马化腾资料 图。本文来源：澎湃新闻 记者 范佳来 马化腾官宣：自研龙虾、本地虾、云端虾等陆续赶 来，外媒曝还在开发智能体 腾讯即将成为"养虾"大户？ 3月11日，澎湃新闻记者获悉，腾讯CEO马化腾在朋友圈转发腾讯全系"龙虾"矩阵产品相关内 容，并介绍：自研龙虾、本地虾、云端虾、企业虾、云桌面虾，安全隔离虾房、云保安、知识 库… 还有一批产品陆续赶来。 在个人用户层面，腾讯主打"零部署、开箱即用"的轻量化体验解决方案。此次新上线的自研全 场景桌面智能体WorkBuddy，能够完全兼容OpenClaw技能包，用户无需具备代码基础与环境配 置经验，即可跨软件执行自动化任务。 同日，腾讯基于OpenClaw开源生态打造的本地AI助手QClaw也正式开启内测，无需服务器、无 需额外配置IM工具，即可在个人电脑上一键部署"小龙虾"。 3月9日，腾讯宣布，全场景AI智能体WorkBuddy正式上线，该产品能力与"小龙虾"OpenClaw类 似，完全兼容"小龙虾"的功能且更易用，安装过程类似于普通的App安装，无需复杂部署，可 以调用国内不同的模型。 当天，腾讯系编程工具CodeBuddy发布致歉信称，因流量激增出现登录及服 ...

Core Viewpoint - Tencent is expanding its product ecosystem in the AI domain, particularly focusing on the "OpenClaw" (referred to as "小龙虾") initiative, which aims to enhance user experience through various AI-driven solutions [1][3]. Group 1: Product Development - Tencent's CEO, Ma Huateng, shared updates on the company's "dragon shrimp" product matrix, which includes self-developed, local, cloud-based, and enterprise solutions [1]. - The newly launched WorkBuddy is a full-scene AI agent that is compatible with OpenClaw's skill set, offering a user-friendly installation process similar to standard app installations [4]. - The local AI assistant QClaw, based on the OpenClaw ecosystem, has entered internal testing, allowing users to deploy it on personal computers without server requirements [3]. Group 2: Market Response - Following the launch of WorkBuddy, Tencent's stock price saw a notable increase, rising by 7.13% to 578 HKD per share from the previous closing price of 519 HKD [4]. - The surge in user traffic after the public testing of WorkBuddy led to service instability for Tencent's CodeBuddy, prompting the technical team to expand capacity by tenfold [4].

Core Viewpoint - The article discusses various developments in the tech and investment sectors, highlighting significant company announcements, financial results, and market trends. Group 1: Company Announcements - Tencent's CodeBuddy faced service instability due to unexpected user traffic from the launch of WorkBuddy, leading to a 10-fold capacity expansion to restore services [8][10] - Meta clarified rumors regarding Alexander Wang's departure, stating he remains influential within the company [5] - NIO reported Q4 revenue of 34.65 billion yuan, a 75.9% year-on-year increase, and achieved its first quarterly profit [11] - Honor plans to launch a sub-brand in the domestic market to optimize its channel strategy [21][22] - Baidu's autonomous driving service,萝卜快跑, has resumed full testing and operations in Dubai and Abu Dhabi [30] Group 2: Financial Results and Market Trends - NIO's full-year revenue for 2025 reached 87.49 billion yuan, a 33.1% increase, with a total of 326,028 vehicles delivered [11] - The Southeast Asian smartphone market is projected to see a 1% decline in shipments in 2025, with Samsung leading the market [40] - Apple increased its iPhone production in India by approximately 53%, with Indian production accounting for 25% of its total output [40] - The Chinese wearable device market is expected to grow by 20.8% in 2025, with smartwatches and wristbands contributing to this growth [41] Group 3: Investment and Financing - Advanced Machine Intelligence Labs, founded by former Meta AI chief Yann LeCun, raised $1.03 billion in seed funding, marking the largest seed round in Europe [31] - Lingchu Intelligent completed a total of 2 billion yuan in angel and Pre-A financing to enhance logistics applications [31] - Dify.AI secured $30 million in Pre-A funding to support AI application development [31] - OpenAI announced plans to acquire Promptfoo, focusing on AI system security [31]

