广告
加载中

腾讯不能只有WorkBuddy

刘诗雨 2026-11-20 10:37
刘诗雨 2026/11/20 10:37

邦小白快读

EN
全文速览

总:腾讯AI战略与WorkBuddy的关键数据

1. 财报数据:腾讯Q3营收2073亿元,同比增长8%,增速为近两年单季最低;Non-IFRS归母净利润665亿元,同比下降6%;资本开支814亿元,远超市场预期的554亿元。

2. WorkBuddy成绩:日活跃用户突破1000万,每日消耗超过100万亿Token,最近七天收入按年化折算超5亿美元;从3月上线到10月底,7个月内迭代100个大版本,累计完成90亿次任务请求。

3. 产品布局:WorkBuddy支持混元及Kimi、DeepSeek等第三方模型,开放平台上架Skills超20万个,企业版落地50多个行业;推出Buddy One,打通腾讯文档、腾讯会议、ima等办公产品。

总:腾讯战略转向与竞争态势

1. 战略转向:马化腾称AI是生死存亡的问题,腾讯要从AI赋能转向用AI再造一个腾讯,一半兵力守核心业务、一半转型探索AI Native业务。

2. 竞争态势:豆包工作DAU突破800万,企业客户ARR突破20亿元;字节系App使用时长占比41.8%,腾讯系为28.6%。

3. 其他产品:微信智能体小微接入30多个场景,日活3000万;元宝未达预期,已削减投入。

总:腾讯AI产品矩阵与品牌竞争格局

1. 产品研发:WorkBuddy 7个月迭代100个大版本,不押注自建模型,支持Kimi、DeepSeek等第三方模型,开放平台Skills超20万个,企业版落地50多个行业。

2. 品牌渠道:WorkBuddy被定位为办公工具和智能体的统一入口,Buddy One以WorkBuddy为工作台,打通腾讯文档、腾讯会议、ima,支持连接企业内部系统和第三方Agent。

3. 用户行为:微信智能体小微接入朋友圈、公众号、视频号等30多个场景,日活3000万,优先控制推理成本而非追求最强模型能力;元宝不再以扩大用户规模为主要目标。

总:品牌竞争与消费趋势

1. 竞争态势:豆包工作DAU800万、ARR20亿元,与WorkBuddy形成正面较量;字节系App使用时长占比41.8%,腾讯系28.6%。

2. 消费趋势:AI Native产品成为新方向,腾讯管理层将WorkBuddy视为第一个跑出来的AI Native产品,强调从AI时代新需求中成长。

3. 品牌定价参考:WorkBuddy当季收入是二季度的5倍以上,中国百强企业已有60%成为客户。

总:AI办公市场的机会与风险提示

1. 增长市场:WorkBuddy企业版落地50多个行业,中国百强企业60%成为客户,当季收入是二季度的5倍以上;日活突破1000万,ARR超5亿美元。

2. 合作机会:WorkBuddy开放平台上架Skills超20万个,引入外部开发者和合作伙伴;支持混元及Kimi、DeepSeek等第三方模型,卖家可围绕技能开发生态寻找机会。

3. 风险提示:腾讯错过时间窗口后,先进AI计算资源供应受限,管理层承认下一代模型训练将耗时更久;元宝未达预期,已大幅削减投入。

总:可学习点与商业模式

1. 可学习点:WorkBuddy以工作台为统一入口整合办公产品,先兑现技术投入价值再拓展生态;不把所有筹码押在自建模型上,借助第三方模型扩大能力。

2. 商业模式:以ARR衡量AI产品商业化,WorkBuddy七天收入年化超5亿美元;豆包工作以DAU800万、ARR20亿元作为披露口径。

3. 竞争提示:字节与腾讯在办公AI市场正面交锋,豆包工作选择在腾讯财报电话会前发声,狙击意味明显。

总:腾讯AI转型对数字化和产品设计的启示

1. 数字化启示:腾讯从轻资产经营转向重资产打法,资本开支从319亿元一路升至814亿元,连续三个季度环比分别增长63%、65%和54%;剔除新AI产品影响,利润口径为833亿元,同比增长8%。

