Duke Energy Reimagines Finance Workflows to Drive AI Value

Duke Energy重塑财务工作流以释放AI价值 将AI智能体直接融入工作流程，比单纯在现有人工流程上附加自动化能带来大得多的价值。 在本次讨论中，Cindy Lee探讨了财务与会计专业人士面对技术颠覆时的本能焦虑，并强调领导者绝不能闭门造车，而必须积极带领整个团队共同踏上变革之旅。Audi Rowe重点介绍了EY的跨行业AI转型框架，对比了单纯的“外挂式”自动化与“内置式”思维——后者通过“人在回路”的持续人机协作来重塑工作流程。会议最后肯定了Lee的前瞻性领导力，并邀请与会者与EY取得联系，以更详细地探讨其AI转型方法论。 财务和会计专业人士面对技术颠覆和运营转变时自然会产生焦虑，因此领导者切不可惧怕变革，而应主动迎难而上。 (17:40) AI转型不可能在孤立或各自为政的项目中取得成功；领导者必须积极参与，并带领整个团队一同推进转型。 (17:40) EY开发了一套跨行业AI转型框架，重点在于明确成效、设定路线图里程碑，并通过保持“人在回路”的人机协作来重构工作模式。 (18:02) 仅仅将AI“外挂”到现有任务上只能带来渐进式的效率提升，而“内置式”思维则围绕AI重塑价值流，从而推动根本性增长并带来持久的业务价值。 (18:02) Cindy Lee在Duke Energy推行的方法堪称典范，它将AI直接嵌入企业运营模式，而不仅将其视作简单的自动化工具。 (18:57)

Realizing transformative value from enterprise AI requires moving beyond isolated task automation to fundamentally restructure operating models around human-agent collaboration, while actively guiding risk-averse teams through the transition. Leading Change in Risk-Averse Finance Functions (17:40) Cindy Lee observes that finance and accounting professionals frequently experience anxiety regarding technological disruption and automated workflows. Rather than avoiding these shifts, Lee advises leadership to lean into change while avoiding top-down or isolated implementations. She emphasizes that sustainable adoption requires collaborative leadership that actively guides the entire team through the journey to secure buy-in across traditional business functions. Shifting from Bolt-On Automation to Built-In Workflows (18:02) Drawing on EY's collaboration with Duke Energy, Audi Rowe outlines a cross-industry framework centered on defining clear outcomes, establishing roadmap milestones, and redesigning workflows around human-agent collaboration. Rowe contrasts traditional "bolt-on" implementations—which apply basic automation to existing processes for incremental efficiency—with a "built-in" mindset. In Rowe's view, restructuring value streams to keep human workers continuously in the loop alongside AI agents is essential for driving deeper operational transformation and value realization. Rowe notes that Duke Energy's initiative represents an ongoing effort with further value still to be unlocked, framing the embedding of AI directly into corporate operating models as a continuous journey rather than a completed deployment.

