Rewired 2026 第二版解读

Executive Summary

• 第二版不是第一版的小修订。 作者明确说明，约 40%Source for 40%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 内容为全新、另有 25%Source for 25%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 大幅扩展；核心框架仍是六项企业能力，但“什么叫做得好”已经被 Agentic AI、AI 平台、业务领导技术能力和分布式运营模式重新定义。
• 全书实际为 39Source for 39Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章，不是 35Source for 35Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章。 第 35Source for 35Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章是第六部分的结尾；第七部分另有 4Source for 4Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 个案例章。第二版案例是 LATAM Airlines、Toyota、Freeport-McMoRan 和 DBS，第一版的乐高集团案例没有作为独立章节保留。
• 核心经营逻辑是“业务域而非用例”。 企业先确定经济杠杆点，再对 1-3 个优先业务域进行端到端重构，以 5-15 个相互关联的用例、共享数据和平台、一次性的组织变革共同释放价值。
• 领先企业的经济画像很有力度。 书中对 20Source for 20Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 家成功转型企业的分析显示：三年平均 EBITDA 提升约 20%Source for 20%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition，多数落在 10%-30%Source for -30%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition；平均 1-2 年转为现金正贡献；一次性投入每 1Source for 1Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 美元带来约 3Source for 3Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 美元增量 EBITDA；转型核心阶段通常持续 2-4 年，之后进入持续经营节奏。
• AI 工具普及只是桌面生产率，竞争优势来自工作流重构。 Agentic AI 的战略价值主要位于端到端 Agentic workflows 和跨工作流 Agentic systems，而非单点助手；护城河来自专有数据、行业知识、业务逻辑、平台复用和实施能力。

版本核对：为什么本报告整理 39Source for 39Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章

用户要求中写有“共 35Source for 35Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 个章节”，这与第一版结构接近，但与所提供 PDF 的正式目录不一致。2026 年修订扩展第二版的 Section Six 包含第 30-35 章，随后 Section Seven 继续包含第 36-39 章，因此全书合计 39Source for 39Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章。

为忠实于原文，本报告完整整理 39Source for 39Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章；涉及乐高集团时，只说明它未被第二版列为独立案例，不把第一版内容混入第二版摘要。

目录导航

• 引言与六大能力
• Section One：Creating the Transformation Roadmap
• Section Two：Building Your Talent Bench
• Section Three：An Operating Model That Outruns the Competition
• Section Four：Technology for Speed and Distributed Innovation
• Section Five：Embedding Data Everywhere
• Section Six：Adoption and Scaling: Your Force Multipliers
• Section Seven：Transformation Journey Stories
• 六项核心企业能力
• 业务主导的实施路线图
• CEO 与高管责任
• 案例对比与实践经验

章节快速跳转

1 · 2 · 3 · 4 · 5 · 6 · 7 · 8 · 9 · 10 · 11 · 12 · 13 · 14 · 15 · 16 · 17 · 18 · 19 · 20 · 21 · 22 · 23 · 24 · 25 · 26 · 27 · 28 · 29 · 30 · 31 · 32 · 33 · 34 · 35 · 36 · 37 · 38 · 39

引言：Your guide to Rewired（p.1Source for 1Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）与 Transformation manifesto（p.12Source for 12Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

作者把 Tech & AI transformation 定义为：由 CEO 和高管团队领导，发展组织与技术能力，使企业能够持续改善客户体验、降低单位成本，并随时间形成可持续竞争优势。两个经营结果必须同时成立，而且转型永远没有真正的终点。

第二版强调，Rewired 不是员工广泛使用 AI 工具的指南。个人助手会迅速成为行业标配；真正的竞争优势来自把 AI 系统嵌入核心流程、产品和服务，重构企业实际运行方式。

六项能力保持不变：业务主导的路线图、人才、运营模式、技术、数据、采用与扩展。新增变化主要来自两股力量：Agentic AI 带来了非结构化数据、Agent 工作流和新风险要求；管理实践则要求业务负责人更深地参与业务重构，并通过新的运营模式缩短创新周期。

成功转型的 EBITDA 提升分布

下图来自第 2Source for 2Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章对 20Source for 20Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 家成功企业的分析。它显示的是三年后稳定状态下，转型带来的 EBITDA 提升占相关公司或业务单元原 EBITDA 的比例。12Source for 12Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 家企业达到 20%Source for 20%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 以上提升，样本平均约为 20%Source for 20%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition。该图是成功样本的经验画像，不代表所有启动转型的企业都能达到同样结果。

Section One：Creating the Transformation Roadmap（p.19Source for 19Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

本部分是全书的总设计层，依次回答：领导层如何建立认知与信念；转型的经济目标是否值得投入；为什么应以业务域为单位；怎样把业务域重构为可执行路线图；Agentic AI 适合哪些工作流；六项能力如何同步建设；最后如何把责任写入 C-suite 契约。

第 1Source for 1Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：Inspired, aligned, and ready to go: Working with your top team（p.25Source for 25Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

核心论点

Tech & AI 转型不能由一个负责人代替高管团队承担。每位高管必须具备足够技术理解、对共同愿景达成一致，并承诺用不同方式领导。研究显示，只有约 7%Source for 7%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 的大型企业拥有技术理解充分的高管团队；这类领导团队所在公司在收入增长和估值上领先同行超过 48%Source for 48%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition。

关键框架

1. 领导学习旅程：高管基础训练、跨行业 go-and-see、art-of-the-possible 工作坊。每名高管在有效领导转型前至少应投入 20Source for 20Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 小时学习，通常需要更多。
2. 愿景四要素：行业趋势、商业模式经济杠杆、既有竞争优势、战略飞轮。愿景必须写出业务域、经济价值和所需能力，而不是“部署 AI”。
3. 领导承诺校准：优先域负责人投入 20%-40%Source for -40%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 时间；高管团队共同问责；3-5 年目标应追求超过 20%Source for 20%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 的利润改善；每位高管建立持续学习机制。
4. Rewired 领导者四种特质：conviction forger、fearless learner、organizational architect、forever transformer。