2. 产品设计需求:WorkBuddy 7个月迭代100个大版本,累计完成90亿次任务请求;Buddy One以工作台打通腾讯文档、腾讯会议、ima,支持连接企业内部系统和第三方Agent。

3. 商业机会:WorkBuddy企业版落地50多个行业,开放平台Skills超20万个,百强企业60%成为客户;微信小微接入30多个场景,日活3000万。

总:资源约束与生产启示

1. 资源约束:先进AI计算资源供应受限,下一代模型训练将耗时更久,说明算力资源是AI产品生产的关键瓶颈。

2. 生产启示:腾讯提出从智能、应用和基础设施三个层面构建AI赋能的腾讯,基础设施层可以靠买、智能层靠混元迭代和对外开放、应用层靠内部发力。

总:AI行业趋势与市场格局

1. 行业趋势:沙利文报告指出,未来三年行业将从资源竞赛进入效率与生态竞赛,竞争关键在于谁能更高效地将芯片、算力、模型与生态转化为企业可直接使用的智能生产力。

2. 市场格局:2026年上半年阿里云占据中国全栈AI云服务市场42%的份额;字节构建以AI为核心的商业生态,底层为豆包、Seedance、火山引擎,上层为豆包、红果AI短剧、豆包工作。

总:客户痛点与解决方案

1. 客户痛点:企业办公AI需求旺盛,但内部协同存在制约,企微AI助手大圆与WorkBuddy两套体系并行,涉及聊天记录等敏感信息需由企微智能体管理访问权限后再交换信息。

2. 解决方案:WorkBuddy开放平台上架Skills超20万个,Buddy One作为统一入口打通办公产品并支持第三方Agent;微信小微接入30多个场景,在保障隐私安全前提下逐步打通接口权限。

3. 新技术:混元模型、Kimi、DeepSeek等第三方模型、AI Native产品,以及腾讯三层AI能力架构。

总:平台建设做法与招商运营

1. 平台做法:WorkBuddy开放平台引入外部开发者和合作伙伴,上架Skills超20万个;Buddy One以WorkBuddy为工作台,打通腾讯文档、腾讯会议、ima,并支持连接企业内部系统和第三方Agent。

2. 招商与客户:企业版落地50多个行业,中国百强企业60%成为客户,当季收入是二季度的5倍以上;微信小微接入朋友圈、公众号、视频号等30多个场景,日活3000万。

总:运营管理与风向规避

1. 运营管理:马化腾要求每个系统都能与WorkBuddy对接融合;企微AI助手大圆与WorkBuddy并行,敏感信息由企微智能体管理访问权限后再交换。

2. 风向规避:面对微信十几亿用户,小微优先控制推理成本,不追求最强模型能力;微信生态的社交关系、支付、小程序等资源不能随意调用,需保障用户隐私安全和接口权限。

3. 平台启示:腾讯将一半兵力留在原有阵地守核心业务,另一半转型为海军探索AI Native业务,平台需平衡存量运营与新战场开拓。

总:产业新动向与竞争格局

1. 产业新动向:沙利文报告指出未来三年AI行业从资源竞赛进入效率与生态竞赛;阿里围绕自研芯片、云服务、模型平台和应用展开全栈布局,字节构建以AI为核心的商业生态,腾讯提出从智能、应用和基础设施三层构建全新的AI赋能的腾讯。

2. 竞争格局:2026年8月字节系App使用时长占比41.8%,腾讯系28.6%;豆包工作DAU800万、ARR20亿元,WorkBuddy日活1000万、ARR超5亿美元。

总:新问题与商业模式

1. 新问题:腾讯后知后觉,错过时间窗口后先进AI计算资源供应受限,下一代模型训练将耗时更久;元宝未达预期,已削减投入;小微能力暂不及AI助手Muse,系有意控制推理成本。