安永（EY）的 Audi Rowe 与杜克能源（Duke Energy）的 Cindy Lee 认为，企业推进智能体（Agentic）转型的关键在于超越单纯的任务自动化，通过流程重塑与人机协作建立“内生式”架构，并以协作型领导力引领易产生焦虑的团队拥抱变革。 化解传统职能团队对技术转型的阻抗与顾虑 (17:40) Cindy Lee 指出，财务与会计团队面对运营机制及自动化流程的转变时，往往天然抱有风险担忧与抗拒情绪。她主张管理人员与其退缩，不如主动迎难而上（lean into it）。同时她强调，技术采纳无法在信息孤岛中单独实现，成功的转型要求管理者积极引导并带领全体团队成员共同推进，通过全员协作获取共识与支持。 以内生式重构替代外部拼凑的智能体转型框架 (18:02) Audi Rowe 基于与杜克能源的合作总结了一套适用于多行业的转型框架，核心在于明确预期成果与路线图里程碑。Rowe 明确区分了两种实施路径：仅对现有流程中的独立任务进行简单自动化被称为“外部拼凑”（bolt-on），该方式仅能产生渐进式的效率提升；而“内生构建”（built-in）思维则围绕 AI 重新设计端到端价值流。该框架要求建立人机协作（human-agent collaboration）机制并确保人工在环（humans in the loop），以此重塑业务运作模式并释放深层次价值。 Rowe 评价 Cindy Lee 在杜克能源的实践体现了将 AI 深度融入运营模式的价值，但指出该转型仍处于进行中，未来仍有更多潜在价值尚待释放；讨论未进一步披露具体的落地成本、风险控制指标或技术选型细节。

## Background 本视频由 Ai4 于 2026 年 8 月 16 日在 YouTube 平台发布。内容记录了安永（EY）美洲区 AI 体验、战略与转型咨询负责人 Audi Rowe 与杜克能源（Duke Energy）财务转型高级副总裁 Cindy Lee 的炉边谈话，重点探讨了杜克能源如何将财务部门打造为企业级 AI 落地的催化剂，通过在运营模型中深度嵌入 AI、治理与信任机制，推动企业从零散试点迈向 Agentic 智能企业。 ## TL;DR 面对未来 10 年高达 2000 亿美元的电网基础设施改造计划，杜克能源选择以强管控的财务职能为破局点推进 Agentic 智能化转型。依托与安永深度合作打造的 FDI 财务数据湖底座与 SOX 合规认证，团队将复杂手工日记账凭证的处理耗时从近 5 小时缩短至 30 分钟，并在不盲目扩充人员的前提下重塑团队产能。通过坚持“人在回路（Human-in-the-loop）”机制化解一线员工的怀疑，财务人员得以从繁重的数据搜集中解放出来，转变为赋能现金流优化与重大资本决策的战略业务伙伴，生动践行了“内置原生”而非“外挂叠加”的根本性变革范式。 ## Atmosphere & style 整场炉边谈话氛围专业务实且轻松幽默，主持人 Audi Rowe 与在杜克能源任职 24 年的资深财务领军人 Cindy Lee 之间互动默契、侃侃而谈。叙事线从传统财务人员对技术的天然审慎怀疑，自然演进为亲见风控闭环与工作负荷降低后的拥抱与信任，兼具公用事业严谨合规的底色与大胆变革的魄力。听众仿佛置身于一场高管私享复盘会，既能感受到统筹 2000 亿美元庞大资本的宏观战略格局，又能体会到聚焦具体凭证编制与员工福祉改善的真切温度。 ## Key takeaways - [00:04](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=4s) 企业推进 AI 变革应摆脱零散孤立的技术用例堆砌，转向以业务成果为导向的全局价值重塑。 - [01:31](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=91s) 面对未来 10 年 2000 亿美元的能源基础设施现代化投资，必须率先重塑底层陈旧的财务流程，以深度洞察驱动高稀缺资本的最优配置。 - [03:13](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=193s) 转型需以深厚信任为基石，通过在长期外包服务中嵌入转型条款并搭建合规财务数据湖（FDI），为 Agentic AI 落地铺平道路。 - [05:43](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=343s) 推行“内置原生（Built-in）”而非“外挂叠加（Bolt-on）”思维，坚决避免将陈旧低效的旧流程直接上云，而是从底层对跨实体业务流进行彻底重塑与标准化。 - [05:43](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=343s) 财务数据天生具备严密的合规控制与强治理属性，是企业实践与验证 Agentic 架构最稳健可靠的切入场景。 - [07:55](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=475s) 在强监管与严格审计环境下，必须通过 SOX 认证及贯彻“人在回路（Human-in-the-loop）”机制，确保 AI 结果完全可追溯并留存完整审计标记。 - [07:55](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=475s) 智能体深度协同能够化解复杂会计痛点，如将涉及 13 个数据源与 20 个工作表的手工凭证编制耗时从 4 至 5 小时大幅压缩至 30 分钟。 - [10:52](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=652s) 变革管理的关键在于主动吸纳心存疑虑的团队成员，用透明的控制流程消除焦虑，并将一线员工每周工作时长从 60 小时有效降至 45 小时。 - [12:12](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=732s) 加速财务结账与现金流洞察的战略核心，在于让高管层提前研判债务与流动性指标，从源头保障用电客户的能源长期可负担性（Affordability）。 - [14:00](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=840s) 应对业务激增应坚持“重塑产能而非扩充编制”，由 AI 接管繁重的数据搜集与底稿整理，释放财务人员转型为推动“1 变 7”价值创造的业务战略伙伴。 - [15:51](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=951s) 跳出单纯“硬性砍成本”的认知误区，AI 的高阶价值在于决策敏捷度与“避免不必要成本”（Avoided Cost），防止无效支出转嫁给终端客户。 - [18:02](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=1082s) 迈向 Agentic 企业的普适范式在于将 AI、治理规范与信任底座深植于运营模型的核心，从而实现超越传统自动化修补的指数级跃迁。 ## Detailed walkthrough ### [00:04](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=4s) 开场与背景：从孤立用例转向以成果为导向的 AI 变革 会议在轻松的氛围中拉开序幕。安永（EY）负责美洲区 AI 体验、战略与转型咨询的合伙人 Audi Rowe 登台打趣道，他们原本以为只是来做一场私密的炉边谈话，结果现场坐了“大概 500 位最要好的朋友”。在进入正题前，Audi 阐述了当前企业界普遍面临的 AI 困境：企业该如何真正从 AI 中捕获业务价值？如何跳出零散、孤立的技术用例（use cases），转而聚焦于业务成果（outcomes）并重塑实现路径？为了避免单调枯燥地翻阅上百页 PPT，他邀请了 Duke Energy 的转型领军人物 Cindy Lee，通过真实的变革故事来展现这一演进路径。 Cindy Lee 在 Duke Energy 拥有深厚而卓越的履历，已在公司任职 24 年，曾先后担任首席会计官（Chief Accounting Officer）和主计长（Controller），目前全面领导财务转型（Finance Transformation）工作。作为全美领先的能源供应商之一，Duke Energy 当前为超过 1000 万客户提供服务。Audi 表示，正是 Cindy 在财务与管理领域的长期积淀，为这场由财务驱动的企业级 Agentic 智能化转型奠定了关键基石。 ### [01:31](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=91s) 2000 亿美元电网投资下的财务重塑 谈话首先切入了一个极具震撼力的数字：Duke Energy 计划在未来 10 年内投资 2000 亿美元（$200B）用于重构电网与能源基础设施。Audi 指出，过去整个能源基础设施历经了 80 年的漫长建设，而如今要在短短 10 年内完成如此大规模的现代化改造，其挑战与体量不可谓不庞大。面对这一宏大蓝图，变革为何要首先从财务部门破局？ Cindy 解释称，当企业致力于重塑电网并铺设各类尖端设备时，核心问题在于“如何为其提供资金支持”。