执行建议与前后关系

CEO 应先用数月建立团队信念，而不是急于批准项目；用客户价值和经营结果定义 North Star；把技术学习、人才调整、运营模式和持续参与写入个人日程。第 1Source for 1Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章建立领导前提，第 2Source for 2Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章据此量化值得追求的经济结果，第 3-5 章再把愿景收敛为业务域和工作流。

第 2Source for 2Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：The economics of Tech & AI transformations（p.39Source for 39Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

核心论点

没有可信价值就没有领导信念，没有回报就无法持续投资，没有商业案例就无法形成责任对齐。作者研究 20Source for 20Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 家成功企业，用实际路线图、财务结果和高管访谈建立转型经济画像。

经济画像

• 三年稳定状态下，平均 EBITDA 提升约 20%Source for 20%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition，多数在 10%-30%Source for -30%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition。
• 三分之二企业主要依靠 3Source for 3Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 个或更少业务域实现价值。
• 平均 1-2 年转为现金正贡献；聚焦少数高回报业务域的企业更快。
• 一次性投入每 1Source for 1Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 美元，平均产生约 3Source for 3Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 美元年度增量 EBITDA。
• 核心转型通常 2-4 年进入 business-as-usual，之后仍以新一轮数据、AI 或 GenAI 推进持续升级。

Toyota 的端到端供应链重构带来 30%Source for 30%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 利润改善；Freeport 初始项目不足 12Source for 12Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 个月回本。行业研究还显示，Tech & AI 成熟度领先者的五年 TSR 在保险、银行、消费、能源和材料行业显著高于落后者。

执行建议与前后关系

商业案例应以业务域 KPI 和分阶段现金回报为基础，而非工具采用量；投资通过阶段门逐笔释放。第 2Source for 2Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章给第 3Source for 3Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章的“域聚焦”提供经济依据，也为第 7Source for 7Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章的财务计划和高管问责设定标准。

第 3Source for 3Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：Focus on business domains, not use cases（p.49Source for 49Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

核心论点与定义

业务域是端到端业务流程，也可被称为 customer journey、value stream 或 product unit。域要“大到足以产生经营影响，小到可以减少外部依赖”。每个域通常包含 5-15 个相互关联的用例；典型公司约有 10-20 个域。

域方法的三项优势

1. 端到端解决问题，避免单点方案被上下游依赖抵消。
2. 复用同一组数据、技术和人才资产，加快后续用例。
3. 对同一用户群一次性统筹流程、培训和激励，减少碎片化变革疲劳。

域转型架构依次为：业务域 → 价值杠杆/KPI → Tech & AI 解决方案 → 用例/模型/数据 → 数据与技术平台。优先级按价值潜力与可行性评估：客户体验、财务收益、速度和风险；高管赞助、数据技术准备度、采用难度和扩展难度。分析可筛出 2-5 个候选域，但多数企业同一时间重点推进 1-3 个。

新业务组合框架

第二版把新业务视为特殊业务域。机会按战略适配、客户需求、独特优势和可行性筛选，并以组合化阶段门管理：大约 100Source for 100Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 个想法 → 20Source for 20Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 个蓝图 → 6Source for 6Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 个进入建设 → 3Source for 3Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 个规模化 → 1Source for 1Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 个 home run。

章节关系

本章确定转型单位和投资边界；第 4Source for 4Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章把单个优先域变成完整交付路线图，第 5Source for 5Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章深入 Agentic workflow，第 6Source for 6Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章配置支持这些域的企业能力。

第 4Source for 4Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：Have business leaders lead the reimagination of their domain（p.63Source for 63Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

核心论点

域重构最重要的角色是 domain owner：既懂端到端流程和用户，也能理解 Tech & AI、识别改进机会并领导变革。企业应建立统一的 Domain-to-Value Delivery（DVD） playbook，把想法转成落袋价值。

DVD 五阶段

1. Domain reimagination（1-3 个月）：形成域级未来状态、解决方案和用例路线图。
2. Solution discovery（2-6 周）：研究用户痛点和技术可行性，输出 solution brief。
3. Proof of concept（1-3 个月）：快速原型、可行性测试和用户反馈，输出 MVP。
4. Production（2-4 个月）：将原型工程化为安全、可靠、可扩展的生产方案。
5. Adoption and scaling：改流程、培训用户、监测使用并复制，最终兑现价值。

阶段间设置正式 stage gates 和时间上限，允许部分工作并行，但每个方案必须经过价值、可行性、生产和采用检验。

域重构五步法

明确业务问题和价值杠杆 → 设计用例路线图与解决方案架构 → 评估数据和技术缺口 → 估算收益与投资 → 设计实施顺序、团队配置和责任。可用 zero-cost thinking 打破旧假设，并用 Level 1-3 的用例成熟度每 6-12 个月交付新价值。

财务估算应避免虚假精确：Tech & AI 业务创新常有 1x-5xSource for -5xSource: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 回报，早期估算误差约 30%Source for 30%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 可以接受；必须纳入变革管理、持续运行成本和共享平台资金。第 4Source for 4Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章是路线图从战略到执行的主干，第 30-33 章进一步负责采用、扩展、价值追踪和中途调整。

第 5Source for 5Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：Reimagining workflows with agentic AI（p.79Source for 79Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

核心论点与定义

AI agent 是由 GenAI 支持、能够围绕目标自主执行多步行动的软件系统；Agentic workflow 把多个 Agent 连接起来完成端到端流程；Agentic system 同时优化多个工作流和经营目标。真正的价值起点是重构工作流，而不是追求 Agent 本身更“聪明”。

四级 Agentic 自动化

1. Individual augmentation：个人助手，单项任务生产率可提高 20%-30%Source for -30%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition，但通常难形成企业级差异。
2. Task automation：自动完成单个任务或低复杂工作流，早期案例可使周期或处理成本改善 20%-40%Source for -40%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition。
3. Agentic workflows：多个 Agent 端到端协作，人类从 in the loop 转向 above the loop，价值集中在吞吐、周期、体验和成本。
4. Agentic systems：跨工作流优化资源、利润和客户结果，需要 CEO/董事会重构跨职能权责和运营模式。