2. 商业模式:WorkBuddy成为首个AI Native产品,7个月迭代100个大版本,企业版落地50多个行业,百强企业60%成为客户;腾讯以ARR衡量商业化进展,并强调基础设施、混元模型、微信生态与AI应用形成协同。

3. 政策启示:微信生态资源调用需在保障用户隐私和安全的前提下打通接口与权限,AI替个人用户办事仍存难题。

返回默认

声明:快读内容全程由AI生成,请注意甄别信息。如您发现问题,请发送邮件至 run@ebrun.com 。

我是 品牌商 卖家 工厂 服务商 平台商 研究者 帮我再读一遍。

Quick Summary

Overall: Tencent's AI strategy and key WorkBuddy metrics

1. Financial results: Tencent's Q3 revenue was RMB 207.3 billion, up 8% year on year, the slowest single-quarter growth in nearly two years; non-IFRS net profit attributable to shareholders was RMB 66.5 billion, down 6% year on year; capital expenditure was RMB 81.4 billion, far above the market expectation of RMB 55.4 billion.

2. WorkBuddy performance: daily active users exceeded 10 million, daily token consumption surpassed 100 trillion, and revenue over the most recent seven days annualized to more than USD 500 million; from launch in March to the end of October, it iterated through 100 major versions in seven months and cumulatively handled 9 billion task requests.

3. Product layout: WorkBuddy supports Hunyuan and third-party models such as Kimi and DeepSeek; its open platform lists more than 200,000 Skills; the enterprise edition has been deployed across more than 50 industries; and Buddy One was launched to integrate office products including Tencent Docs, Tencent Meeting, and ima.

Overall: Tencent's strategic shift and competitive landscape

1. Strategic shift: Pony Ma called AI a matter of survival, saying Tencent must move from AI empowerment to rebuilding Tencent with AI, with half its forces defending core businesses and half pursuing AI Native businesses.

2. Competitive landscape: Doubao Work's DAU exceeded 8 million and enterprise customer ARR exceeded RMB 2 billion; ByteDance apps accounted for 41.8% of app usage time, versus 28.6% for Tencent apps.

3. Other products: WeChat's AI agent Xiaowei is integrated into more than 30 scenarios with 30 million DAU; Yuanbao fell short of expectations and investment has been cut.

Overall: Tencent's AI product matrix and brand competition landscape

1. Product R&D: WorkBuddy iterated through 100 major versions in seven months, does not bet solely on self-built models, supports third-party models such as Kimi and DeepSeek, lists more than 200,000 Skills on its open platform, and has its enterprise edition deployed across more than 50 industries.

2. Brand channels: WorkBuddy is positioned as a unified entry point for office tools and agents; Buddy One uses WorkBuddy as the workbench, integrates Tencent Docs, Tencent Meeting, and ima, and supports connections to enterprise internal systems and third-party agents.

3. User behavior: WeChat AI agent Xiaowei is integrated into more than 30 scenarios, including Moments, Official Accounts, and Channels, with 30 million DAU; it prioritizes inference cost control over pursuing the strongest model capabilities; Yuanbao no longer has user-scale expansion as its primary goal.

Overall: Brand competition and consumer trends

1. Competitive landscape: Doubao Work has 8 million DAU and RMB 2 billion ARR, forming a direct rivalry with WorkBuddy; ByteDance apps account for 41.8% of app usage time, versus 28.6% for Tencent apps.

2. Consumer trends: AI Native products are becoming the new direction; Tencent management views WorkBuddy as the first AI Native product to break out, emphasizing growth from new AI-era demand.

3. Brand pricing reference: WorkBuddy's quarterly revenue was more than five times its Q2 revenue; 60% of China's top 100 enterprises are already customers.

