⌘K
Change language Switch ThemeSign In
Narrow Mode
当失败成为企业重塑的催化剂 =============
哈 哈佛商业评论 @哈佛商业评论
One Sentence Summary
This article explores how companies, using practical cases from Amazon, Intercom, and Intuit, can view failure as an iteration signal and achieve organizational reinvention by rebuilding psychological safety, adjusting AI strategies, and decoupling technical architectures.
Summary
The article deeply analyzes how top companies handle 'failure' in the era of AI and transform it into a driving force for innovation. The core view is that failure is the twin brother of innovation, and companies should establish a 'fail fast' mechanism. The article unfolds from three dimensions: Culturally, it warns against 'watermelon goals' that hide problems, emphasizing that psychological safety is the foundation of accountability. Strategically, Intercom's case demonstrates the determination to proactively break old SaaS models and reorganize cross-functional teams to embrace Agentic AI. In terms of strategy and architecture, Intuit's case proves that business implementation needs to shift from 'conversation' to 'automation,' while Amazon solved the 'NPI tax' caused by bureaucratic processes through API-driven decoupling.
Main Points
* 1. Psychological safety is the foundation for building true accountability; a culture of fear leads to information distortion.When management severely punishes non-'green light' statuses, teams create 'watermelon goals' (green on the outside, red on the inside) to hide risks. Only trust allows problems to be exposed early and resolved. * 2. In the AI era, enterprises need the strategic resolve to break 'Conway's Law' and reinvent themselves.Intercom, by abandoning its mature seat-based pricing model, reorganizing cross-functional strike teams, and adopting Vibe Coding, proved that proactive 'self-disruptive' reinvention is key to sustaining growth in the face of a technological tsunami. * 3. Technical success does not equate to business success; AI implementation must return to the customer's real workflow.Intuit's experience shows that customers don't need to chat with software, but rather end-to-end automation. Through the 'follow-me-home' research method, AI strategy shifted from conversational to silently executing Agents in the background. * 4. Bureaucratic failures in processes are often a signal of excessively high coupling in the underlying technical architecture.Amazon, by treating cumbersome NPI processes as architectural governance signals, completely eliminated dependencies on cross-team coordination through API-fication and service decoupling, thereby unleashing productivity.
Metadata
AI Score
86
Website mp.weixin.qq.com
Published At Today
Length 2969 words (about 12 min)
Sign in to use highlight and note-taking features for a better reading experience. Sign in now
在商业世界中，“失败”往往是一个令人讳莫如深的词汇。但正如杰夫·贝佐斯所言：“失败与创新是不可分割的孪生兄弟。”亚马逊云科技过往经历了多次大规模转型，每一次转型的主要拐点，在某种程度上都是由失败推动的。
对于大多数企业来说，道理也类似。实际上，失败能激发学习，能加速创新并催生更好的产品，能帮助你排除选项，还能帮助企业构建韧性。因此，我们应该更早、更频繁地失败——这就是“快速失败”概念的由来。
那么，我们通常是如何失败的呢？ 文化陷阱： 西瓜式目标”与虚假的安全感
几年前，亚马逊内部有一个负责重大交付的团队。为了强化问责制，该团队的领导者建立了一种高压管控机制：每周召集全部门 200 人开会，逐一过问项目状态——是红灯、黄灯还是绿灯？
