从3万亿美元市值增至4万亿美元市值，英伟达用了13个月。今年年初，英伟达股价曾一度遭遇DeepSeek"黑天鹅"的冲击。 7月9日美股开盘，英伟达的市值首次突破4万亿美元，成为全球首家4万亿美元市值公司。 从3万亿美元市值增至4万亿美元市值，英伟达用了13个月。 2024年6月18日美股收盘，英伟达市值飙升至超过3.4万亿美元，首次成为全球市值最大公司。这距离该公司挺进3万亿美元俱乐部不到两周。 2024年6月5日，英伟达首次收盘市值突破3万亿美元，从2万亿美元市值增至3万亿美元市值，英伟达仅用了约3个月的时间。 2023年5月30日，英伟达市值首次突破1万亿美元，并成为美股有史以来首家万亿美元市值的芯片公司。2024年3月1日，英伟达市值突破2万亿美元，从1万亿 美元市值增至2万亿美元市值，英伟达用了9个月时间。 今年年初，英伟达股价曾一度遭遇DeepSeek"黑天鹅"的冲击。市场担心，由于DeekSeek的人工智能模型R1仅需更少的芯片，就能实现与ChatGPT大模型相当 的能力，可能减少未来对英伟达芯片的需求。 对此，黄仁勋向市场多次回应。他强调，市场对R1的理解完全不对，这种AI推理模型对算力的需求实 ...

本文来自微信公众号：MacTalk，题图来自：AI生成 Manus 裁员，改新加坡公司了。 中国区员工 120 多人裁了 80 左右，也算大规模。 总部迁走，裁员开始，新加坡这边也在同步招人。 在 AI Agent 赛道里，Manus 是一个典型的样本。从中国起势，用英文命名，快速出海，背后站着资本 的推力和政策的红线。这类公司的走向，某种程度上也是整个行业的一面镜子。 至于 Manus 将总部迁入新加坡的原因，主要有两种声音： 社交平台信息显示，目前 Manus 在中国区的员工总数 120 人左右，除了四十多名核心技术人员迁往新 加坡总部之后，其余员工都将会进行裁员优化，给予 N+3 或者 2N 的赔偿。 今年初在全球爆火的中国 AI Agent 产品 Manus，4 个月后，主体迁到新加坡。动作之快，堪称"闪电迁 都"。 早在今年 6 月，Manus 的合伙人张涛就在一场活动上透露，公司总部已经迁往新加坡。查了一下资料， 新总部名叫"Butterfly Effect"，去年 8 月就已经在新加坡注册，背后的母公司设在开曼群岛。 根据 The information 的报道，Manus 已完成硅谷顶级风投 ...

SoundHound AI TodaySOUNSoundHound AI$12.94 +0.23 (+1.79%) 52-Week Range$3.94▼$24.98Price Target$10.50Add to WatchlistSoundHound AI NASDAQ: SOUN shares surged more than 30% in early July as a combination of factors rapidly reduced the available share count. They include inflows into agentic AI names, such as Salesforce, Palo Alto Networks, and Meta Platforms, as well as short-covering, which is robust. In June, short interest in SoundHound was over 30%, making it a powerful market force for share prices. Ge ...