Overall: Opportunities and risks in the AI office market

1. Growth market: WorkBuddy's enterprise edition has been deployed across more than 50 industries, 60% of China's top 100 enterprises are customers, and quarterly revenue was more than five times Q2 revenue; DAU exceeded 10 million and ARR exceeded USD 500 million.

2. Partnership opportunities: WorkBuddy's open platform lists more than 200,000 Skills and brings in external developers and partners; it supports Hunyuan and third-party models such as Kimi and DeepSeek, so sellers can look for opportunities around the skill-development ecosystem.

3. Risk warnings: After missing the time window, Tencent faces constrained supply of advanced AI compute resources, and management acknowledges that training next-generation models will take longer; Yuanbao fell short of expectations and investment has been significantly cut.

Overall: Key takeaways and business model

1. Key takeaways: WorkBuddy uses the workbench as a unified entry point to integrate office products, monetizing technology investment first before expanding the ecosystem; it does not put all its chips on self-built models and expands capabilities through third-party models.

2. Business model: ARR is used to measure AI product commercialization; WorkBuddy's seven-day revenue annualized to more than USD 500 million; Doubao Work discloses 8 million DAU and RMB 2 billion ARR.

3. Competitive signal: ByteDance and Tencent are facing off directly in the office AI market; Doubao Work chose to make its announcement before Tencent's earnings call, a clear sniping move.

Overall: Implications of Tencent's AI transformation for digitalization and product design

1. Digitalization implications: Tencent is shifting from an asset-light model to an asset-heavy approach, with capital expenditure rising from RMB 31.9 billion to RMB 81.4 billion and growing 63%, 65%, and 54% quarter on quarter for three consecutive quarters; excluding the impact of new AI products, the profit measure was RMB 83.3 billion, up 8% year on year.

2. Product design needs: WorkBuddy iterated through 100 major versions in seven months and cumulatively handled 9 billion task requests; Buddy One uses the workbench to integrate Tencent Docs, Tencent Meeting, and ima, and supports connections to enterprise internal systems and third-party agents.

3. Business opportunities: WorkBuddy's enterprise edition has been deployed across more than 50 industries, its open platform lists more than 200,000 Skills, and 60% of top 100 enterprises are customers; WeChat Xiaowei is integrated into more than 30 scenarios with 30 million DAU.

Overall: Resource constraints and production implications

1. Resource constraints: Supply of advanced AI compute resources is constrained, and training next-generation models will take longer, showing that compute resources are a key bottleneck in AI product production.

2. Production implications: Tencent proposes building an AI-empowered Tencent across three layers—intelligence, applications, and infrastructure; the infrastructure layer can be bought, the intelligence layer relies on Hunyuan iteration and opening up, and the application layer relies on internal efforts.

Overall: AI industry trends and market landscape

1. Industry trends: A Frost & Sullivan report notes that over the next three years the industry will move from a resource race to an efficiency and ecosystem race; the key will be who can more efficiently convert chips, compute, models, and ecosystems into intelligent productivity that enterprises can use directly.

2. Market landscape: In the first half of 2026, Alibaba Cloud held 42% of China's full-stack AI cloud services market; ByteDance is building an AI-centric business ecosystem, with Doubao, Seedance, and Volcano Engine at the bottom layer and Doubao, Hongguo AI short dramas, and Doubao Work at the upper layer.

Overall: Customer pain points and solutions

1. Customer pain points: Enterprise demand for office AI is strong, but internal collaboration is constrained; the WeCom AI assistant Dayuan and WorkBuddy run in parallel, and sensitive information such as chat records must have access permissions managed by the WeCom agent before being exchanged.

2. Solutions: WorkBuddy's open platform lists more than 200,000 Skills; Buddy One serves as a unified entry point that integrates office products and supports third-party agents; WeChat Xiaowei is integrated into more than 30 scenarios and is gradually opening interface permissions while ensuring privacy and security.

3. New technologies: Hunyuan, third-party models such as Kimi and DeepSeek, AI Native products, and Tencent's three-layer AI capability architecture.

