作者｜尹小明 编辑｜李忠良 策划｜AICon 全球人工智能开发与应用大会 在大模型技术飞速发展的当下，大数据领域的各类应用如雨后春笋般涌现，从数仓开发到 ChatBI 问数，再到深度分析 Agent，这些领域的大模型 应用极大地提升了数据处理和分析的效率。但与此同时，如何科学、准确地评估这些应用的效果，成为了行业面临的重要难题。 InfoQ 荣幸邀请到了字节跳动 / 数据平台大模型评测技术负责人 尹小明 在 AICon 全球人工智能开发与应用大会·深圳站上分享了《 评测也很酷 ——Agent 自动化评测技术创新与实践 》。作为字节跳动数据平台的大模型效果评估团队，他们深耕数据应用 Agent 领域，构建了覆盖从数据开 发到数据应用垂直领域 Agent 应用的评测技术体系，尤其在自动化评测算法、Agent 级评测框架等方面形成了可落地的技术方案。本次分享将聚焦 这一领域的技术细节与实践经验。 12 月 19～20 日的 AICon 北京站 将锚定行业前沿，聚焦大模型训练与推理、AI Agent、研发新范式与组织革新，邀您共同深入探讨：如何构建起可 信赖、可规模化、可商业化的 Agentic 操作系统，让 AI ...

Group 1: Investment Outlook - HubSpot Inc. (NYSE:HUBS) is considered one of the best falling stocks to buy, holding a Strong Buy consensus from 27 analysts, with 23 Buys, 4 Holds, and no Sells [1] - The average price target for HubSpot is $589.08, with a range from $450 to $800, indicating a potential upside of 59.71% from the current price of $368.84 [1] Group 2: AI Integration Strategy - HubSpot's CEO, Yamini Rangan, emphasized a hybrid monetization strategy focused on making artificial intelligence accessible to small and medium businesses during the Wells Fargo's 9th Annual TMT Summit [2] - The plan includes embedding AI technology across all product lines, such as Customer Agent and Data Agent, while utilizing extensive customer data to enhance product effectiveness [3] Group 3: Market Challenges - Rothschild Redburn downgraded HubSpot's stock to Neutral from Buy and reduced the price target from $610 to $450, citing concerns over the company's resilience amid potential AI disruption and expectations of slower net customer additions [4] - HubSpot provides a customer platform for marketing, sales, and customer service, with a core offering that includes a cloud-based Customer Relationship Management (CRM) tool [5]

来源：环球网 【环球网科技报道 记者 李文瑶】2025年，行业的聚光灯打在人工智能与大数据议题上。台上的讨论热 烈而宏大——大模型参数、算力竞赛、算法突破——这几乎是今年所有科技论坛的标准配置。 而在此之外，Aloudata大应科技创始人&CEO周卫林却对另一个更为"朴素"的问题着迷：当所有人追逐 模型的智能时，谁来解决数据的"脏活累活"？ 图：Aloudata大应科技创始人&CEO周卫林 "数据、算力、算法，这三要素中，数据是根本，但也是最苦最累的一环。"周卫林在接受采访时直言不 讳。这位从蚂蚁集团出来的技术创业者，选择了一条看似不那么性感的赛道——数据工程。 回归数据基础 在AI成为全民热潮的今天，周卫林的创业选择显得颇为"反常规"。他没有去做大模型应用，也没有追逐 算力基础设施，而是回到了最基础的"数据清洗、整合、管理"领域。 "这不是偶然，而是必然。"他分析道，"大模型爆发后，数据需求呈指数级增长，但数据供给方式还停 留在30年前——靠ETL工程师手工劳作。需求增长百分之几百，人力不可能同步跟上。" 他提出了一个比喻："这就好比在电动车时代，你还在用马拉车的动力系统。" 2021年，周卫林带着团队独立 ...

HubSpot FY Conference Summary Company Overview - **Company**: HubSpot (NYSE:HUBS) - **Event**: Ninth Annual Wells Fargo TMT Conference - **Date**: November 19, 2025 Key Points Industry Context - The software industry is currently facing challenges but presents significant opportunities, particularly in AI integration [8][9][10] AI Strategy - HubSpot's AI strategy focuses on embedding AI into all products to assist small and medium businesses (SMBs) in growth [8][9] - Three featured AI agents were launched: Customer Agent, Prospecting Agent, and Data Agent, along with a Breeze Assistant for go-to-market employees [9][10] - The differentiation of HubSpot's AI offerings lies in its deep understanding of SMBs and the context it provides for sales and marketing activities [10][11] Customer Base and Adoption - HubSpot has 280,000 customers utilizing its platform for marketing, sales, and service, which enhances AI effectiveness through contextual data [11] - AI adoption is driven by the presence of C-suite leaders prioritizing AI initiatives within companies, rather than company size [58][59] Data Management - The transition to an agentic architecture allows HubSpot to handle unstructured data, enhancing the context and quality of AI outputs [22][23] - Data Hub was rebranded to emphasize its role in improving data quality and facilitating AI workflows [66][67] Monetization Strategy - HubSpot employs a hybrid monetization strategy for AI, charging for both seats and credits based on the type of AI usage [48][49] - The Customer Agent is the leading consumer of credits, resolving over 60% of support tickets for more than 6,000 customers [49][50] Marketing and SEO Adaptation - HubSpot has diversified its marketing channels in response to changes in lead generation, focusing on platforms like YouTube and podcasts [94][95] - The concept of Answer Engine Optimization (AEO) is emerging as a new strategy to capture leads through AI-driven responses [95][96] Future Outlook - The adoption curve for AI technologies is expected to improve as customers gain trust in data quality and the effectiveness of AI solutions [102][103] - HubSpot aims to continue enhancing its AI capabilities and providing value to its customers through iterative feedback and improvements [102][103] Conclusion - HubSpot is well-positioned to leverage its extensive customer base, data management capabilities, and AI integration to drive growth and enhance customer experiences in the evolving software landscape [72][73]

