梦晨 发自 凹非寺 量子位 | 公众号 QbitAI 用Claude Code写代码的人，终于不用每次开新会话都从头解释项目背景了。 顶GitHub开源热榜的一款持久化记忆系统Claude-Mem，直击AI编程助手最致命的痛点：跨会话失忆。 Claude-Mem本身100%免费，还能帮你省token钱。 它通过"三层渐进式披露"的检索架构，常规使用下能节省90% Token，测试阶段的"无尽模式"更是能把Token消耗砍掉95%，工具调用次数上 限直接拉高20倍。 给Claude Code装上"长期记忆" 传统AI编程助手有个绕不开的问题，每次新会话都是一张白纸。 昨天刚聊完的架构设计、上周敲定的编码规范、刚刚那些踩过的坑，AI统统不记得。开发者只能一遍遍重复解释，时间和Token都在这种"复 读"中白白流失。 Claude-Mem的解法是在本地环境搭建一套完整的记忆系统。 它采用事件驱动架构，通过五个生命周期钩子 （SessionStart、UserPromptSubmit、PostToolUse、Stop、SessionEnd） 在后台静默运 行。 每当Claude Code执行文件读写、代码编辑、命令执 ...

AI叙事分化 最近北美大厂的财报又陆续出来了，亚马逊、谷歌、Meta 和微软计划在 2026 年投入约 6600 亿美元 的资本支出。 但这次，市场对大厂们激进的资本开支和前段时间甲骨文的RPO的反应，跟去年完全不同了，只画 大饼已经不再被认可了。 现在市场对AI价值的评判标准已从技术先进性或用户规模转向对收入结构、盈利能力及现金流的实 际贡献。AI不再是统一估值加分项，市场开始区分"已兑现收益"与"投入期远景"，定价差异更多反 映企业所处商业化阶段。能够率先证明AI投入转化为可持续盈利的公司更易获得市场认可，ROI与 自由现金流成为关键定价指标。随着AI需求扩张，市场对玩家的定价差异将聚焦于商业化阶段，而 非长期价值分歧。 SemiAnalysis又是连着两天出了两篇文章，内容都比较长，我们还是只讲里面的主要观点。 AI 智能体发展的核心转折点 1、Claude Code 是 AI 智能体发展的核心转折点 这并非单纯因其实现了 AI 写代码的功能升级，而是它完成了从 AI 模型被动响应 到 AI 智能体主动 执行的范式突破，彻底重构了人机交互的底层逻辑，同时成为 AI 从单一工具向通用协作体进化的标 志性产 ...

Core Viewpoint - Anthropic's enterprise-level product strategy has achieved a significant breakthrough, prompting a reevaluation of the AI competitive landscape and leading to substantial market reactions across various sectors [1][2]. Group 1: Company Overview and Market Position - Anthropic, founded by former OpenAI researchers, is valued at approximately $350 billion and plans to initiate an IPO within the year [1]. - The company has seen its annualized revenue surge from about $1 billion at the beginning of 2024 to over $9 billion by the end of 2025, with projections to exceed $30 billion by the end of 2026 [1]. - Anthropic's AI model API spending market share has reached nearly 80% as of January, indicating a rapid increase in its appeal among enterprise users [4]. Group 2: Product Strategy and Development - Anthropic has chosen a distinct commercialization path, focusing on developer and enterprise tools rather than consumer products, which has been validated by recent market performance [3]. - The launch of industry-specific tools, including plugins for legal, sales, finance, marketing, and customer support, has deepened its enterprise service ecosystem [6]. - The Claude Code tool, which can read existing codebases and autonomously plan and execute tasks, has emerged as a leader in software engineering tools, showcasing the initial realization of AI "agent" capabilities [7]. Group 3: Competitive Landscape and Industry Impact - The rapid adoption of Anthropic's models is seen as the most impactful event in the AI sector since the launch of ChatGPT, with investors betting on its potential to transform white-collar workflows [2]. - Despite facing fierce competition from Google and OpenAI, Anthropic's tools are expected to extend beyond coding to reshape broader white-collar work processes [9]. - The company has committed to not introducing advertisements in its products, differentiating itself from competitors like OpenAI, which has tested ads in its offerings [14][15]. Group 4: Safety and Management Strategy - Anthropic has maintained a cautious and safety-focused public image, emphasizing the importance of responsible AI development [11][12]. - The company's unique approach to AI feedback, utilizing AI systems to review AI-generated content, has improved model iteration efficiency and reduced human bias [12][13]. - Investors view Anthropic's clear product focus and stable management as factors that make it a more robust long-term investment compared to OpenAI [13].