E.ON Next 案例

八个 Agent 共同完成客户识别、意图分类、数据查询、账户更新、开票和人工转接。实施一年后，客户满意度提高 6Source for 6Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 个百分点，平均处理时间下降 8%Source for 8%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition，交易成功率提高 14%Source for 14%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition，超过 40%Source for 40%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 来电由 AI 完成识别验证，入站电话平均成本下降接近 50%Source for 50%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition。企业同时建设 AI 平台、系统接口、25Source for 25Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 项质量与护栏指标，并用小模型满足实时延迟、大模型做事后复核。

工作流筛选五项标准

价值潜力、是否适合推理与有界自主、正确性要求、数据与系统准备度、实施复杂度。高频、高差异、高价值且结果可观测的流程优先；错误代价高的场景需要更多人工监督和验证投入。

六条实施经验

从工作流而非 Agent 出发；并非所有任务都该用 Agent；像培训新员工一样持续评估 Agent；让每一步可追踪可验证；把重复能力做成复用模块；重新设计人类的监督、例外处理和数据策展职责。评估指标包括任务成功率、F1/精确率与召回率、检索准确率、语义相似度、LLM-as-judge、偏差、幻觉率和校准误差。

第 6Source for 6Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：You're only as good as the capabilities you build（p.97Source for 97Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

核心论点

概念和用例容易被复制，组织能力很难复制。企业必须同时具备六项能力，不能把一项推到行业领先而让其他能力停在 novice。作者的基准观察是：约 50%-70%Source for -70%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 企业处于 novice，30%-40%Source for -40%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 达到 competent，约 5%Source for 5%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 是行业最佳。

六项能力成熟度

业务路线图从点解决方案进化为端到端域重构和 Agentic systems；人才从零散专家进化为业务与 IT 均完成升级；运营模式从 side-car 进化为业务、技术、运营和控制职能融合；技术从单体集中化进化为平台化和 AI 驱动开发；数据从难用的数据湖进化为自助消费和专有数据优势；采用扩展从“最后一公里”进化为流程重构和 Agent orchestration。

能力建设按 Crawl（0-18 月）、Walk（18-36 月）、Run（36Source for 36Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 月以后）协调推进；一般从一个成熟度等级提升到下一级需要 2-3 年。眼前能力要 fit-for-purpose，同时从一开始按规模化设计。

Talent 与 Technology 的 Make/Buy 逻辑

企业无法靠外包获得 Tech & AI 卓越。软件系统分为三类：Differentiating systems 应端到端内部拥有；High-velocity systems 以内部为主、选择性补充；Utility systems 优先 SaaS 或外包，“adopt, don't customize”。技术则默认购买，只有能够合理地建立持续领先时才自建。

合作伙伴

五类伙伴是战略统筹、人才供给、数字工程、技术平台和数据伙伴。合作应以经营结果、人才质量、能力转移和逐步减少依赖为契约，而不是替代内部能力。

第 7Source for 7Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：The ultimate corporate team sport（p.111Source for 111Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

核心论点

路线图最终要汇总成 2-3 年的实施计划和财务计划：业务域与用例按短期价值排序，基础能力与所支持的价值同步建设，变革管理和风险进入同一时间轴，C-suite sponsor 对每一条工作流负责。这是高管团队“stack hands”的执行契约。

领导责任

• CEO：统一愿景、处理跨职能冲突、盯关键指标和激励，每月投入约 2-4 天，启动期更多。
• Chief Transformation Officer：通常为内部资深业务领导，直接向 CEO 汇报，负责 2-3 年的日常推进，直至转型进入经营常态。
• 业务/职能负责人：亲自重构所属业务域，定义价值、赞助交付并对采用负责。
• CIO/CTO/CDO：分别覆盖核心系统与架构、客户技术产品、数字解决方案与现代工程；角色可合并，但能力不能缺失。
• Chief AI Officer：可提供专业聚焦并负责 AI 平台和风险，但不能夺走业务领导的转型责任，也不应把 AI 工程师孤立为中央团队。
• Chief Data Officer：连接域路线图和数据需求，建设易消费数据与治理体系。
• CHRO：高管升级、技术人才招聘、技能体系、职业路径和 IT 人才重构。
• CFO：商业案例、价值兑现、动态资金和季度重配；常参与新运营模式设计。
• CRO：重构分布式创新下的风险流程；董事会应有 2-3 名真正经历过 Tech & AI 转型的成员。

章节关系

本章把前六章的认知、经济目标、域选择、DVD、Agentic 工作流和能力计划合成为责任系统；Section Two-Five 分别建设人才、运营模式、技术和数据，Section Six 负责兑现价值。

Section Two：Building Your Talent Bench（p.125Source for 125Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

第二版把人才问题从“招聘数字人才”扩展为两项更难的规模化任务：培养能领导域重构的 N-2Source for -2Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition/N-3Source for -3Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 业务负责人，以及对传统 IT 人才结构进行系统手术。最终目标是让业务领导、工程人才和未来的 Human-Agent 协作能力同时成熟。

第 8Source for 8Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：Building the tech muscle of your business leaders（p.131Source for 131Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

技术型业务负责人应能制定客户导向愿景、把愿景转为用例路线图、监督技术交付，并亲自负责端到端采用。典型大型公司约有 15-30 个核心流程，需要培养约 75-150 名 N-2Source for -2Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition/N-3Source for -3Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition domain owners。

领导者通过持续学习、聚焦 20%Source for 20%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 以上价值机会、理解技术与组织系统、深入 sprint、识别顶尖技术人才、接受不确定性和阶段门来建立“第二块肌肉”。企业层面的三项动作是：选对人并调整激励；通过真实业务 capstone 进行战略升级；采用 persistent funding、业务领导下的跨职能团队和 KPI 绩效制度。

第 9Source for 9Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：The big IT talent 'surgery'（p.141Source for 141Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

规模化 AI 的最大人才瓶颈常在传统 IT。作者提出三个“70%Source for 70%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition+”结构性转变：70%Source for 70%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 以上工程人才内部化；70%Source for 70%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 以上是直接交付的 doer engineers 而非管理协调岗位；70%Source for 70%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 以上达到 competent 或 expert 水平。

实施从技能 taxonomy、角色/技能/熟练度盘点和多种评估开始，再把当前供给与路线图需求比较，决定招聘、升级、退出和外部补充。目标组织从底部庞大的低技能金字塔转为中间熟练人才更厚的钻石形，并利用 AI 提高小型高质量团队的产出。