Core Viewpoint - The emergence of AI agents capable of making payments autonomously is a significant trend in the AI application and business model landscape, with various companies developing solutions to facilitate this capability [4][20]. Group 1: Steps for Agent Payment - The process of enabling agent payment involves several steps, including research tools for inventory, communication tools for supplier interaction, note-taking for financial tracking, customer interaction capabilities, and price adjustment functionalities [7]. - Recent developments indicate that companies like Mastercard and Visa have launched AI agent payment solutions, while PayPal introduced its first MCP server, allowing LLMs to generate invoices and share payment links automatically [9][20]. Group 2: Key AI Products with Payment Integration - Perplexity Pro Shopping allows users to complete purchases directly within a chatbot interface, representing an early attempt at integrating agent and commerce [11]. - Stripe's Agent Toolkit provides virtual cards with customizable spending limits, addressing security and spending control for agent transactions [12]. - Shopify Sidekick automates product descriptions, promotions, and order processing, serving as an AI assistant for merchants [13]. - Adyen Uplift offers middleware services for AI agents, optimizing payment routes and retry mechanisms [14]. - Operator from OpenAI marks the beginning of a general agent framework, although it currently lacks payment integration [15]. - Mastercard's AgentPay distributes virtual cards to agents, enhancing their role in payment networks [16]. - Visa's Intelligence Commerce uses network tokens for transactions, ensuring security and budget control for AI agents [17]. - PayPal's MCP Server simplifies invoice generation and payment link sharing, making it easier for small businesses to implement payment solutions [18]. Group 3: Challenges in Achieving Autonomous Agent Payment - Three core challenges in achieving agent payment autonomy include defining the agent's role and scope, addressing fraud and KYA (Know Your Agent) issues, and clarifying liability in transactions [21][23][24]. - The ambiguity surrounding the agent's authority and the merchant's ability to verify agent interactions complicates the establishment of a secure payment framework [23]. - The responsibility for costs and liabilities in transactions involving agents remains unclear, particularly in scenarios like returns [26]. Group 4: Future Models of Agent Payment - Potential future models for agent payment include collaborative checkout with human oversight, authority derived from user wallets, limited payment capabilities through virtual cards, and agents possessing their own wallets funded by stablecoins [27].

Core Viewpoint - The article discusses the rising interest in AI agent startups founded by Chinese entrepreneurs, highlighting their innovative products and the attention they are receiving from major players like OpenAI [3][4]. Group 1: AI Agent Startups - Manus, an AI agent product developed by Chinese founders, gained significant attention earlier this year and received funding from Benchmark [4]. - Other notable AI agent products include Genspark, Lovart, Flowith, and Fellou, which aim to automate various tasks such as data analysis and scheduling [4][5]. - Lovart, founded by former ByteDance executive Chen Mian, attracted over 100,000 registered users within five days of its limited release [7]. Group 2: Product Features and Performance - Genspark's Super Agent, launched in April, can analyze raw data and create presentations, and even make phone calls for reservations [7][8]. - Within 45 days of its launch, Super Agent achieved an annual recurring revenue (ARR) of $36 million, indicating a strong market demand with at least 144,000 paying customers [8]. - Genspark has raised $160 million in funding and has been recognized by OpenAI and Anthropic for its use of advanced AI models [8][9]. Group 3: Strategic Moves and Market Positioning - Many of these startups are establishing headquarters in regions like Singapore to mitigate regulatory risks associated with their operations [9][10]. - Manus has relocated its headquarters to Singapore and has opened offices in California and Tokyo, while Genspark operates from both Singapore and Palo Alto [10][11].

Workday Research Reveals 76% of Employees Don't Know Who Owns Contracts, Resulting in Lost Revenue and Missed Opportunities PLEASANTON, Calif., July 9, 2025 /PRNewswire/ -- Workday, Inc. (NASDAQ: WDAY), the AI platform for managing people, money, and agents, today released "The Contract Intelligence Index Report." This research uncovers a widespread problem: many companies don't clearly define who is responsible for their contracts. This lack of clarity results in significant financial losses, inefficient o ...

Zoom AI Companion now connects with 16 third-party apps, including ServiceNow, Jira, Asana, Box, and others, to complete tasks and orchestrate agents without leaving ZoomCustom AI Companion add-on is now available to online customers, delivering extended capabilities, including AI meeting summaries across top meeting platforms and customized meeting summariesZoom expands core AI Companion capabilities, including real-time call questions, voice recorder for in-person meetings, and meeting agenda creation, in ...