尽管 2000 亿美元听起来是一笔天文数字，但分摊到公司业务覆盖的 6 个州之后，资本仍然是高度稀缺的资源（scarce resource）。为了帮助领导层在海量项目中做出最优抉择，财务必须提供由深度洞察驱动的决策支持（insight-led decisions）。Cindy 风趣地自嘲道，公司原有的财务流程“可能和我们的电网一样古老，甚至比电网还要老”，因此，若要在基础设施上实现跃迁，必须先从最底层的财务流程重构起步。 - **投资规模与周期**：未来 10 年计划投资 2000 亿美元，用于全面重构能源基础设施。 - **业务覆盖版图**：跨越全美 6 个州，服务客户总量超过 1000 万户。 - **财务核心诉求**：在资本约束下，以洞察驱动投资决策，优先保障最佳项目的资源倾斜。 ### [03:13](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=193s) 奠定信任基石：从税务与主计托管到 Agentic 智能演进 两人的对话进一步深入到转型的切入策略。Audi 提到，Duke Energy 并没有盲目追逐光鲜亮丽的“时髦技术”，也没有“拿着 AI 这把锤子到处找钉子”，而是采取了极具针对性的战略推进，其中“信任”成为了至关重要的底色。Cindy 回顾了 Duke 与安永长达数年的合作演进历程： 早在 2018 年，双方就启动了税务托管服务（Tax Managed Service）合作。当时 Duke 进行了大力度的组织重塑，将几乎全部税务职能外包托管给安永，内部仅保留了 4 名核心人员。作为重资产的公用事业企业，Duke 涉及庞大的财产税（property tax），通过与安永团队的深度协作与税务优化，直接为公司创造了数亿美元（hundreds of millions of dollars）的真实真金白银现金节省。这一战役让安永深入理解了 Duke 的业务系统、运营架构乃至团队文化。基于此，大约 4 年前，Duke 在寻找主计业务托管（Controllership Managed Service）伙伴时再次选择了安永，并在布宜诺斯艾利斯和印度建立了支持团队。 值得注意的是，主计业务的合同中一开始就明确嵌入了“转型框架条款（transformation construct）”，并非单纯的人力外包。团队借此在过去几年中推进了数百项流程优化与业务标准化。在此坚实基础上，Duke 进一步引入了名为 FDI（Finance Data Infrastructure）的数据湖层工具，将管控严格、高度合规的总账（General Ledger, GL）及各类财务数据注入数据湖，从而能够以远超企业单打独斗的速度，迅速将 Agentic AI 能力无缝嫁接到核心财务场景中。 ```mermaid flowchart TD accTitle: Duke Energy 财务数据与 Agentic 架构演进 subgraph S1["数据底座与标准化"] A["6家子公司统一 GL 总账"] --> B["FDI 财务数据湖层<br/>(SOX 合规认证)"] end subgraph S2["知识智能与 Agentic 协作"] B --> C["智能体应用层 (Agentic AI)"] C --> D["人机协同审核 (Human-in-the-loop)"] end subgraph S3["业务价值释放"] D --> E["复杂手工凭证编制<br/>(工时降低90%)"] D --> F["资本决策与现金优化"] end ``` ### [05:43](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=343s) “内置”而非“外挂”：重塑业务流程与分层架构 Audi 总结道，Cindy 在主计长视角的转型中展现了一种鲜明的思维范式——专注于“业务成果”而非单纯将现有操作自动化。换言之，Duke Energy 践行的是将 AI 深度植入流程的“内置（built-in）”思维，而非浮于表面的“外挂（bolt-on）”模式，同时始终将“人”置于核心位置。 Cindy 对此高度认同。她强调，即使只是将总账系统迁移上云，也绝不能简单地把陈旧低效的旧流程照搬上去。Duke 旗下拥有 6 家不同的公用事业实体，尽管此前已将它们整合迁移到了同一套 GL 总账系统上，但各实体的具体操作流程依然存在差异。因此，转型的首要任务是夯实底层基础，使所有团队“朝同一个方向划桨”，在流程标准化的同时进行彻底重塑，甚至“有时需要将原有流程彻底颠倒过来”。财务之所以是绝佳的切入点，是因为财务数据天生具备严密的治理与受控特性，一旦将坚实的数据资产与 FDI 智能层结合，Agentic AI 和软件智能体就能以最高的效率释放出自动化与分析潜力。 - **转型方法论**：拒绝“旧流程上云”与“外挂式 AI”，采取“内置式重塑”（Built-in Transformation）。 - **标准化挑战**：统筹 6 家公用事业子公司，在统一总账的基础上消除流程差异。 - **财务天然优势**：具备高度管控与受控的数据资产，是落地 Agentic 架构最稳健的试验田。 ### [07:55](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=475s) 强监管下的负责任 AI 与人机协同实践：30分钟完成手工日记账 公用事业属于强监管行业，合规与审计要求极其严苛。Cindy 直言，财务人员天生带着审慎甚至“充满怀疑（distrustful）”的职业习惯，对系统的受控性要求极高。因此，如何构建负责任的 AI（Responsible AI），确保智能体按照预期精准执行，成为了整个项目的生命线。 Cindy 透露，团队首先联合外部审计师及内部审计团队，使 FDI 平台全面获得了 SOX（萨班斯-奥克斯利法案）合规认证，并在网络安全与权限治理上构筑了严密防线。更关键的是在整个链条中始终贯彻“人在回路（Human-in-the-loop）”机制，让一线财务团队全程深度参与，拥有充分的话语权。这种信任并非一蹴而就，而是在一次次可感知的实操成果中逐渐沉淀。 | 评估维度 | 传统手工编制模式 | FDI 驱动的 Agentic 模式 | |:--- |:--- |:--- | | **支持数据源** | 散落于约 13 个独立数据支撑文件 | 智能体自动关联拉取多源文件 | | **底稿复杂度** | 包含多达 20 个工作表（tabs）的复杂 Excel 表格 | 智能体结构化输出并保留所有计算公式 | | **透明度与审计轨迹** | 需人工逐项检查勾稽关系与标记（tick marks） | 完整回溯数据来源，清晰呈现所有推导逻辑与勾稽标记 | | **端到端耗时** | 约 4 至 5 小时 | 约 30 分钟（效率提升近 90%） | Cindy 分享了一个极具代表性的实战案例：在处理一笔复杂的手工日记账凭证（manual journal entry）时，传统流程需要调用大约 13 个不同的辅助数据文件，并在一份包含近 20 个工作表的庞大 Excel 中进行复杂计算，通常需要耗费资深人员近 5 个小时。引入 FDI 智能体后，Agent 能够自主读取并拉取全部所需数据，自动生成记账凭证。最令会计经理信服的是，智能体不仅给出了结果，而且完整保留了所有计算公式，清晰标注了数据来源以及会计核查所需的全部勾稽标记（tick marks）。审核人员能够像审查人类底稿一样，一目了然地追溯智能体的推导全过程。这让单笔复杂凭证的处理耗时从 4 到 5 小时骤降至仅需 30 分钟，彻底打消了一线团队的疑虑，将“怀疑”转化为了拥抱 AI 变革的巨大热情。 ### [10:52](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=652s) 变革管理与信任构建：让怀疑者入局并重塑工作体验 在探讨人机协作的角色演变时，变革管理与组织文化往往决定了技术落地的成败。针对如何让一线财务人员建立对新技术的信心，杜克能源（Duke Energy）在推进财务转型过程中积累了深刻的实践心得。建立信任绝不能仅靠单向的说教与宣贯，而是必须让员工亲身参与到整个转型流程之中，亲眼目睹技术背后的控制机制与实际带来的改变。当团队成员能够清晰看到完整的审计追踪（Audit Trail）与风控闭环时，内心的疑虑才会真正消解。 在推动文化变革的过程中，不仅要调动那些对技术充满热情、迫不及待想要上手尝试的“积极派”，更关键的是要把那些心存疑虑的“怀疑派”一同带入这场旅程。通过让他们亲身参与并切实体会到“人类依然在环”（Human-in-the-loop），这些员工能够直观感知到自己的岗位并非被替代，而是在升级与演进。这种演进不仅让他们有精力去思考更具战略意义的议题，还带来了切实的个人福祉——许多员工得以从每周高压工作 60 小时的泥潭中解脱出来，回归到 45 小时左右更为平衡健康的工作节奏。 - **全员参与机制**：既要激发技术先锋的热情，更要将怀疑者纳入转型流程，用透明的控制流程消除顾虑。 - **改善工作与生活平衡**：通过自动化与智能代理分担重负，将部分岗位每周工作时长从 60 小时有效降至约 45 小时。 --- ### [12:12](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=732s) 支撑 2000 亿美元资本计划：以敏捷洞察赋能财务结账与现金流 在推进大规模技术投资时，向首席财务官（CFO）清晰证明转型的商业价值是维持战略定力的关键。