第 10Source for 10Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：Creating an environment where tech talent thrives（p.159Source for 159Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

这一章把招聘、认可和成长整合为技术人才环境。企业需要有说服力且真实的 EVP、懂技术的招聘团队、快速候选人流程、第一周即可进入实际工作的 onboarding，以及用技能而非职称管理人才。

核心制度包括按 skill family 和 proficiency 定义 competency markers；用市场稀缺性和实际影响定薪；建立专家与管理双职业通道；由资深技术人员参与绩效与晋升；针对开发、云、UX 等不同 craft 建独立学习旅程。HR 的角色是建立可扩展系统，而不是长期依赖例外政策。

第 11Source for 11Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：Rethinking your talent model for an agentic world（p.175Source for 175Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

Agentic AI 不只是给员工增加工具，而是改变工作内容和组织单元。AI Agent 的输入、推理和输出都需要人类定义数据、权限、业务规则、模型、工作流、质量标准和升级边界。

三类角色将演变：管理者从 supervisor 变为 hybrid workflow orchestrator；专家从 practitioner 变为把隐性知识编码进 Agent 的 coach；业务领导从 domain owner 变为跨系统、流程和经济目标的 integrator。组织可能从职能转向工作流、从金字塔转向钻石、从观察式学徒制转向模拟与判断训练。

作者建议 HR 不要追逐 hype，而采取六项 no-regret actions：普及 AI fluency；专业化 workflow thinking；轮岗进入 3-5 个 agentic lighthouse；提高领导判断和系统思维要求；识别 orchestrator/coach/integrator 人才；加速数据平台和跨职能运营模式等基础能力。

Section Three：An Operating Model That Outruns the Competition（p.193Source for 193Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

运营模式不是组织架构图，而是结构、决策系统、工作流程、人才配置和文化规范的组合。竞争优势来自资源能否流向重要机会、决策能否靠近工作、团队能否减少依赖并快速迭代。

第 12Source for 12Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：Go team! The ground floor of your operating model（p.199Source for 199Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

跨职能交付团队是运营模式的最小单元。团队围绕业务使命长期存在，拥有业务、产品、设计、数据科学、软件/数据工程、AI 风险与敏捷支持等能力，并对从 discovery、POC、production 到 adoption 的完整生命周期负责。

团队配置应按 solution archetype 设计，而不是平均分配人头；至少每个解决方案一个团队，复杂方案可有多个团队，并按季度滚动更新未来 12-18 个月需求。敏捷的本质是短周期用户反馈和小批量交付，必须与自动化 SDLC 配合，不能只复制仪式。

第 13Source for 13Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：The revolutionary idea: Embedding engineering talent in the business（p.217Source for 217Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

本章的“革命性想法”是把工程交付能力嵌入业务域，同时通过 chapters 保持专业标准、人才发展和资源流动。业务和技术不再以需求单/项目交接协作，而是共同对产品、流程和 KPI 负责。

三种分布式模式是 Digital Factory（约 10-50 个团队）、Domain & Platform（约 50-1,000+ 团队）和 Enterprise-wide model（1,000Source for 1,000Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition+ 单元）。选择取决于技术在战略中的角色和组织准备度。D&P 转型通常 1-2 年，再需 1-2 年达到成熟；平台提供复用，域直接创造客户和经营价值。

第 14Source for 14Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：The blueprint for a faster enterprise（p.237Source for 237Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

新运营模式必须先有经过高管共同认可的 target-state blueprint，通常至少需要三个月严肃设计。蓝图涵盖 scope and unit design、management system、ways of working、people and culture、tooling。

关键设计包括域/平台边界与负责人、应用和预算归属、跨职能团队与运营单元、明确 decision rights、OKR/QBR、persistent funding、专业人才归属和统一工具。目标不是把现有人原样移进新盒子，而是借机重新定义技能、责任与人才质量。

第 15Source for 15Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：Making the transition in a controlled way（p.257Source for 257Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

运营模式转型会影响 20%-100%Source for -100%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 员工，不能一次性“大爆炸”。应由有业务威望、能亲自解决问题的转型领导带队，按 waves 推进，并在每一波完善人员识别、域平台边界、沟通和资源配置 playbook。

每个域和平台经过 incubation、mobilization、maturation 三步建立。到第三波左右，组织应开始因可见成果而主动加入。CEO 要防止前两波后的能量下降，定期 revisit、reset、recommit。

第 16Source for 16Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：Building a new business? There's an operating model for that（p.265Source for 265Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

新业务需要像 start-up 一样独立运行，同时调用母公司的品牌、客户、数据、渠道和资本优势。它需要独立领导、专属跨职能团队、自己的 P&L/融资节奏和快速决策边界，避免被母公司年度预算和控制流程拖慢。

执行上必须在 scrappy 与 scalable 之间保持平衡：早期用低成本验证需求，但要把技术债视为必须偿还的贷款；验证后及时建设平台、治理和可靠性。成熟后可有选择地并回母公司，但整合时既不能扼杀速度，也不能长期形成毫无协同的平行组织。

Section Four：Technology for Speed and Distributed Innovation（p.278Source for 278Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

现代技术环境的目标是让分布式团队无需反复申请基础能力，就能安全、快速、可复用地开发和部署。业务领导不必掌握底层实现，但必须理解平台、接口、工程自动化和核心系统取舍，因为这些决定创新速度。

第 17Source for 17Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：The platforms that power the enterprise（p.285Source for 285Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

平台不是一组软件，而是被产品化管理的组织能力。企业级 solution stack 包括用户应用/工作流、API 平台、数据平台、AI 平台、工程平台、基础设施与安全平台，以及核心企业系统。

平台团队需要明确内部客户、产品负责人、路线图、服务水平、采用和单位成本；只有被域团队实际复用，平台才创造价值。高管应按“让多少团队更快、多少能力被复用、价值实现时间是否缩短”管理平台投资。

第 18Source for 18Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：Make your tech flexible and scalable—hello APIs!（p.293Source for 293Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

API 把核心系统、数据和业务能力从点对点依赖中解耦，使应用可以像积木一样组合和替换。企业应建立 API-first 原则、统一分类与标准、网关、访问控制、开发者目录、复用与性能指标。