在这种文化下，任何非“绿灯”状态的汇报都会引发严厉的批评，甚至项目会被当场剥夺并转交给他人。
恶果迅速显现：所有项目指标一夜之间全变为“绿色”。但这并非效率提升，而是团队学会了制造“西瓜式目标”——外表是绿的，里面全是红的。团队开始刻意隐瞒问题，导致项目交付延期，客户满意度大幅下滑。核心原因在于，这种管理模式催生了恐惧文化，传递的信号是：承认风险就意味着失去项目主导权。这位领导者混淆了“指责文化”与“问责文化”，初衷是提升执行力与交付能力，结果却适得其反。
最终，该团队领导层更迭，管理思路随之彻底调整。新领导者后来回忆，他花了整整六个月时间，才让团队真正相信：他的核心目标不是追责解雇，而是助力团队成功。也正是从这时起，局面开始扭转——管理焦点从“惩罚问题”转向“赋能解决”，从纠结“谁的责任”转向关注“你需要什么支持”。
心理安全感才是构建真正问责制的基础。恐惧只能制造“可控”的假象，唯有信任才能搭建起高效的土壤。问责制从不来自恐吓，而根植于授权——这一认知的转变，最终重塑了团队活力与业绩表现。 战略重塑： Intercom如何在AI浪潮中自杀式”重生
如果说文化决定了能否发现失败，那么战略决心则决定了能否利用失败。SaaS 独角兽 Intercom 就曾经历过一次惊心动魄的“濒死体验”。
在 ChatGPT 发布之前，Intercom 正处于舒适区，拥有数亿美元的营收，甚至还在讨论 IPO（首次公开募股），并试图在所有战线上全线出击。然而，随着生成式 AI 的兴起，他们原本向企业出售“按座席收费”的客户服务软件模式面临灭顶之灾——AI 可以不知疲倦、以近乎零成本的方式提供服务。
当时所有的业绩数据依然向好，但这只是惯性使然。面对即将到来的技术海啸，他们做出了一个极端的决定：彻底放弃过去 15 年赖以成功的路径，在此刻“重新创办”公司，全面投入 AI。
Intercom 采取了几个反常规的举措：
打破康威定律（Conway's Law）：康威定律遵循围绕产品设计组织的原则，但为了构建 Agentic AI 工作流，必须审视产品的不同部分和不同方面，因而必须从各个部门“抢人”，组建跨职能的突击队。
革新传统设计模式：过去，设计环节多是设计师“闭门造车”，先拿出完整方案再小幅微调；而在 Agentic AI 时代，借助技术能力可同步探索数十种方向。如今，Intercom 通过 Vibe Coding（氛围编码）规划产品路线图，预判技术落地效果与市场适配性。
重构核心客户群体：过去，产品主要面向客服部门负责人销售；而 AI 带来的效率提升直接关联企业战略成本，销售对象随之转向 CFO、CIO 等更关注经济效益的核心决策层。
这场转型被团队称为“前所未有的疯狂与混乱”，过程极其艰难，但成果显著：其推出的 AI 客服机器人“Fin”，问题解决率从初始 23%飙升至 66%；推出首年便创造出 1 200 万美元年度经常性收入，且有望在两个季度内突破 1 亿美元。更重要的是，这场转型推动 Intercom 重回高增长轨道，跻身同阶段上市公司中的增长领军者行列。 策略校准： Intuit的AI转型如何从技术导向到商业落地
并非所有 AI 转型都如 Intercom 般激进。金融科技巨头 Intuit（旗下拥有 QuickBooks、TurboTax 等核心产品）的故事，揭示了另一重核心真理：技术层面的成功，不等于商业层面的成功。
Intuit 早在 7 年前就启动 AI 布局，累计投入约 10 亿美元。生成式 AI 爆发后，他们迅速推出“Intuit Assist”——一款嵌入财务管理软件 QuickBooks 边栏的 AI 聊天机器人。从技术维度看，该产品非常成功；但从商业维度看，QuickBooks 的小企业客户并不买账。
客户并不想与会计软件“聊天”，而是希望软件能帮他们高效完成工作。面对这一挫折，Intuit 回归其经典的“跟我回家（Follow Me Home）”调研法：深入客户工作现场，全程观察其实际工作流程。调研后发现，客户的核心痛点是需盯着两个屏幕，手动完成跨平台数据搬运。
基于这一核心洞察，Intuit 调整了核心 AI 策略：从“对话式 AI”转向“端到端自动化 Agent”。AI 不再局限于对话交互，而是在后台静默完成数据录入、分类与核对；仅在任务完成后，通过对话界面向用户同步结果。
为支撑这一转型，Intuit 甚至重构了技术架构：从中央集权式开发模式，转向“收割模式”（Harvesting Model）——让各产品团队自主探索 AI 应用创新，中央平台则负责“收割”成功组件并推动标准化落地。
新 AI 产品的成效立竿见影：小企业回款周期平均缩短 5 天，每月可节省 12 小时工作时间；78%的客户反馈 AI 让业务运营更轻松，68%的客户表示这使他们能将更多时间投入业务发展，而非繁杂的账目处理。 架构解耦：从“NPI税”到架构自由
当企业规模持续扩大，流程往往会演变为扼杀创新的官僚主义枷锁。亚马逊云科技曾推行“新项目倡议”（NPI）流程，初衷是规划并协调跨团队复杂项目。
但 NPI 逐渐演变为吞噬资源的烦琐机制：团队需耗费大量精力填写模板、做数据预测、召开协调会，甚至出现“项目经理管理项目经理”的荒诞局面。这一流程因严重占用资源、拖累效率，在内部被戏称为“NPI 税”。
有趣的是，亚马逊云科技并未通过优化 NPI 流程解决问题，而是将这种“流程失败”视为关键的“架构信号”：若开发一个功能需几十个团队反复开会协调，本质是系统耦合度过高所致。
因此，亚马逊云科技未止步于流程优化，而是反向发力架构治理——通过 API 化改造与服务解耦，彻底消除了对跨团队会议协调的依赖。技术架构理顺后，NPI 流程自然失去了存在的必要。
在如今的商业环境中，失败不再是一个终结性结果，而是组织迭代的关键输入信号。当项目状态全“绿”时，要主动探寻被隐藏的“西瓜式目标”；当流程推进受阻时，要拆解背后的架构依赖；当新技术落地遇冷时，要深入现场观察客户真实行为；当组织结构僵化时，要打破部门墙，围绕核心结果重组。
对于决策者而言，当下的核心任务并非预测未来，而是搭建一个能快速从错误中学习、不惜推翻过往成功经验的组织。因为在这个时代，唯一确定的失败，就是用昨天的逻辑维系当下的现状，却妄图迎接明天的到来。
斯蒂芬·布罗佐维奇（Stephen Brozovich）德斯·特雷纳（Des Traynor）亚历克斯·鲍洛什（Alex Balazs）| 文
斯蒂芬·布罗佐维奇是亚马逊云科技全球企业战略总监。德斯·特雷纳是Intercom 联合创始人兼首席战略官。亚历克斯·鲍洛什是Intuit 执行副总裁兼首席技术官。
哈 哈佛商业评论 @哈佛商业评论
One Sentence Summary
This article explores how companies, using practical cases from Amazon, Intercom, and Intuit, can view failure as an iteration signal and achieve organizational reinvention by rebuilding psychological safety, adjusting AI strategies, and decoupling technical architectures.