以下文章来源于纵所周知101 ，作者纵所周知 101 纵所周知101 . 每一个与 AI 协作方式，都值得纵所周知。 作者 | 纵所周知101 来源 | 纵所周知101 导语： "谁愿意为此付费，为什么？"先验证需求，再优化技术。 深夜12点，朋友圈又炸了。 从爆火到裁员，只用了4个月。这是必然。 我为什么没"蹭"那波流量？ 3月7日，当整个科技媒体都在为Manus写赞歌的时候，我选择了另一条路。 不是因为我清高， 说实话，在全网都在为Manus叫好的时候，我选择唱反调是需要勇气的。 毕竟，谁也不想当那个"扫兴的人"。 这次不是因为新的AI突破，而是因为一条裁员新闻——那个4个月前让整个科技圈彻夜不眠的Manus，正在"优化"员工。还记 得3月6日那个不眠之夜吗？整个科技圈都为Manus疯狂，邀请码炒到5万元，200万人排队等候。所有人都在狂欢，说这是"AI Agent的GPT时刻"。 而我，写了一篇文章，泼了一盆冷水。 当时有人说我酸葡萄心理，有人说我不懂技术趋势。现在回头看，那篇文章里的疑问，都在今天得到了印证。 但我看到了一些让我不安的东西： 「没有用到有效内测码，对于没有亲自用过的产品，没有判断依据。 ...

以下文章来源于晚点科技 ，作者晚点团队 晚点科技 . 晚一点，好一点 今年 4 月，《晚点 LatePost》曾访谈 AI Agent 创业公司 Pokee.ai 创始人朱哲清， 他分享了强 化学习如何用于构造 Agent 。 我们了解到，Pokee.ai 近期完成种子轮融资，募资额约 1200 万美元。投资人包括对冲基金 Point72 旗下风险投资机构 Point72 Ventures、芯片公司高通的风投部门 Qualcomm Ventures， 以及英特尔 CEO 陈立武、前 Adobe CTO Abhay Parasnis 和前 Tinder CTO Maria Zhang。 Pokee.ai 去年 10 月成立，最近也只有 7 名员工。创始人朱哲清告诉我们，完成融资后 Pokee.ai 不计划大规模扩招，短期内研发员工数量不会超过 10 人。资金将用于扩张其 Agent 产品 Pokee 的可选工具集、加快对大公司客户销售。 创立 Pokee.ai 之前，朱哲清曾任 Meta "应用强化学习" 部门负责人，用强化学习算法改善内容推荐系 统，把上任前只剩 3 人，一度要关停的部门扩张至 10 余人，为 ...

作者 | 邓咏仪 编辑 | 苏建勋 2025年，还有人不知道AI是新时代潮流吗？过于Fancy的概念，已经0人在意。 无论是企业还是许多个人用户，对AI的关注重点，已经从大模型迅速转向落地。如何能够在真实的工作、学习、生活场景，让AI帮助降本增效——才是现 在AI圈子里的讨论重点。 7月9日，飞书发布并升级了知识问答、AI会议、飞书Aily、飞书妙搭等多款AI产品。 飞书CEO谢欣 来源：飞书 相较往年，这可以说是飞书又一次的海量更新——不仅是对多维表格、知识问答等已有产品进行了全新升级；而Aily和妙搭等新上线的AI全家桶，则是顺应 当下大模型落地难点的新产品。 此前几年，飞书已经走过从产品到市场覆盖的早期阶段，也积累了大量零售、高科技、先进制造等等行业的客户。 根据飞书披露的最新数据，在新能源汽车行业里，销量前30的品牌中，已有6成开始使用飞书；在茶饮行业的6家上市品牌中，也有5家是飞书的用户。 一场AI场景、AI新用户的竞赛也开始了。这两年，如何占据AI、具身智能领域的新玩家们的心智，成为协同办公赛道的竞争重点——这两年中，飞书也已 经拿下了包括DeepSeek、智元机器人在内的新客户。 △来源：作者拍 ...