Core Viewpoint - Tencent has launched its all-scenario AI assistant WorkBuddy, which simplifies remote operations for users by eliminating complex cloud deployment processes, making it more user-friendly and secure compared to OpenClaw [1][2]. Group 1: Product Features and Usability - WorkBuddy is compatible with OpenClaw's functionalities and allows users to connect via WeChat for quick remote operations, with setup taking as little as one minute [2][5]. - The product addresses the challenges faced by users with OpenClaw, such as high installation costs and security vulnerabilities associated with cloud deployment [4]. - WorkBuddy can perform complex tasks like batch renaming files, processing invoices, creating presentations, and managing social media content, showcasing its capability to enhance productivity [7]. Group 2: Market Demand and Trends - The popularity of OpenClaw indicates a strong demand among professionals for AI products that integrate into workflows and solve real productivity issues, moving beyond simple chat functionalities [6][7]. - The launch of WorkBuddy is expected to popularize the "养龙虾" (raising lobsters) smart office model, further embedding AI into everyday work environments and making it more accessible to the general public [8]. Group 3: Future Developments - Tencent plans to release an AI-assisted programming tool, CodeBuddy, in 2025, which will allow users to program through conversation, indicating a broader strategy to enhance AI capabilities in various domains [7]. - WorkBuddy is built on the same Agent architecture as CodeBuddy, positioning it as a scalable and extensible desktop AI platform rather than just a simple chat interface [7].

GAPO团队 投稿 量子位 | 公众号 QbitAI 程序员们又能少掉头发了！ 新研究通过过滤掉训练中的噪声和异常值，显著提升代码大模型在实际编辑任务中的准确性和效率。 在AI辅助编程成为软件开发核心生产力的今天，大语言模型 （LLMs） 已深度融入代码编辑、调试与优化全流程。 然而，当企业试图用 真实复杂用户环境中采集的数据 开展强化学习 （RL） 训练时，一个棘手的实际问题浮出水面：复杂上下文 （context） 导致大模型的输出答案频繁出现异常内容，即rollout噪声更普遍，使得reward出现异常值 （outliers） ，直接造成优势值 （advantage） 估计不准确，严重拖累强化学习效果。 上海交通大学、腾讯CodeBuddy等团队联合提出的 Group Adaptive Policy Optimization（GAPO） 方法，精准直击这一产业落地关键 瓶颈，为代码LLM的工业化训练提供了兼具科研创新性与工程实用性的突破方案，引发AI科研界与产业界广泛关注。 真实场景的核心梗阻：复杂上下文→rollout噪声→优势估计失真 代码编辑的核心难点在于，真实用户场景的输入提示绝非简单的代码片段， ...

Core Viewpoint - The article discusses the rise of "Vibe Coding," a concept that democratizes programming by allowing individuals with little to no coding experience to create applications using AI tools, thus reshaping the landscape of technology and entrepreneurship [4][5]. Group 1: Vibe Coding and Its Impact - Vibe Coding, introduced by Andrej Karpathy, allows users to develop applications without deep coding knowledge, making it accessible to a broader audience, including children and non-technical individuals [4][5]. - The popularity of Vibe Coding has led to a surge in AI programming tools, with companies like Baidu and Tencent reporting significant portions of their code being generated or assisted by AI [11][12]. - The article highlights various success stories of individuals using Vibe Coding, such as a student who earns substantial income by leveraging AI tools for development and sharing accounts on platforms like Xianyu [19][22]. Group 2: Entrepreneurial Opportunities - The rise of Vibe Coding is seen as beneficial for "one-person companies," enabling individuals to start businesses with minimal resources and technical skills [36][39]. - Success stories include a programmer who founded a Vibe Coding company and was later acquired for a significant sum, illustrating the potential for high returns in this new landscape [37]. - However, the article also notes the challenges faced by solo entrepreneurs, such as customer service demands and the need for unique value propositions to stand out in a crowded market [40][39]. Group 3: Demographics and Perspectives - The article features a diverse range of users, from young students to middle-aged professionals, all finding value in Vibe Coding for personal and professional development [32][43]. - It emphasizes that while technical skills are becoming less critical, creativity, business insight, and resource integration remain essential for success in the AI-driven economy [45]. - The fast-paced nature of the AI industry requires continuous learning and adaptation, as many individuals are actively engaged in sharing knowledge and experiences late into the night [46].

