application scenarios, shifting market focus from infrastructure to application value [1] - The necessity of establishing a foundational platform for intelligent agents is emphasized to achieve real-time understanding of user needs and efficient data analysis [1] Group 2: Company Focus - Saiyi Information (300687) is accelerating the integration of AI technology with vertical industries, exemplifying the trend of deep fusion between AI capabilities and specific business scenarios [2] - The company has developed an "AI + PCB manufacturing solution" leveraging its extensive experience in digital smart factory construction and Huawei's Ascend AI computing power [2] - Saiyi Information has initiated the "AI + supply chain" era, utilizing a dual-engine approach of "low-code + large model" to enhance supply chain intelligence through various technological steps [2] Group 3: Solutions and Impact - Saiyi Information has launched four intelligent agent scenario solutions focusing on order, planning, procurement, and logistics, addressing inefficiencies and enhancing supply chain transparency and resource scheduling [3] - The intelligentization of supply chains is becoming a critical competitive metric for enterprises, with Saiyi Information aiming to reconstruct supply chain value chains to achieve improvements in customer experience, cost efficiency, and risk control [3] - The demand for AI toolchains in the manufacturing sector is on the rise, marking a golden development period for AI application deployment in manufacturing software [3]

课程内容涵盖了 从智能体的基础概念到实际应用场景 的多个方面，包括但不限于： 基于知识库搭建 AI客服智能体 在近日， 未 可知人工智 能研究院与三节课携手合作 ，成功举办了主题为" 定制你的智能体 "的AI课程。 本次课程由未可知人工智能研究院资深讲师Jeffrey老师授课，吸引了众多对AI技术感兴趣的 学习者参与， 共同探索如何利用AI技术定制专属智能体，提升工作效率和业务能力。 课程亮点：深度解析智能体定制与应用 Jeffrey老师凭借其在人工智能领域的深厚背景和丰富经验，为学员们带来了一场精彩的知识 盛宴。 Jeffrey老师首先介绍了 如何基于知识库搭建AI客服智能体 。 他以三节课的会员权益为例，展示了如何通过知识库让智能体更好地理解用户需求并提供精准 服务。学员们了解到，通过准备详细的业务知识库，智能体可以实现与用户的高效互动，提升 用户体验和服务质量。 飞书多维表格批量分析运营周报 课程中， Jeffrey老师还分享了 如何利用飞书多维表格和AI工具批量分析运营周报。 他通过实际数据，展示了如何快速生成流量数据、转化率数据、销售数据等关键指标的分析报 告。 学员们学会了如何利用 AI工具提高数 ...