企业架构职能由静态审批者转成与域和平台团队共同设计模块化能力的实践者。解耦优先级应受业务域路线图驱动，而不是追求微服务或 API 数量。

第 19Source for 19Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：Engineering practices for speed and quality（p.301Source for 301Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

软件工程卓越依赖工程平台与自动化 SDLC：标准开发环境、版本控制、CI/CD、测试、制品管理、可观测性、IaC 和 DevSecOps。平台团队把这些能力做成自助产品，把开发者视为内部客户。

敏捷必须与工程自动化同步推广，否则团队只是更频繁地等待。生产环境要可审计、可靠、可扩展；安全向左移动到每次提交。技术债应被持续量化和偿还，不能用“速度”作为长期降低工程质量的理由。

第 20Source for 20Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：The AI revolution in software development（p.313Source for 313Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

AI 对开发的真正影响不是多一个编码助手，而是重构交付模式：从建议与补全，发展到多步开发 Agent，再到能够异步运行的 Agent factory。人类更多负责方向、架构、判断、验收和质量，Agent 承担大量生成、测试和修复。

落地需要三类准备：工程师实战升级；可被 Agent 理解的模块化代码、API 和文档；强测试、观测、安全和治理。书中提出“白天用于判断、夜间让 Agent 工作”的新节奏，并提醒不能靠聊天直接获得 production-grade 软件。作者要求领导思考：如果同样预算带来 20Source for 20Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 倍软件产出，哪些过去不经济的产品和流程将变得可行？

第 21Source for 21Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：An AI platform to accelerate model development（p.325Source for 325Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

AI 平台让团队复用模型、数据、工具、评估和护栏，把 AI 从手工项目变成工业化能力。六层架构包括 application patterns、development workflow、solution building blocks、AI models、automated guardrails 和 AI mesh。

获取路径有三种：开启 SaaS 内置 Agent，速度快但形成孤岛；购买平台，治理成熟但锁定和定制成本高；组合自建平台，投入最大但支持跨系统工作流和长期差异化。典型企业前 6-12 个月投入 500万-1,000Source for -1,000Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 万美元建立可用版本，完整平台约 12-24 个月、总投入 1,000万-3,000Source for -3,000Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 万美元。

先用 2-3 个高价值用例验证最小平台，同时从第一天建设开发工作流和 guardrails；用采用率、time-to-market、复用率、质量、安全和每千次请求成本管理。DBS 借助数据与 AI 平台把模型交付从 12-15 个月缩短到 2-3 个月。

第 22Source for 22Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：The art of developing agentic AI solutions（p.337Source for 337Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

Agent 适合动态、模糊、需要判断和多步协作的任务，不适合确定性强、规则明确、低价值或难以审查的流程。设计时应把端到端工作流拆成边界清晰的 constrained agents，为每个任务限定工具、数据、权限、输出格式和升级条件。

输入质量决定输出质量，应结合 RAG、语义层、业务规则和上下文工程。由于 Agent 非确定性，书中经验是约三分之一时间用于构建、三分之二用于测试；测试覆盖任务成功、边缘情况、幻觉、偏差、权限、成本和人工接管。购买 Agent 也要计算集成、训练、监控、token 和复用的完整成本。

第 23Source for 23Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：A more surgical approach to modernizing core systems（p.359Source for 359Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

核心系统现代化不是“先全部重写”或“完全不动”的二选一。正确问题是：哪些遗留约束正在阻塞优先业务域的价值？现代化顺序由域路线图、API 解耦、稳定性、数据访问和经济性决定。

先诊断真正痛点，再选择 retire、replace、replatform、refactor 或 retain；优先处理高业务影响、高故障风险和高变化频率的部分。韧性、云成本和 FinOps 必须进入商业案例。现代化是价值驱动的持续组合，而不是多年技术清仓项目。

Section Five：Embedding Data Everywhere（p.373Source for 373Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

数据工作的目标不是“建湖”，而是让结构化和非结构化数据可发现、可理解、可治理、可被应用和 Agent 安全消费。第二版进一步加入语义层/本体、LLM 数据保护和专有数据优势。

第 24Source for 24Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：Start with the value, then get the data（p.379Source for 379Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

数据路线图从业务价值倒推：业务域 → 解决方案 → 用例 → 数据产品/元素。多数企业已经拥有启动 MVP 所需的大部分数据，不应先做全量清洗；先识别关键元素，再按准确性、及时性、一致性、完整性、唯一性、连贯性、可用性、安全性和可解释性设定用途相关标准。

非结构化文档、邮件、图像和对话需要被分类、分块、嵌入、标注权限与来源。GenAI 可以辅助发现数据、生成元数据、检测质量问题和提出治理动作，但最终责任仍属于业务数据所有者。

第 25Source for 25Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：No data architecture, no AI advantage（p.391Source for 391Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

现代数据架构包含八项连接能力：data ingestion、data platform、semantic layer、data products、data consumption、access and governance、metadata、data operations。它应以最小可行架构支持优先用例，并保持模块化、可演进和实时链路一致。

GenAI 新增向量库/embeddings、Agent orchestration 与 retrieval、AI gateway/guardrails、自动化质量安全血缘、模型反馈闭环等能力。语义层把数据转成知识：ontology 定义什么是有意义和允许的，knowledge graph 动态连接现实实体和事件。二者让模型复用业务语义、减少幻觉、提高可解释性并支持治理。

第 26Source for 26Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：Data products: The reusable building blocks for scaling（p.401Source for 401Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

数据产品把一组经过治理的数据连同语义、质量标准、接口、SLA、元数据和负责人一起包装，使应用和团队可直接消费。它把数据从一次性交付改为持续经营的产品。

只有被多个高价值用例复用的数据才值得产品化；书中经验是企业大部分价值通常集中在 5-15 个数据产品。数据产品通过 pod 建设和迭代，以消费、复用、质量、满意度和所支持用例价值衡量；复用可使新用例交付最高加快约 90%Source for 90%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition。

第 27Source for 27Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：Organize to get the most from your data（p.417Source for 417Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