【导读】华为云码道代码智能体已开启公测，AI编程领域再添国产强手 AI智能编程领域，迎来重磅消息！ 2月11日，中国基金报记者获悉，华为云码道（CodeArts）代码智能体，已在马年春节前夕开启公测，参与公测人数已超1万人。 华为云码道代码智能体定位于实干派AI编程工具，其核心是高效替代高频重复的工程化开发工作，从而大幅度降低编程门槛，有助于加速AI编码技术的 产业化落地。 无编程基础者通过自然语音输入需求，即可借助华为云码道代码智能体搭建小程序等应用。 据悉，华为云码道代码智能体提供代码生成、研发知识问答、单元测试用例生成、代码解释、代码注释、代码调试、代码翻译、代码检查、代码优化等场 景功能。 业内人士透露，华为云码道代码智能体在项目级代码生成等关键场景实现突破，可适配企业研发需求，助力压缩开发周期、提升编码效率。 与华为云码道代码智能体同类的AI编程工具包括Cursor、GitHub Copilot等海外产品，以及字节跳动的Trae、腾讯云的CodeBuddy等国产产品。 华为云码道代码智能体的核心竞争力来自华为数十年积淀的高质量研发数据，整合了代码大模型、AI IDE（集成开发环境）、自主开发模式、代 ...

以下文章来源于20社 ，作者王晓玲 贾阳 20社 . 关注年轻人的钱包、工作和生活。 来源丨 20社 （ quancaijing_20she ） 作者丨王晓玲 贾阳 图源丨 pixabay 就在跨年之际，马斯克的"旧世界1000天倒计时"宣言，又给AI恐慌狠狠地加了一把柴。 巨头和寓言，离普通人太远，也只是AI时代水面上的冰山一角。我们关注的是普通人能够复制的成功 经验。 最近，我们找到了4位"吃螃蟹的人"，他们的共同点是，不太专业，甚至都是文科生。但是，都抓住 了AI带来的时代红利。 AI的爆发对王博源来说，更多的不是"被替代"的焦虑，而是终于等到了能够让他"创造宇宙"工具的兴 奋。在2025年最后一天，这个完全没有IT背景的38岁前媒体记者，自己用AI做出来的杜比全景声音 乐APP"全景新声"上线App Store的TestFlight 。 根本不会画画的连续创业者，用AI创作了品质相当不错的职场讽刺短片，创造出了深受网友认可 的"牛马"IP。 还有成熟的商业广告导演，借助AI，让作品更具特色，成为AI创业公司喜欢的广告合作商。但他也会 在深夜悲伤，因为他引以为傲的、几十年训练出来的创作风格，"在未来的A ...