Financial Data and Key Metrics Changes - Revenue grew 18.4% year over year in constant currency, reaching $810 million, with an operating margin of 20% [4][18] - Subscription revenue increased by 21% year over year, while services and other revenue rose by 19% [18] - Net income was $140 million in Q3, translating to $2.66 per fully diluted share, with free cash flow at $147 million, or 18% of revenue [19] - Customer dollar retention remained in the high 80s, and net revenue retention was flat at 103% [18][19] - Calculated billings were $804 million, growing 19% year over year in constant currency [19] Business Line Data and Key Metrics Changes - Total customers increased by 10,900 in Q3, bringing the total to 279,000, a 17% year-over-year growth [18] - Multi-hub adoption is evident, with 43% of Pro Plus customers subscribing to three core hubs, up 4% year over year [5] - Deals over $5,000 in monthly recurring revenue grew 35% year over year, indicating strong up-market momentum [6] Market Data and Key Metrics Changes - Domestic revenue grew 17% year over year, while international revenue growth was 20% in constant currency [18] - International revenue represented 49% of total revenue, highlighting the company's global reach [18] Company Strategy and Development Direction - The company is focused on platform consolidation, multi-hub adoption, and up-market momentum as key growth drivers [5][15] - AI innovation is central to the company's strategy, with new features launched across all hubs to enhance customer experience [8][9] - The introduction of the Loop playbook aims to guide customers in leveraging AI for growth, with significant engagement metrics reported [12][13] Management's Comments on Operating Environment and Future Outlook - Management expressed confidence in the company's ability to achieve durable growth, emphasizing the strength of core growth drivers and emerging levers [25][17] - The company anticipates a step-up in net revenue retention in Q4, with expectations for a couple of points improvement for the full year of 2025 [18][19] - Management highlighted the importance of AI as a multi-year tailwind for growth, with ongoing investments in product innovation [24][17] Other Important Information - The company repurchased 780,000 shares of common stock under its share repurchase program, representing $375 million [19] - The acquisition of XFunnel aims to enhance brand visibility in AI-generated answers, integrating it into HubSpot's platform [14] Q&A Session Summary Question: Path to 20% Growth - Management believes it can grow faster than current levels, focusing on durable growth through strong core drivers and emerging growth levers [22][23] Question: Impact of AI Overviews - Management sees AI overviews as a significant opportunity for customers to enhance visibility and engagement through new marketing strategies [26][27] Question: Platform Consolidation Drivers - Customers cite total cost of ownership, unified data access, and AI adoption as key reasons for consolidating on HubSpot's platform [29][30] Question: Billings Growth and Revenue Inflection - Management noted that billings growth is influenced by various factors, including mix shifts and duration, and expects revenue growth to follow net new ARR trends [31][32] Question: Customer Adoption of AI Products - The company is seeing strong adoption of AI features, with positive trends in credit consumption and customer engagement [38][39] Question: ASP Growth Challenges - ASP growth has been impacted by new customer pricing and robust starter additions, but there are tailwinds from large deals and multi-hub adoption [45] Question: Customer Data Quality - The state of customer data varies, with higher quality observed among those using multiple hubs, and the introduction of Data Hub aims to improve data quality [52][54]

Core Viewpoint - The implementation of Data Agents appears straightforward but is fraught with challenges, primarily due to software engineering difficulties. A unified semantic layer is crucial for success, and neglecting aspects like scenario focus, iterative technical frameworks, or semantic models can lead to stagnation in prototype stages [2][6][12]. Group 1: Importance of Semantic Layer - The significance of building a semantic layer for Data Agents is widely recognized, with both domestic and international investments increasing in this area. Tencent Cloud WeData has been an early investor in this domain [7][12]. - The semantic layer encompasses four main aspects: concepts, data relationships, metrics, and dimensions, which are essential for providing accurate and unified data access interfaces for Agents [8][12]. Group 2: Technical Challenges and Solutions - The primary technical challenges in integrating Data Agents into existing enterprise platforms include data governance issues and the difficulty in evaluating the effectiveness of Data Agents [14][15]. - To address these challenges, a focus on specific scenarios for unified semantic layer construction and evaluation systems is recommended [15][18]. Group 3: Future of Data Roles - Data Agents are not expected to replace data engineers or scientists but will automate some execution tasks. This will lead to a fusion of roles, requiring professionals to possess a broader skill set related to Agents and large language models (LLMs) [10][11]. - Understanding the basic principles of Agents and LLMs is essential for effectively utilizing large model technologies [11]. Group 4: Recommendations for Enterprises - Companies are advised to focus on scenario-specific semantic abstraction and address existing data governance issues to build a robust semantic layer [16][17]. - It is crucial to establish an iterative technical framework and a comprehensive Agent operation system to monitor, evaluate, and modify the Data Agent effectively [18].