Overall: Platform-building practices and merchant recruitment and operations

1. Platform practices: WorkBuddy's open platform brings in external developers and partners and lists more than 200,000 Skills; Buddy One uses WorkBuddy as the workbench, integrates Tencent Docs, Tencent Meeting, and ima, and supports connections to enterprise internal systems and third-party agents.

2. Recruitment and customers: The enterprise edition has been deployed across more than 50 industries, 60% of China's top 100 enterprises are customers, and quarterly revenue was more than five times Q2 revenue; WeChat Xiaowei is integrated into more than 30 scenarios, including Moments, Official Accounts, and Channels, with 30 million DAU.

Overall: Operations management and risk avoidance

1. Operations management: Pony Ma requires every system to connect and integrate with WorkBuddy; the WeCom AI assistant Dayuan runs in parallel with WorkBuddy, and sensitive information has access permissions managed by the WeCom agent before being exchanged.

2. Risk avoidance: Facing WeChat's more than 1 billion users, Xiaowei prioritizes inference cost control rather than pursuing the strongest model capabilities; WeChat ecosystem resources such as social relationships, payments, and mini programs cannot be accessed at will, and user privacy and interface permissions must be protected.

3. Platform implications: Tencent keeps half its forces on its original positions to defend core businesses and transforms the other half into a navy exploring AI Native businesses; platforms need to balance existing operations with opening new battlefields.

Overall: New industry developments and competitive landscape

1. New industry developments: A Frost & Sullivan report notes that over the next three years the AI industry will move from a resource race to an efficiency and ecosystem race; Alibaba is pursuing a full-stack layout around self-developed chips, cloud services, model platforms, and applications; ByteDance is building an AI-centric business ecosystem; and Tencent proposes building a new AI-empowered Tencent across intelligence, applications, and infrastructure.

2. Competitive landscape: In August 2026, ByteDance apps accounted for 41.8% of app usage time and Tencent apps for 28.6%; Doubao Work had 8 million DAU and RMB 2 billion ARR, while WorkBuddy had 10 million DAU and more than USD 500 million ARR.

Overall: New issues and business model

1. New issues: Tencent was slow to act; after missing the time window, supply of advanced AI compute resources is constrained, and training next-generation models will take longer; Yuanbao fell short of expectations and investment has been cut; Xiaowei's capabilities temporarily lag behind the AI assistant Muse, due to intentional inference cost control.

2. Business model: WorkBuddy has become the first AI Native product, iterating through 100 major versions in seven months, with its enterprise edition deployed across more than 50 industries and 60% of top 100 enterprises as customers; Tencent uses ARR to measure commercialization progress and emphasizes synergies among infrastructure, Hunyuan, the WeChat ecosystem, and AI applications.

3. Policy implications: Calls to WeChat ecosystem resources must open interfaces and permissions while protecting user privacy and security; AI acting on behalf of individual users remains difficult.

Disclaimer: The "Quick Summary" content is entirely generated by AI. Please exercise discretion when interpreting the information. For issues or corrections, please email run@ebrun.com .

I am a Brand Seller Factory Service Provider Marketplace Seller Researcher Read it again.

11月18日,互联网大厂间的较量,来到了真正的争分夺秒时刻。

19时57分,赶在腾讯Q3财报电话会开始前三分钟,豆包工作高调宣布自家DAU突破800万,企业客户ARR(注:年化收入)突破20亿元。

最近半年,腾讯WorkBuddy在办公AI产品中持续领先,唯一的「不确定性」便是从未公布过的商业化进展。

豆包工作选择此时发声,狙击意味不言而喻。

腾讯也很快亮出了自己的成绩单。随后的财报电话会上,腾讯总裁刘炽平同步了一组数据:WorkBuddy日活跃用户突破1000万,每日消耗超过100万亿Token,最近七天的收入按年化折算(ARR)超过5亿美元。