Core Viewpoint - The competition among cloud service providers in China, particularly in the AI cloud sector, is intensifying, with companies like Alibaba Cloud, Volcano Engine, and Tencent Cloud adopting different strategies to capture market share and achieve profitability [1][6][26]. Group 1: Advertising and Market Position - Since 2025, the number of AI cloud-related advertisements in major airports has significantly increased, with major players like Alibaba Cloud and Volcano Engine dominating the advertising space [1]. - In Shenzhen's Bao'an Airport, Alibaba Cloud claims to lead the AI cloud market share, while Volcano Engine asserts a 46.4% share of the public cloud large model market [2][5]. - Tencent Cloud's absence from the advertising scene is notable, especially as AI is seen as a critical growth area for cloud computing [5]. Group 2: Profitability vs. Scale - Tencent Cloud achieved overall profitability in its cloud business by 2025, a rare feat in an industry plagued by a "profit black hole" due to prolonged price wars [6][8]. - The industry has seen a trend where higher revenues often correlate with greater losses, with Alibaba Cloud's latest report showing a profit margin of less than 9% [8]. - Tencent Cloud has shifted its focus from absolute revenue to sustainable income structures, leading to a reduction in low-profit business segments [10]. Group 3: Business Strategies - Tencent Cloud's strategy emphasizes high-quality growth and a focus on high-value clients, resulting in a decrease in market share but a healthier revenue model [10][12]. - Volcano Engine's approach involves aggressive pricing for Token sales to attract AI customers, but this model has been criticized for lacking customer retention [14][15]. - Tencent Cloud's business structure is differentiated, with a focus on 40% IaaS, 40% PaaS, and 20% SaaS, highlighting its commitment to AI SaaS products [17]. Group 4: Market Growth and Competition - The AI cloud market in China reached 22.3 billion yuan in the first half of 2025, with Alibaba Cloud holding a 35.8% market share, followed by Volcano Engine at 14.8% [28]. - Tencent Cloud's revenue from partnerships has shown double-digit growth, with AI and SaaS product orders doubling [24]. - The competition is not just about market share but also about long-term sustainability and the ability to adapt to industry changes [26]. Group 5: Ecosystem and Partnerships - Tencent Cloud emphasizes a partnership-first approach, valuing long-term relationships with partners over short-term gains [24]. - The company has collaborated with over 11,000 partners, maintaining a focus on mutual respect and shared benefits [24]. - The competitive landscape has seen some cloud providers exploit partnerships, leading to a breakdown in trust and collaboration [21][22].

Summary of the Conference Call on the Chinese Internet Industry Industry Overview - The report focuses on the **Chinese Internet Industry** and its outlook for **2026**, highlighting the potential activation year for consumer-level AI, particularly chatbots transitioning from novelty to habitual use [1][5]. Key Insights and Arguments - **Market Readiness**: The market is not yet prepared to consistently price "AI application winners and losers." Early market share shifts reflect distribution routes, marketing rhythms, and product iteration speeds rather than solid competitive advantages [4][8]. - **Investment Strategy**: The recommended investment strategy is to avoid binary bets on application-level winners and instead target beneficiaries of AI integration in industries, focusing on broader adoption rather than individual application market shares [4][8]. - **AI Infrastructure Growth**: The growth in daily multi-turn chatbot interactions is expected to boost demand for reasoning workloads and token throughput, benefiting companies like **Alibaba** and **Baidu** [4][8]. - **Advertising Platforms**: Increased marketing spending to cultivate chatbot user bases is anticipated to benefit platforms like **Kuaishou Technology**, as they compete for chatbot mindshare [4][8]. - **Gaming Sector**: The gaming industry is expected to benefit from AI adoption, with companies like **Tencent** and **NetEase** leveraging AI to accelerate content creation and improve return on investment [4][8]. Additional Important Points - **Chatbot Adoption**: The rise of chatbots is seen as a significant shift in consumer behavior, with **ByteDance's Doubao** reaching approximately 100 million daily active users, indicating a milestone in mainstream application [5][7]. - **Global Context**: The report draws parallels with global trends, noting that once a reliable adoption path is established, Chinese platforms are unlikely to underinvest in consumer-level AI [5][7]. - **AI as a Business Tool**: Companies are increasingly productizing AI as enterprise toolkits rather than merely as R&D capabilities, with Tencent's **CodeBuddy** reducing coding time by 40% [11]. - **Regulatory Environment**: The regulatory landscape is seen as favorable for content supply, with a stable approval rhythm for game licenses, indicating a healthy content reserve for 2026 [19]. - **Valuation Dynamics**: The report emphasizes that valuation multiples are being driven by narrative changes, particularly regarding AI capabilities, rather than immediate profit forecasts [24][28]. Company-Specific Insights - **Core Stocks**: The preferred stocks in the industry include **Baidu**, **Alibaba**, **Tencent**, **NetEase**, and **Kuaishou Technology** due to their strong positioning in the AI landscape [1][4]. - **Valuation Comparisons**: The report highlights significant valuation discounts for leading Chinese internet platforms compared to their U.S. counterparts, suggesting potential for narrowing these gaps as narratives around AI and platform indispensability evolve [32][33]. This summary encapsulates the key points from the conference call, providing a comprehensive overview of the current state and future outlook of the Chinese internet industry, particularly in relation to AI adoption and its implications for investment strategies.