Summary
The article deeply analyzes how top companies handle 'failure' in the era of AI and transform it into a driving force for innovation. The core view is that failure is the twin brother of innovation, and companies should establish a 'fail fast' mechanism. The article unfolds from three dimensions: Culturally, it warns against 'watermelon goals' that hide problems, emphasizing that psychological safety is the foundation of accountability. Strategically, Intercom's case demonstrates the determination to proactively break old SaaS models and reorganize cross-functional teams to embrace Agentic AI. In terms of strategy and architecture, Intuit's case proves that business implementation needs to shift from 'conversation' to 'automation,' while Amazon solved the 'NPI tax' caused by bureaucratic processes through API-driven decoupling.
Main Points
* 1. Psychological safety is the foundation for building true accountability; a culture of fear leads to information distortion.
When management severely punishes non-'green light' statuses, teams create 'watermelon goals' (green on the outside, red on the inside) to hide risks. Only trust allows problems to be exposed early and resolved.
* 2. In the AI era, enterprises need the strategic resolve to break 'Conway's Law' and reinvent themselves.
Intercom, by abandoning its mature seat-based pricing model, reorganizing cross-functional strike teams, and adopting Vibe Coding, proved that proactive 'self-disruptive' reinvention is key to sustaining growth in the face of a technological tsunami.
* 3. Technical success does not equate to business success; AI implementation must return to the customer's real workflow.
Intuit's experience shows that customers don't need to chat with software, but rather end-to-end automation. Through the 'follow-me-home' research method, AI strategy shifted from conversational to silently executing Agents in the background.
* 4. Bureaucratic failures in processes are often a signal of excessively high coupling in the underlying technical architecture.
Amazon, by treating cumbersome NPI processes as architectural governance signals, completely eliminated dependencies on cross-team coordination through API-fication and service decoupling, thereby unleashing productivity.
Key Quotes
* Failure and innovation are inseparable twin brothers. * Teams learned to create 'watermelon goals'—green on the outside, red on the inside. * The management focus shifted from 'punishing problems' to 'empowering solutions,' and from dwelling on 'whose fault it is' to focusing on 'what support you need.' * The only certain failure is to maintain the current status quo with yesterday's logic while attempting to embrace tomorrow. * When process advancement is blocked, dissect the underlying architectural dependencies; when new technologies face cold reception, delve into the field to observe real customer behavior.
AI Score
86
Website mp.weixin.qq.com
Published At Today
Length 2969 words (about 12 min)
Tags
Corporate Transformation
AI Strategy
Organizational Culture
Agentic AI
Architectural Decoupling
Related Articles
* The Global Phenomenon of Chinese Short-Form Dramas: Unpacking Their Underlying Logic * Tech Enthusiast Weekly (Issue 381): What Are China's AI Large Model Leaders Thinking? * Major Update! Jensen Huang's First Keynote of the Year: The World is Undergoing Two Platform Revolutions! New Rubin Computing Architecture Debuts! First Autonomous Driving AI Open-Sourced, AI-Native Application Framework Released! Grok 5 Parameter Scale Revealed * How Leading Companies Scale AI: A Four-Quadrant Strategy * Sam Altman's Deep Dive: Countering Google's Challenges, Disregarding IPOs, and Unveiling the Ultimate Path to 'Superintelligence' * 127. Large Model Quarterly Report New Year Dialogue: Guang Mi's Prediction of the AI War's Two Major Alliances and the Third Paradigm of Online Learning * “Who You Work With Matters More Than Position and Money” * GLM-5 Technical Report Deep Dive | a16z: "The Best Open-Source Model" * Jensen Huang's CES 2026 Keynote: Three AI Turning Points in 2025 and the Rise of Agentic AI * #423. Silicon Valley Godfather Ben Horowitz: American Ambition, Management Philosophy, and Culture Building in the AI Era HomeArticlesPodcastsVideosTweets
When Failure Becomes a Catalyst for Corporate Reinvention... ===============