按此折算,WorkBuddy年化日均收入约为137万美元,折合人民币约900多万元。

虽然两家披露的统计口径未必一致,仅从ARR来看,腾讯总算在与字节的AI较量中扳回一局。

只是资本市场未必有心情跟着庆祝,次日收盘,腾讯股价跌去6个点。

过去,稳定的现金流、丰厚的利润和持续增长的核心业务,构成了腾讯的投资价值。如今,腾讯有了更大的「梦想」——推动AI战略转向、走出熟悉的经营模式,资本市场不得不重新审视这场转型的成本与回报。

在战略叙事转折之际,一个WorkBuddy跑出来,还不足以安抚外部的不安。

因为腾讯的AI拼图还未成型。

01 腾讯也得交AI的学费

从整体业绩来看,腾讯三季度交出了一份营收增速放缓、净利润下滑的财报。

腾讯当季营收2073亿元,同比增长8%,增速为近两年单季最低水平。Non-IFRS口径下,归母净利润665亿元,同比下降6%;经营利润为681亿元,同比下降6%。

AI投入是利润承压的重要原因。腾讯披露,若剔除「新AI产品」的影响,该口径利润为833亿元,同比增长8%。

与此同时,腾讯当期资本开支达到814亿元,远超市场预期的554亿元。

投资者们内心可能五味杂陈:以前怕腾讯舍不得在AI上花钱,现在怕腾讯太舍得在AI上花钱。

但如果拉长时间轴来看,真正的转折出现在2026年一季度:当季腾讯资本开支环比增长63%,此后连续两个季度分别增长65%和54%,从319亿元一路升至814亿元。

这背后是腾讯正在改变自己的经营方式。

过去很长一段时间,腾讯是一家以流量和产品为核心的互联网公司。微信积累了庞大的用户群,游戏、广告营销、支付等业务负责将用户规模和活跃度转化为收入。

凭借成熟的产品体系和商业化能力,腾讯可以在轻资产的模式下获得丰厚利润。

AI刚刚兴起时,腾讯也一度沿着这条熟悉的路径前进:将AI能力嵌入已有业务,通过提升广告推荐、内容分发和游戏开发效率,提升整个腾讯生态的赚钱效率。

2025年三季度,马化腾便表示,AI已在广告精准定向、游戏用户参与度,以及编程和内容制作等领域发挥作用。

相比阿里、字节在AI上的激进投入,腾讯的转型姿势被外界调侃为「囤最少的卡,赚最多的钱」。这套策略并非没有道理:先让AI服务于成熟业务,尽快兑现技术投入的价值,再寻找新的增长机会。

不过这个选择本身充满争议。支持者赞其「务实」,批评者却认为腾讯在顶层设计上缺乏「梦想」。

现在回看,真正的问题在于,无论初衷如何,AI产业的竞争节奏没有给腾讯留下太多观望时间。

AI颠覆来得太快,竞争环境的变化也很快。用马化腾最近在腾讯28周年员工沟通会上的话说,「浪已经杀过来了」。

AI已不再是单点技术或产品的较量。

沙利文最新发布的《中国全栈AI云服务市场报告2026年H1》指出,未来三年,行业将从「资源竞赛」进入「效率与生态竞赛」。竞争的关键,在于谁能更高效地将芯片、算力、模型与生态转化为企业能够直接使用的智能生产力。

腾讯的主要竞争对手,已经围绕AI搭建起新的能力组合——

阿里围绕自研芯片、云服务、模型平台和应用展开全栈布局,C端产品有千问,B端有千问办公。沙利文数据显示,2026年上半年,阿里云占据中国全栈AI云服务市场42%的份额。阿里集团CEO吴泳铭为云和AI业务定下的目标:未来五年年收入突破1000亿美元。

字节也在抖音之外构建起一个新的、以AI为核心的商业生态。这个生态的底层是大语言模型豆包、多模态视频生成模型Seedance、云与AI服务平台火山引擎,上层是现象级C端产品豆包、红果AI短剧,B端应用豆包工作……