领先企业采用强中央能力与业务域责任结合的 federated model。中央团队负责战略、架构标准、平台、人才池、治理政策和风险框架；业务域负责数据产品、日常质量和具体消费。

关键角色包括 data domain owner、data product owner、data architect、data steward/quality roles 和平台/工程团队。治理通过明确所有权、统一定义、自动化监控和分级论坛建立信任；其目的不是增加审批，而是让数据能够更快、更安全地被复用。

第 28Source for 28Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：Protecting data in an LLM world（p.427Source for 427Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

LLM 和 Agent 扩大了数据移动、推断和二次使用的风险面。问题通常不是模型本身，而是身份、权限、检索、提示、输出、缓存、外部工具和 Agent-to-Agent 数据交换没有被系统治理。

企业需要机器身份与最小权限、AI gateway、内容过滤、敏感信息检测、可追溯日志、动态策略和自动化控制；对合作数据建立 rights catalog，明确用途、地域、保留、衍生内容和模型训练权利。保护必须随数据和模型调用自动执行，manual review 无法规模化。

第 29Source for 29Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：Your next strategic duty: Create a proprietary data advantage（p.437Source for 437Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

专有数据是 AI 时代的重要资产和竞争壁垒。业务领导应明确哪些数据支撑差异化决策，并像管理资本资产一样制定多年度获取、丰富、质量、权利和回报计划。

优先通过核心流程 instrumentation 自动捕获新数据，而不是周期性清理；将内部数据与外部伙伴、传感器、客户互动和模型反馈结合。价值不在数据量，而在独特性、相关性、持续更新、合法权利和能否改善业务模型。

Section Six：Adoption and Scaling: Your Force Multipliers（p.451Source for 451Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

MVP 演示成功并不等于价值成功。失败常见于只覆盖流程一小段、基于一次性技术和数据、没有用户激活方案。书中经验法则仍是：每 1Source for 1Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 美元开发投入，至少再准备 1Source for 1Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 美元用于采用和扩展。

第 30Source for 30Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：Make adoption stick（p.457Source for 457Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

采用从开发期开始，让最终用户和 subject matter experts 成为“共同驾驶员”，持续参与研究、测试和工作流设计。影响模型包括领导示范、基于角色的培训、可信变革故事、采用与绩效度量。

真正价值还要求改造上下游业务模式：流程、组织、技能、目标、激励和产能。Domain owner 对采用最终负责；复杂场景应建立长期 adoption team，在生产前准备培训、沟通、现场支持和反馈机制。

第 31Source for 31Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：Design solutions for scale and reuse（p.469Source for 469Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

规模化首先定义 scale unit：市场、工厂、客户群、产品线或业务单元；再按价值、准备度和实施难度决定 linear waves、exponential waves 或 big bang。

Assetization 把方案拆成可复用核心与可配置部分，并同时标准化代码、数据接口、模型、实施 playbook、培训、运维和支持。复用应在架构和预算中从第一天设计，不能等试点完成后再“硬化”。

第 32Source for 32Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：Ensuring impact by tracking what matters（p.483Source for 483Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

绩效架构把企业目标连接到业务/运营 KPI、解决方案和团队 OKR。应同时追踪三类指标：价值创造、team health、六项能力/变革进展；避免只统计上线数量和活动量。

用价值驱动树建立因果链，用阶段门决定继续、调整或停止，用 QBR 根据用户反馈、OKR 和财务证据重新拨款。工具应自动汇集交付、依赖、采用和价值信息，让转型办公室聚焦清除障碍而非制作报表。

第 33Source for 33Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：Plan for midstream adjustments（p.501Source for 501Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

转型在 1-2 年后常遇到中段问题：早期灯塔价值不足、核心系统约束显现、人才供给和平台成熟度落后、数据质量问题扩大、组织热情衰减。路线图必须是 living contract，而不是初期承诺的固定甘特图。

领导层应定期重估业务域价值、依赖、能力和投资节奏；区分可修复执行问题与需要改变路线的事实变化；保留高目标但调整路径。持续传播真实成果、解决顽固瓶颈和重新集中资源，是通过中段低谷的关键。

第 34Source for 34Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：Risk, trust, and the right to deploy AI（p.511Source for 511Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

部署 AI 的权利来自数字信任。流程包括风险分诊和分级、更新数据/模型/第三方/透明度政策、把政策转为自动化控制、提升全员风险识别能力。

GenAI 风险涵盖错误与幻觉、偏差、隐私/IP、网络安全、第三方和模型行为变化。风险专家应嵌入团队，缓解项进入 backlog；政策尽量转为 compliance/security/policy as code。Agentic AI 方案中，多达 50%Source for 50%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 代码可能用于 guardrails 和降低风险，说明安全不是附加成本，而是产品主体的一部分。

第 35Source for 35Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：So, what about culture?（p.523Source for 523Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

文化是六项能力长期运行后的结果，不是单独的口号项目。企业要从“技术是支持部门、年度大项目、定制系统、外包工程”转向“技术进入战略、持续迭代、平台/API 复用、内部工程能力”。

文化改变的抓手包括领导评价和激励、全员基础学习、关键领导深度升级、对受 AI 影响最大的 pivotal roles 进行 3-9 个月再培训，以及用真实业务成果塑造故事。心态改变最快的方式，是领导改变奖励、任用和资源配置。

Section Seven：Transformation Journey Stories（p.538Source for 538Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

第二版用四家公司展示六项能力的组合效果：LATAM 与 Toyota 为新增案例，Freeport 和 DBS 为第一版案例的“第二季”。共同点是先建设基础能力，再以业务域持续扩大价值；没有一家认为转型已经完成。

第 36Source for 36Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：LATAM Airlines: Creating an 'airline in your pocket'（p.543Source for 543Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

LATAM 先围绕 passenger experience 建立独立数字组织，首批约 120Source for 120Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 名工程、架构和 UX 人才组成产品团队，以“airline in your pocket”重构端到端旅客体验。技术基础、产品运营模式和客户数据使多市场复制成为可能；转型 18Source for 18Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 个月达到盈亏平衡，36Source for 36Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 个月时收益 run-rate 超过最初商业案例两倍。