作者 | Patrick Farry 译者 | 田橙 GitHub CEO Thomas Dohmke 近日发出了一则措辞严厉的警告："要么拥抱 AI，要么离开这个职业。"但所谓拥抱 AI，并不只是使用代码自动补全 工具那么简单。它意味着我们核心能力的一次转移——从对语法的熟练掌握，转向系统思维（Systems Thinking），学会把问题不断拆解，直到小 到可以交由 AI 去解决。一句话概括：我们现在都是架构师了。 我正在开发一个 IoT 应用，整体由设备端固件、后端系统以及 Web UI 组成。尽管我本身具备软件工程背景，但在这个项目中，我一直在使用 Claude Code 来提升开发效率，并帮助我应对一些并不十分熟悉的语言和框架。我的技术栈包括：设备端使用 Python + PyTorch，前端采用 React + TypeScript，后端则由 MQTT + Node.js + Postgres 构成。起初，与 Claude 的协作并不顺利。我的请求经常会引发对整个代码库的大规模改 动。随着我逐渐学会如何更合理地组织代码结构、并对提示词进行调整和约束，情况开始好转。现在，我已经可以在不进行逐行代码审 ...

Core Insights - The article discusses the launch of CodePilot, a desktop client for Claude Code, highlighting its comprehensive features and user-friendly design [1][3]. Group 1: Key Features of CodePilot - CodePilot supports all core functionalities of Claude Code, including folder selection, model switching, slash commands, Skills invocation, and MCP server integration, providing a significantly improved user experience [3]. - The client offers enhanced chat history management, allowing users to easily access previous conversations, with each message displaying the associated cost for transparency [5][6]. - A visual configuration management interface has been introduced, enabling users to modify configuration files, Skills, MCP, and plugins without needing command line interaction [8]. - Users can preview the contents of folders directly within the application, making it easier to access text files and other resources [9]. - Third-party API configurations are supported, allowing flexibility for users who may not have direct access to the official API [11]. - The connection status of Claude Code is clearly displayed, providing guidance for users in case of connectivity issues [13][14]. Group 2: Agent Teams Collaboration - The article introduces the Agent Teams mode, which allows a main intelligent agent to delegate tasks to multiple sub-agents, enabling parallel work and real-time communication between agents [19][20]. - Enabling Agent Teams is straightforward, requiring users to update to the latest version of Claude Code and follow simple instructions to configure it [21]. - Tips for utilizing Agent Teams include having Claude assist in writing planning prompts, emphasizing the importance of preliminary research for role definition, and advocating for flexible role design tailored to specific tasks [23][25][27]. - The article emphasizes that the current era allows for rapid development of fully functional applications, with the use of Opus 4.6 proving to be cost-effective due to its efficiency and reduced need for corrections [30][31].

在这次发布之前，Anthropic 内部和部分早期用户已经开始让 Opus 4.6 参与一项持续时间很长的工 程任务：从零开始，用 Rust 编写一个完整的 C 编译器，并要求它能够编译 Linux 内核。 这项实验持续了约两周时间，期间累计运行了近两千次 Claude Code 会话，最终产出了一个规模约 10 万行代码的编译器。该编译器不仅能够在多种架构上构建 Linux 6.9，还可以编译 FFmpeg、 Redis、PostgreSQL、QEMU，并通过了 GCC 自身 99% 的 torture test，甚至能够成功编译并运行 Doom。整个实验的 API 成本约为 2 万美元。 作者 | Tina Anthropic 正在升级它"最聪明的模型"。 随着新一代旗舰模型 Claude Opus 4.6 的发布，Anthropic 释放出的信号十分明确：这并不是一次常 规的性能小修小补，而是一轮围绕长任务、复杂工作，以及智能体（agent）如何真正干活展开的系 统性升级。 为了让外界更直观地理解这一成果的尺度，有网友在社交平台上给出了一个对照： GCC 的开发从 1987 年开始，历经 37 年，投入 ...