QuestMobile数据显示,2026年8月,中国移动互联网使用时长排名前50的App中,字节系占比41.8%,腾讯系为28.6%。

继续在原有业务上做AI加法,已经不足以应对下一阶段的竞争形势。

由此不难理解,为什么一向温和的马化腾在司庆现场显得如此凝重。他向11万名腾讯员工强调,即便一些核心业务看似地位稳固,也可能被飞速发展的AI浪潮颠覆,「这是一个生死存亡的问题」。

换言之,腾讯现阶段的梦想,必须从「老树发新芽」式的AI赋能,转向用AI再造一个腾讯。

这意味着,腾讯需要从过去熟悉的轻资产经营,转向适应重资产的打法。今年一季度开始的资本开支激增,以及并行的组织架构变革、产品整合,其实都是最新战略意志的落地。

AI这笔学费,腾讯已经开始补交了。

02 一个WorkBuddy还不够

腾讯在今年二季度财报中首次明确提出,要从智能、应用和基础设施三个层面,构建一个「全新的、AI赋能的腾讯」。

三层能力中,基础设施层可以靠买,智能层可以通过混元迭代和对外开放两条腿走路,应用层却只能靠腾讯内部发力。

细数腾讯的AI应用版图,WorkBuddy是最先跑出来的产品。

从今年3月上线到10月底,WorkBuddy在7个月内迭代了100个大版本,累计完成90亿次任务请求。

在产品策略上,WorkBuddy没有把所有筹码押在自建模型上。除混元外,它还支持Kimi、DeepSeek等第三方模型。9月上线开放平台后,WorkBuddy进一步引入外部开发者和合作伙伴,目前平台上架Skills超过20万个,企业版已落地50多个行业。

10月29日,腾讯又推出Buddy One,以WorkBuddy为工作台,打通腾讯文档、腾讯会议、ima等办公产品,并支持连接企业内部系统和第三方Agent。

很明显,腾讯试图将WorkBuddy打造成办公工具和智能体的统一入口。

不过,这个入口仍在整合当中。有腾讯员工跟《降噪NoNoise》提到一个观察:马化腾要求每个系统都能与WorkBuddy对接融合。但不同于飞书、钉钉直接在对话框中内嵌AI助手的模式,企业微信有自己的AI助手「大圆」。WorkBuddy接入企微后,两套体系会并行。涉及聊天记录等敏感信息时,由企微智能体管理访问权限,获得授权后再和WorkBuddy交换信息。

当然,内部协同的制约,在WorkBuddy亮眼的成绩单面前可以被暂时忽略。刘炽平在三季度财报电话会上提到一个数据:WorkBuddy当季收入是二季度的5倍以上,中国百强企业已有60%成为它的客户。

在腾讯28周年员工沟通会上,管理层将WorkBuddy视为第一个跑出来的AI Native产品。按照管理层的思路,腾讯要将一半兵力留在原有阵地,像陆军一样守住核心业务、借助AI提高效率;另一半则转型为海军,主动寻找新战场,探索更多AI Native业务。

腾讯需要的显然不止一个WorkBuddy。从目前的产品布局来看,WorkBuddy开辟了企业办公市场,在更广泛的个人用户场景,腾讯还没找到突破。

在C端场景,元宝曾被寄予厚望,但刘炽平坦言,元宝并没有达到腾讯此前的预期,腾讯目前已不再将扩大用户规模作为元宝的主要目标,而是转向提升产品能力,并大幅削减了相关投入。

相较之下,微信智能体「小微」有机会成为连接个人需求与微信服务生态的AI入口。

今年6月开始灰度测试的小微,目前已接入微信朋友圈、公众号、视频号等30多个场景的触点,日活跃用户达到3000万。未来,它还将进一步连接小程序、微信小店等服务,让用户直接通过AI完成任务和交易。

