作者 | Tina Data Agent 看似轻松上手，但真正落地却充满挑战。虎兴龙在专访中指出，90% 的难点源于软件工程，而统一语义层建设是成功的关键。企业如果忽 略场景聚焦、技术框架的可迭代能力或语义模型和运营体系，即便投入几个月，也可能无法监控、评估或修改，最终停留在原型阶段。掌握统一语义 层、完善的技术框架和运营体系，才能让 AI 代理真正理解数据、快速迭代、落地应用，显著提升企业数据智能化效率。 采访嘉宾： InfoQ：过去 BI 系统很多时候是"看数"，如今进化到从"看"到"做"，从系统工程角度看，这背后意味着什么样的技术支撑？ 虎兴龙： Agent 的交互形态可以比作是数据分析的新"head" ，新 Head 一定需要新的技术支撑。我认为必须必要的两方面技术革新：1、需要有数据语 义工程的平台化能力，数据工程、数据平台不止交付数据内容，还应该交付可被分析的语义。需要有数据语义层为 Data Agent 提供可靠的知识、高性 能的数据操作接口；2、需要有完善的 Agent Ops 平台基础，目前 Agent 开发框架发展很快，但是 Agent 的运营体系其实很关键并且是存在不足的， Agent ...

Core Insights - The article emphasizes the transformative role of Data Agents in the era of Generative AI, highlighting their ability to compress the data lifecycle into a rapid "data → insight → action" loop, achieving over 60% efficiency gains and significant cost savings in the millions of dollars [3][4][10]. Industry Trends - Data Agents redefine "Data" as any digital asset that can be accessed and utilized in real-time, moving away from traditional static databases [5][7]. - The global data volume is projected to reach 149 ZB in 2024 and exceed 181 ZB in 2025, with approximately 80% being unstructured data that requires immediate structuring for algorithmic use [5][7]. - Generative AI is expected to contribute an additional $2.6 to $4.4 trillion in value annually, with nearly 75% of this value coming from functions heavily reliant on structured data [5][7]. Data Agent Definition and Functionality - Data Agents are AI entities that automate the entire data lifecycle, capable of planning, executing, and verifying tasks based on natural language inputs [7][8]. - They are positioned as core infrastructure rather than mere BI tools, directly impacting business KPIs and productivity [7][8]. Efficiency Gains and Market Acceptance - Early adopters of Data Agents have reported productivity increases of over 60% and annual savings of millions of dollars [7][8]. - The cost of LLM inference has dramatically decreased from $60 per million tokens to $0.06, indicating a significant technological shift [10][13]. - AI search and query traffic in the U.S. has reached 5.6%, reflecting a growing acceptance of natural language interactions for structured answers [13][14]. Market Demand and Investment Trends - The demand for Data Agents has surged, with a 900% increase in global search interest for "AI agent" and a tripling of investment in the AI Agent sector, reaching $3.8 billion in 2024 [45][46]. - Major acquisitions by companies like Databricks and Snowflake indicate a strong focus on data-driven AI platforms [13][14]. Development Stages of Data Agents - The evolution of Data Agents is expected to occur in three stages: 1. Human-led with AI empowerment, transforming data interaction and decision-making processes [36][37]. 2. Scenario-driven applications that allow for rapid development of customized systems based on existing data [38][40]. 3. Autonomous intelligence where Data Agents manage data collection, governance, and analysis, acting as a digital COO [41][42]. Conclusion and Future Outlook - The current landscape presents a unique opportunity for Data Agents to become the default interface for digital work, akin to the Office suite in the 1990s [45][46]. - The integration of Data Agents into business processes is anticipated to enhance organizational efficiency and responsiveness, marking a significant shift in how data is utilized across industries [48][49].

Investment Rating - The report maintains an "Increase" rating for the industry [7] Core Insights - Major players like Alibaba and Tencent are significantly increasing their capital expenditures (Capex) for AI infrastructure, indicating a positive outlook for the industry [12][16] - The demand for high-performance computing is rapidly increasing, driven by AI applications, which is expected to further expand cloud computing needs [12][16] - The report emphasizes that computing power is a critical infrastructure for the development of AI agents, which will support long-term growth in the industry [42][51] Summary by Sections Capital Expenditure Growth - Alibaba's Capex for Q1 2025 reached 24.612 billion RMB, a year-on-year increase of 120.68%, with cloud revenue of 30.127 billion RMB, up 17.71% [13][16] - Tencent's Capex for Q1 2025 was 27.476 billion RMB, a 91.35% increase from 14.4 billion RMB in Q1 2024 [16][19] AI Application Acceleration - Major cloud providers are enhancing their capabilities to accelerate AI application deployment, with significant upgrades announced at various conferences [21][26] - Alibaba Cloud's ninth-generation ECS has improved computing power by up to 20% while reducing prices by 5% [28][30] - Huawei Cloud introduced the CloudMatrix 384 super node, designed to meet the massive computing demands of the AI era [36][39] Computing Power as a Key Driver - The report identifies several reasons for the high demand for computing power in AI agents, including the need for long context processing, external data integration, and complex task verification [42][51] - The increasing complexity of AI models and the need for high concurrency access further exacerbate the demand for computing resources [51] Investment Opportunities - The report suggests focusing on companies involved in computing power such as Cambricon, Alibaba, and Inspur, as well as those in the AI agent space like Kingsoft Office and Kingdee International [4][53][54]