Core Insights - The emergence of Agentic AI, particularly through tools like Anthropic's Claude Code, is causing significant shifts in the software and data center industries, leading to a notable sell-off in software stocks [1][2] Group 1: Impact of Agentic AI - Claude Code is predicted to account for over 20% of daily code commits on GitHub by year-end, a substantial increase from the current 4%, marking a pivotal moment in AI development [2] - The transition from manual coding to "vibecoding," where code is generated through natural language prompts, is expected to accelerate the deployment of AI agents that can operate independently [3] Group 2: Companies Benefiting from Agentic AI - Arm Holdings is positioned to benefit from the growth of Agentic AI, as it is known for licensing CPU instruction sets, which are becoming increasingly vital for handling workloads in this new AI landscape [5][6] - The demand for CPUs is anticipated to rise significantly due to their role as agents in orchestrating tasks, which could lead to increased revenue for Arm [6] Group 3: Financial Performance and Future Outlook - Arm's data center royalty revenue has more than doubled year-over-year, with expectations that its data center business will surpass mobile as the largest revenue category in the coming years [7] - Despite a guidance for slower revenue growth in the fiscal fourth quarter, the strong performance in the third quarter reassures investors about Arm's growth potential [7] - The company is also expected to benefit from increasing royalty rates as it introduces newer products like Armv9, enhancing its revenue streams [10]

AI初创公司Anthropic本周凭借其企业级产品战略实现关键突破，引发市场对AI竞争格局的重新评估。 其工具已在全球股市触发连锁反应，软件、法律、金融数据及地产等多个板块出现显著市值蒸发，显示 出企业软件市场正面临结构性重构。 Menlo Ventures合伙人Matt Murphy透露，Claude Code工具最初为Anthropic内部使用，在验证其显著效 果后，公司迅速将其转化为对外产品。本周，Anthropic进一步发布了一系列面向法律、销售、金融、 营销及客户支持等垂直行业的专用"插件"工具，持续深化其企业服务生态。 编码能力成为核心竞争力 这家成立仅五年的旧金山公司正以约350亿美元估值进行规模相当的融资轮，并计划年内启动IPO。自 2024年初以来，其年化收入从约10亿美元跃升至2025年底的逾90亿美元。据媒体援引知情人士，该公司 预计2026年底年化收入将突破300亿美元，并预计在2028年首次实现盈亏平衡，这一时点较OpenAI提前 两年。 本周Anthropic发布的多款行业专用工具持续引发市场震荡。在最新旗舰模型Claude Opus 4.6推出后， Salesforce、Intu ...

SemiAnalysis分析指出，Claude Code已占据GitHub代码提交的4%，预计2026年底超20%，标志着AI智能体技术进入规模化商用拐点。它正重构全 球15万亿美元的信息工作市场，并推动Anthropic季度营收增量超越OpenAI。传统软件商业模式面临根本冲击，而微软等巨头陷入平衡云服务增长与保 护核心产品的战略两难。智能体竞争已从生成响应转向实现可交付成果，整个行业正转向效率为王的实战阶段。 该报告强调， Claude Code不仅改变了编程形态，更预示着AI智能体将重塑全球约15万亿美元的信息工作市场。 目前，Anthropic依托该技术实现营收 快速增长，其季度年度经常性收入（ARR）增量已超越OpenAI，反映出AI智能体赛道的竞争格局正发生结构性变化。 这一趋势已引发产业链的深度响应。埃森哲已签署协议，计划在金融、医疗及公共部门培训3万名专业人员使用Claude，成为目前规模最大的企业级部 署案例。与此同时， 传统软件与SaaS商业模式面临根本性挑战。 分析指出，相当多的领域都将受到影响。 随着Claude Code（和Cowork）的兴起，智能体的总市场规模远大于大型语言模型本 ...

Anthropic旗下AI编程工具Claude Code已占GitHub公共代码提交量的4%，并预计到2026年底该比例将超20%。研究机构SemiAnalysis最新分析指 出，Claude Code是AI"智能体"的转折点。 该报告强调，Claude Code不仅改变了编程形态，更预示着AI智能体将重塑全球约15万亿美元的信息工作市场。目前，Anthropic依托该技术实现 营收快速增长，其季度年度经常性收入（ARR）增量已超越OpenAI，反映出AI智能体赛道的竞争格局正发生结构性变化。 对科技巨头而言，战略平衡变得更为复杂。微软既通过Azure为OpenAI、Anthropic等提供算力支持，又需防御AI代理工具对其Office 365等核心 产品的冲击。据悉，CEO Satya Nadella已亲自参与AI产品战略调整，可见该领域对公司未来发展影响深远。 随着Claude Code等AI代理工具持续渗透，从开发流程到企业运营的智能化重构进程正全面加速。 Claude Code与智能体未来 智能体将成为人类与人工智能交互的主要方式。Claude Code的出现，不仅展示了人类如何引导AI，更揭示了智能 ...