在三季度财报电话会上,腾讯承认小微的能力目前还不及爆火的AI助手Muse。管理层解释,这其实是有意为之:面对微信十几亿用户,小微必须优先控制推理成本,而不是一味追求最强的模型能力。

另外,微信生态虽然拥有社交关系、支付、小程序等资源,但「小微」并不能随意调用。如何在保障用户隐私和安全的前提下,打通不同业务的接口和权限,让AI替个人用户办事,仍是一道难题。

这也解释了为什么马化腾越来越强调AI Native。腾讯需要更多像WorkBuddy这样从AI时代的新需求中成长起来、能够独立创造收入的产品。

好消息是,腾讯还有相对充足的资金以及微信生态这座富矿;坏消息是,腾讯也要为自己的后知后觉买单——错过时间窗口后,先进AI计算资源供应受限。管理层承认,下一代模型训练将耗时更久。

这意味着,腾讯需要同时解决技术、产品、组织和资源配置上的问题。

站在全局来看,WorkBuddy至少证明,腾讯可以在AI时代做出一款有市场、有商业化潜力的产品。接下来便是持续夯实智能、应用和基础设施三层,让基础设施、混元模型、微信生态和各个AI应用形成协同,逐渐建立起AI驱动的增长新模型。

对于腾讯而言,真正值得期待的,不只是下一个WorkBuddy什么时候出现,还有腾讯能否在AI时代找到微信量级的新入口。

毕竟,一款产品可以跑在前面,一个完整的新生态才能决定一家大公司的时代排序。

注:文/刘诗雨,文章来源:降噪NoNoise,本文为作者独立观点,不代表亿邦动力立场。

文章来源:降噪NoNoise

广告
微信
朋友圈

FAQ回顾

腾讯WorkBuddy目前的商业化表现如何?

截至2026年11月披露,WorkBuddy日活跃用户突破1000万,每日消耗超过100万亿Token,最近七天收入按年化折算ARR超过5亿美元;当季收入是二季度的5倍以上,中国百强企业已有60%成为其客户。

WorkBuddy是什么产品,主要覆盖哪些办公场景?

WorkBuddy是腾讯2026年3月推出的AI Native办公产品。它除混元外支持Kimi、DeepSeek等第三方模型,9月上线开放平台,上架Skills超过20万个,企业版已落地50多个行业;10月推出Buddy One,以WorkBuddy为工作台打通腾讯文档、腾讯会议、ima等办公产品,并连接企业内部系统和第三方Agent。

腾讯为什么需要更多WorkBuddy这类AI Native产品?

腾讯管理层将WorkBuddy视为第一个跑出来的AI Native产品,并提出一半兵力守核心业务、一半转型探索AI Native新业务。继续在原有业务上做AI加法已不足以应对下一阶段竞争;马化腾称AI浪潮可能颠覆核心业务,是生死存亡问题,腾讯需从AI赋能转向用AI再造一个腾讯。

中国全栈AI云服务市场的竞争趋势是什么?

沙利文《中国全栈AI云服务市场报告2026年H1》指出,未来三年行业将从“资源竞赛”进入“效率与生态竞赛”。竞争关键在于谁能更高效地把芯片、算力、模型与生态转化为企业可直接使用的智能生产力。2026年上半年,阿里云占中国全栈AI云服务市场42%的份额。

腾讯在个人AI入口上有哪些进展和挑战?

腾讯个人AI入口主要有元宝和微信智能体小微。元宝未达预期,已不再以扩大用户规模为主要目标,转向提升产品能力并削减投入;小微2026年6月灰度测试,接入朋友圈、公众号、视频号等30多个场景,日活3000万,未来连接小程序、微信小店。管理层称其能力暂不及Muse,并需优先控制推理成本、处理隐私与权限。

这么好看,分享一下?

朋友圈 分享

APP内打开

赞 +1
+1
微信好友 朋友圈 新浪微博 QQ空间
关闭
收藏成功
发送
/140 0