编辑：杨文 字节拿出了国际顶尖水平的视觉–语言多模态大模型。 5 月 13 日，火山引擎在上海搞了场 FORCE LINK AI 创新巡展，一股脑发布了 5 款模型和产品，包括豆包・视频生成模型 Seedance 1.0 lite、升级后的豆包 1.5・视 觉深度思考模型，以及新版豆包・音乐模型。同时，Data Agent 和 Trae 等产品也有了新进展。 其中，全场最吸睛的就是豆包 1.5・视觉深度思考模型（以下称 Seed 1.5-VL）。 相比于之前版本，Seed1.5-VL 具备更强的通用多模态理解和推理能力，不仅视觉定位和推理更快更准，还新增了视频理解、多模态智能体能力。 举个例子。仅需一张图，再来个提示词，Seed1.5-VL 就能精准识别观众、棒球、座椅、围栏等多种元素，还能正确分类并给出坐标。 再比如，喂它一段监控，问一句：「今天小猫干了哪些坏事？」几秒钟后，它就丢过来几段视频，「抓包」了小猫遛弯、玩球、巡视、搞破坏等各种「作案」现 场。 机器之心报道 尽管 Seed1.5-VL 的激活参数仅有 20B，但其性能可达到与 Gemini2.5 Pro 相当的水平，在 60 个公开评测基准中，有 ...

Core Insights - ByteDance's Volcano Engine is focusing on practical and specialized large model products, moving away from grand innovations to more incremental improvements in 2023 [1][2] - The newly launched Seedance 1.0 lite video generation model emphasizes small size, cost-effectiveness, and high-quality output, supporting video generation of 5s and 10s at resolutions of 480P and 720P [1][3] - The Doubao 1.5 Thinking Vision Model has a parameter size of only 20 billion but excels in multimodal understanding and reasoning, achieving top performance in 38 out of 60 public evaluation benchmarks [3][4] Product Features - The Seedance 1.0 lite model allows for precise control over video generation, including character expressions and clothing, enhancing its application in e-commerce advertising and entertainment [2][3] - The Doubao 1.5 model introduces GUI Agent capabilities, enabling complex interactions across different platforms, such as automated testing of new app features [3][4] AI Ecosystem Layout - Volcano Engine has established a broad AI ecosystem, impacting various industries including automotive, finance, education, and retail, with coverage of 4 billion devices and partnerships with major banks and universities [4][6] - The introduction of Data Agent aims to help enterprises unlock data asset value through intelligent analysis and marketing [4][6] - The upgrade of the AI-native IDE product Trae allows developers to utilize AI more efficiently, with the integration of the Model Context Protocol (MCP) for external tool invocation [4][5]

很多人低估了 Manus 的影响力。 就像 DeepSeek 降低了大模型商业化门槛、引领新一波 AI 创业一样，Manus 让大众真正看到了 Agent 的重要性和可能性。 「看到」本身比任何观点、任何演示都有说服力。 看到了，才有落地的意愿。 但企业内落地不能靠对业务缺乏了解的通用 Agent，需要垂直领域，真正懂场景的 Agent。火山引擎 4 月份发布的 Data Agent 就是垂直领域 Agent 的 代表。 Founder Park 在第一时间进行了深度试用，体验后还和项目团队进行了交流，试图理清楚 Data Agent 在当下落地的现状和难点，以及未来的想象力有 多大。 点击「阅读原文」，可申请【Data Agent 样板间】体验。 进群后，你有机会得到： 01 企业内的数据难题 企业内如何利用数据这件事，一直都没有被很好地解决。 这些问题都有望在 Data Agent 里得到解决：统一的数据管理平台、兼容非格式数据、自然语言查询数据，以及让数据本身从单纯的查询功能向解释功 能、预测功能转变。 Founder Park 正在搭建「 AI 产品市集」社群，邀请从业者、开发人员和创业者，扫码加群： ...