随后能力扩展到飞机维护。企业建设数据平台和业务主导团队，用健康数据、零件模型和 Agent 辅助维护；零件库存减少超过 1Source for 1Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 亿美元，同时降低缺件风险。案例说明，成功的客户域可以成为进入运营域和 Agentic workflow 的能力跳板。

第 37Source for 37Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：Toyota: From kaizen improvements to big leaps in value（p.555Source for 555Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

Toyota North America 从渐进 kaizen 迈向端到端供应链域重构，初始商业机会约 5Source for 5Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 亿美元。团队围绕需求预测、车型组合、生产、库存、促销和经销商透明度建立相互连接的模型和工作流。

数字孪生可在 15Source for 15Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 分钟内处理 10Source for 10Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 亿个决策变量，为每年约 250Source for 250Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 万辆、约 5,000Source for 5,000Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 种配置优化计划；过去过程需要大团队、约 30Source for 30Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 天和大量表格。项目首年工作资本下降 20%Source for 20%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition、利润提高约 2Source for 2Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 亿美元；数字孪生累计带来约 8Source for 8Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 亿美元利润，库存从约 60Source for 60Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 天降至 20-25 天。业务专家嵌入团队和可解释界面使模型被规划人员信任。

第 38Source for 38Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：Freeport-McMoRan: Harnessing Tech & AI for mining innovation（p.567Source for 567Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

第一阶段 Americas' Concentrator 已让六大资产产量提高 10%-15%Source for -15%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition，无重大新增资本投入，每年增加约 2Source for 2Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 亿磅铜。第二版重点描述 Leach to the Last Drop：把 AI、物理模型和数字孪生用于持续 150-300 天、传感器更少、更复杂的浸出过程。

HELENA 平台连接数据、专利算法、3D stockpile 表示和现场经验，帮助企业重新评估约 390Source for 390Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 亿磅过去被认为不可回收的存量铜。项目首年试点增加约 3,500Source for 3,500Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 万磅铜，最终超过每年 2Source for 2Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 亿磅目标，并把长期目标提高到每年额外 8Source for 8Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 亿磅、超过 10Source for 10Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 亿美元 EBITDA 潜力。核心经验是复用人才和平台、把现场操作员嵌入建模、用实际价值建立信任。

第 39Source for 39Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 章：DBS Bank: A multinational bank goes all in on AI（p.581Source for 581Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition）

DBS 把 AI 当作长期经营能力，而非模型项目。统一数据平台、数据产品、自动化权限、AI 平台、平台化运营模式和广泛技能建设，使模型从 2018 年的 12-18 个月交付缩短到 2-3 个月。

银行在客户营销、风险、运营和服务中工业化 AI：2022 年一个解决方案创造近 1.24Source for 1.24Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 亿新元价值；2023 年已有 300Source for 300Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition+ 生产模型，创造近 4.54Source for 4.54Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 亿新元；到 2025 年 AI 经济影响估计超过 10Source for 10Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 亿新元。DBS 继续发展 GenAI 助手与 Agent 工作流，说明能力复利比单个模型更重要。

六项核心企业能力：第二版的升级含义

| 能力 | 第二版“行业最佳”状态 | Agentic 时代的新前沿 | 主要管理指标 | |---|---|---|---| | 业务路线图 | 业务负责人端到端重构域 | Agentic workflows/systems 重构 | 域 KPI、现金回报、stage-gate 通过率 | | 人才 | 业务与 IT 领导均具技术能力，核心人才内部化 | Orchestrator、agent coach、system integrator | 70-70-70、关键技能密度、学习与流失 | | 运营模式 | 业务、工程、运营和控制职能嵌入域/平台 | Human-Agent workflow 作为新组织单元 | 决策速度、资源重配、团队交付周期 | | 技术 | 现代化分布式环境和复用平台 | AI 驱动软件开发、AI mesh、Agent factory | time-to-production、复用率、自动化与单位成本 | | 数据 | 业务和模型可自助消费的数据产品 | 专有数据优势、语义层、非结构化数据 | 数据产品消费、质量、权利、模型反馈 | | 采用与扩展 | 流程被重构，方案从设计时可复制 | Agent orchestration 嵌入经营系统 | 使用率、规模化周期、运营/财务价值、信任 |

六项能力必须同步推进。平台没有高质量人才不会被用；模型没有数据产品无法扩展；产品没有业务流程和激励变化不能兑现价值；业务域没有财务目标又会退化为试点集合。

业务主导的转型实施路线图

0-3 个月：建立信念与经济边界

1. CEO 组织至少 20Source for 20Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 小时的高管共同学习和跨行业参访。
2. 用行业趋势、经济杠杆、竞争优势和战略飞轮形成 3-5 年愿景。
3. 以成功画像校准目标：追求约 20%Source for 20%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 级别的经营改善，而不是 1%-2%Source for -2%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 的项目组合。
4. 列出 10-20 个业务域，按价值和可行性选出 2-5 个候选，实际先推进 1-3 个。

3-6 个月：完成域重构与 C-suite 契约

1. 由 domain owner 运行五步 reimagination，形成价值杠杆、5-15 个用例和 Level 1-3 路线图。
2. 按 DVD 五阶段设定时间、团队、stage gates、收益和 build/run 成本。
3. 同步评估六项能力成熟度，制定 Crawl/Walk/Run 能力计划。
4. 把域、平台、变革管理、财务与 C-suite sponsor 写入 2-3 年实施计划。

6-18 个月：交付灯塔价值并建设基础能力

1. 建专属跨职能团队；以短 sprint、用户反馈和工程平台进入生产。
2. 内部化关键工程人才，培养 N-2Source for -2Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition/N-3Source for -3Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition domain owners，启动 70-70-70 IT 人才转型。
3. 按优先用例建设 API、数据产品、最小数据/AI 平台和自动化 guardrails。
4. 在 POC/production 阶段启动采用团队、流程重构和扩展设计。

18-36 个月：扩大业务域与运营模式

1. 资产化已验证方案，按 scale unit 分波复制。
2. 从 digital factory 逐步进入 Domain & Platform，采用 persistent funding 和 QBR。
3. 用价值树、stage gates 和自动化绩效系统重新拨款；及时修订 living roadmap。
4. 将临时转型办公室的机制嵌入日常经营。

36Source for 36Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 个月以后：Agentic 与专有数据复利