Core Insights - The article discusses the upgrades and new features of Volcano Engine's AI tools, including PromptPilot and Data Agent, aimed at enhancing AI application efficiency and data utilization for enterprises [1][2][4]. Group 1: PromptPilot Upgrade - PromptPilot has been upgraded to support prompt optimization for any model, including public cloud models, private models, and custom-trained models [2][3]. - The tool utilizes natural language interaction to understand user needs, extract evaluation criteria, and generate improved prompts, thereby continuously optimizing based on online traffic and bad case analysis [2][3]. - The integration with Volcano Engine's knowledge base allows for precise content retrieval, enhancing the model's understanding and output in specialized fields [3]. Group 2: Data Agent and One-Customer-One-Strategy - Data Agent is a vertical intelligence tool that deeply understands and utilizes enterprise data assets, enabling proactive analysis and action [5]. - The "One-Customer-One-Strategy" feature allows for personalized marketing by analyzing multi-dimensional data related to customers, internal knowledge bases, and public data, generating tailored marketing plans [5][6]. - This feature has shown significant results, with conversion efficiency from Marketing Qualified Leads (MQL) to Sales Qualified Leads (SQL) increasing by up to 300%, and data utilization rising from 10% to 95% [6]. Group 3: AI Data Lake Service and AI Operator Square - The AI Data Lake service has been enhanced with the launch of "AI Operator Square," which facilitates the management of multi-modal data, including text, images, and audio-visual content [7]. - The AI Operator Square provides over 100 standardized operators for various data processing tasks, allowing users to create modular workflows through a visual drag-and-drop interface [8][9]. - This upgrade aims to transform scattered data into knowledge assets, promoting automated circulation and value addition of knowledge assets [9].

Core Insights - The article discusses the upgrades of Volcano Engine's PromptPilot and Data Agent, which enhance AI applications and data management for enterprises. Group 1: PromptPilot Upgrade - PromptPilot has been upgraded to support prompt optimization for any model, including public cloud models, private models, and custom-trained models [2][3] - The tool utilizes natural language interaction to understand user needs, extract evaluation criteria, and generate better prompts, thus improving AI application performance [2][3] - After deployment, PromptPilot can sample online traffic, analyze bad cases, and autonomously optimize prompts, creating a cycle of continuous improvement [2][3] Group 2: Data Agent and Multi-modal Data Lake - Data Agent is introduced as a vertical intelligent agent that deeply understands and utilizes enterprise data assets, enabling proactive analysis and action [4] - The "One Customer, One Strategy" capability of the intelligent marketing agent integrates three types of data for precise customer profiling and targeted marketing strategies [4][5] - The effectiveness of "One Customer, One Strategy" includes a 300% increase in conversion efficiency from MQL to SQL, a rise in data utilization from 10% to 95%, and a reduction in customer analysis time from 30 minutes to 2 minutes [5] Group 3: AI Operator Square - The AI Data Lake service has launched the "AI Operator Square," which integrates management of multi-modal data, including text, images, and audio-visual content [5][6] - The platform offers over 100 standardized operators and supports the integration of mainstream open-source operators, providing a comprehensive framework for custom operator development [6] - Users can visually drag and drop to quickly assemble modular workflows, transforming scattered data into knowledge assets for automated circulation and value addition [7]

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].

Group 1 - The core viewpoint is that the commercialization of AI technology is steadily advancing, with 2025 expected to be the year of large-scale commercial deployment of AI Agents [1] - ByteDance has upgraded multiple models and launched the all-scenario intelligent agent Data Agent, enhancing its capabilities in video and music generation [1] - The report emphasizes that AI technology iteration and the deployment of Agents are progressing across major companies, indicating ongoing development in AI commercialization [1] Group 2 - The China Securities Regulatory Commission (CSRC) has revised the Major Asset Restructuring Management Measures, aiming to deepen the reform of the mergers and acquisitions market [2] - Key changes include establishing a phased payment mechanism for restructuring shares and increasing regulatory tolerance for financial condition changes and related transactions [2] - The CSRC's actions are expected to invigorate the mergers and acquisitions market and accelerate the integration of segments driven by digital transformation in the computing sector [2] Group 3 - Google DeepMind has launched the general-purpose AI system AlphaEvolve, which autonomously generates and improves algorithmic code [3] - AlphaEvolve has successfully addressed significant challenges in mathematics and computer science, demonstrating its potential in enhancing AI chip design and optimizing global computing resource utilization [3] - The consensus among major tech companies is that the deployment of Agents is becoming a reality, with expectations for accelerated future development [3]