用既有平台、语义层、业务专家和治理能力进入 Agentic workflows/systems；围绕专有数据和模型反馈建立持续增强的业务飞轮。转型由项目变成经营方式。

CEO 与高管团队的角色和责任

| 角色 | 不可委托责任 | 典型投入/节奏 | |---|---|---| | CEO | 形成信念、选业务域、解决跨职能矛盾、共同问责和激励 | 每月 2-4 天，启动期更多 | | Domain/业务负责人 | 重构流程、定义 KPI、监督交付、采用与价值落袋 | 优先域可投入 20%-40%Source for -40%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 时间 | | Transformation Officer | 维护路线图、stage gates、资源、依赖和价值纪律 | 直接向 CEO 汇报约 2-3 年 | | CIO/CTO/CDO | 核心系统、平台、架构、现代工程、数字产品 | 每 90Source for 90Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 天展示可见平台与交付进展 | | CAIO/CDAO | AI 平台、模型能力、风险和数据协同 | 不能替代业务 owner 或孤立 AI 人才 | | CHRO | 域负责人升级、70-70-70、技能体系、职业路径、Human-Agent 准备 | 将人才指标纳入转型仪表盘 | | CFO | 商业案例、动态资金、价值验证、运营模式财务机制 | 季度重配，不只年度预算 | | CRO/法务/安全 | 风险分诊、自动化 guardrails、政策更新与数字信任 | 从设计期进入 backlog 和流水线 | | 董事会 | 长期资本、领导团队质量、技术与 AI 风险监督 | 至少 2-3 名有真实转型经验的董事 |

高管不需要成为程序员，但必须懂到足以做经济取舍、识别关键人才、理解平台依赖并挑战团队。任命一个 CAIO、买一个平台或批准一批 Agent 试点，都不能替代业务负责人和 CEO 的经营责任。

案例对比与实践经验

| 公司 | 优先业务域 | 能力建设主线 | 代表性结果 | |---|---|---|---| | LATAM Airlines | 旅客体验，后扩至维护 | 独立数字组织、产品团队、数据平台、跨市场复用、Agent 辅助维护 | 18Source for 18Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 个月盈亏平衡；36Source for 36Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 个月收益超过原案例两倍；零件库存减少 1Source for 1Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 亿美元以上 | | Toyota North America | 端到端供应链 | 业务专家+数据科学、数字孪生、统一界面、经销商协同 | 首年利润 +2Source for +2Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 亿美元；数字孪生约 +8Source for +8Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 亿美元利润；库存 60Source for 60Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 天降至 20-25 天 | | Freeport-McMoRan | 选矿与浸出 | 现场专家嵌入、数据/AI 平台、物理模型、3D 数字孪生、持续实验 | 六大资产产量 +10%-15%Source for -15%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition；浸出目标提升至每年额外 8Source for 8Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 亿磅铜 | | DBS Bank | 客户、风险、运营和服务 | 平台运营模式、数据产品、AI 平台、自动权限、广泛技能建设 | 模型交付 12-18 月降至 2-3 月；2025 年 AI 价值估计超过 10Source for 10Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 亿新元 |

共同经验

1. 从经营问题和经济杠杆出发，不从模型或供应商出发。
2. 先以一个业务域证明价值，再复用平台、数据、人才和治理进入新域。
3. 业务专家必须嵌入团队，既提高模型质量，也建立一线信任。
4. 价值是能力复利：第一阶段的 API、数据平台和工程人才，使第二阶段 AI/Agentic 项目更快、更大。
5. 采用和规模化与开发同等重要，组织、流程和激励必须同步改变。

关于乐高集团

乐高集团是第一版的案例之一，但不在这份 2026 第二版 PDF 的 Section Seven 中。第二版以 LATAM 和 Toyota 替换/扩展案例组合，并保留 Freeport 与 DBS 的后续进展。因此本报告没有把第一版乐高内容冒充为第二版原文。

建议的企业自检动作

1. 统计当前 10-20 个业务域及其经济杠杆，而不是先统计 AI 用例数量。
2. 选一个域，用第 4 章 DVD 模板在 8-12 周内完成重构、价值树和 stage-gated business case。
3. 盘点 N-2/N-3 中真正能监督技术交付并对采用负责的 domain owners。
4. 用 70-70-70 检查 IT 人才结构，并确认 differentiating/high-velocity/utility 系统的内外部边界。
5. 对 Agentic workflow 使用价值、适配性、正确性、准备度和复杂度五项筛选，不以 Demo 效果决定投资。
6. 把 API、数据产品、语义层、AI 平台和 guardrails 的投资绑定到优先域用例，而不是先做全量平台。
7. 在预算中为采用和扩展配置不低于开发投入的资源，并明确 domain owner 对价值落袋负责。

值得继续研究的问题

• 你所在企业的经济杠杆点是什么，哪些 1-3 个业务域最接近书中“高价值、高可行”的甜点区？
• 哪些系统是 differentiating、high-velocity 或 utility，现有内外包边界是否与战略意图一致？
• 哪些工作流适合 Agentic automation，哪些其实用规则引擎、传统 ML 或 RPA 更可靠、更便宜？
• 当前数据语义是否足以让人和 Agent 对“客户、订单、风险、资产”等核心对象形成同一理解？
• 当软件开发产出可能提高 10-20 倍时，组织真正的瓶颈会转向需求判断、数据、采用、风险还是资本配置？

阅读说明与边界

• 本报告仅基于用户提供的 2026 年修订扩展第二版 PDF，书内页码按正式目录和正文标注。
• 原书将 20Source for 20Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 家成功企业作为“成功转型画像”，存在选择成功样本的目的，不应把 20%Source for 20%Source: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition EBITDA、3xSource for 3xSource: Rewired: McKinsey's Playbook on How Leading Companies Win with Technology and AI, Revised and Expanded Second Edition 回报机械当作所有企业的预测值。
• Agentic AI 章节描述的是截至成书时的早期实践。作者多次强调技术仍不成熟，完整工作流难以生产化，企业应保持 optionality 并通过实际业务试验学习。
• 报告中的跨章路线图和自检动作是对原文框架的结构化整合；具体行业、监管和组织约束仍需单独评估。