杜克能源目前已经成功将月度财务结账时间缩短了整整一天，未来的目标是进一步再缩短数天。然而，比单纯节省结账时间更具战略价值的，是对公司庞大资本支出的敏捷支持。杜克能源当前面临高达 2000 亿美元的宏观资本规划（$200B Capital Plan），在这一规模下，现金管理以及运营资金对债务比率（FFO-to-debt metrics）等关键财务指标显得尤为重要。 历史上，杜克能源的现金流数据汇总往往相对滞后。为了改变这一被动局面，团队在现金流流程内部部署了多个蓝图与智能分析方案，大幅加快了现金流信息的生成速度。这使得 CFO 能够在向公开市场投资者与分析师做汇报之前，有充裕的时间提前审视、消化这些关键洞察。转型的真正价值不仅在于节约成本，更在于洞察获取的速度与质量——帮助业务部门更快地为客户筛选出最具性价比的优质资本项目，从源头上保障能源公用事业的“可负担性”（Affordability），确保这 2000 亿美元的每一分投入在长期周期内都合理高效。 ```mermaid flowchart TD accTitle: 杜克能源财务转型价值流转图 subgraph S1["数据与流程优化"] A["月末结账流程提速<br>(已缩短1天，目标再减数天)"] B["多维度现金流分析蓝图"] end subgraph S2["核心财务与决策指标"] C["FFO-to-Debt 指标优化"] D["CFO 提前获取关键洞察"] end subgraph S3["战略与业务成果"] E["支撑 2000 亿美元资本计划"] F["严控客户用能成本<br>(确保 Affordability)"] end A --> D B --> C C --> D D --> E E --> F ``` --- ### [14:00](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=840s) 人才升级与工作重构：让 AI 负责数据整合，释放人员战略思考力 面对 2000 亿美元的庞大资本开支，传统的管理惯性往往是大幅扩招会计与财务人员来支撑激增的业务量。然而，杜克能源确立的核心战略是“不增人头，重塑产能”（Repurpose Capacity）。通过智能工具重构工作流，企业可以果断停止那些不再创造实质价值的繁琐杂务，让现有团队在技能重塑（Reskilling）后流向最需要人才的战略高地。 长期以来，会计与财务行业耗费了海量工时在繁重的数据搜集与报告撰写上，根本没有充裕的时间停下来深入思考。借助 AI Agent（智能代理），这些数据汇总与基础分析工作可以在极短时间内高质量完成。这让财务人员得以转型为业务部门的深度战略伙伴，共同推演“如何花好手头的每一块钱，从而为股东争取赚回七块钱”的价值创造路径，实现从后勤记账向业务增长引擎的根本跨越。 - **产能重塑而非人员膨胀**：在 2000 亿美元资本计划的重压下，坚持不依靠大规模扩充编制，而是通过 AI 释放现有团队潜能。 - **剥离低价值劳动**：将大量数据搜集、底稿汇总与报表起草任务移交 AI Agent 快速完成。 - **业务战略合伙人**：财务团队转向深度赋能业务决策，推动“投入 1 美元创造 7 美元回报”的高质量资本配置。 --- ### [15:51](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=951s) 走出单纯降本误区：以决策速度与避免不必要成本驱动业务增长 在与众多企业高管（C-suite）沟通时，普遍存在一种认知偏差：过多管理者将 AI 转型窄化为“如何从业务中硬生生砍掉 1 亿美元成本”，而忽略了技术在创新与增长维度上的巨大潜力。对于刚开启 AI 转型之旅的企业，降本固然是一个考量因素，但绝非全部。真正的高阶价值在于依托深度洞察所驱动的敏捷决策——在某项试验可能面临巨额亏损之前迅速识别并及时止损，以及赋能领导层以极高的响应速度采取行动。 此外，“避免不必要成本”（Avoided Cost）同样至关重要。对于杜克能源这样的公用事业企业而言，避免无效开支就意味着不必将不必要的成本转嫁给广大用电客户。财务团队的核心价值，正是通过实时、高质量的信息流动，全力支持各职能业务主管在基础设施建设中挑选出最优项目、用好每一笔资本。正如现场给出的诚恳建言：“不要害怕变革。会计和财务人员可能会对变革感到焦虑，但请主动拥抱它，并把整个团队带在身边——你绝不可能在孤岛中完成转型。” --- ### [18:02](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=1082s) 走向 Agentic 企业的核心逻辑：“内置原生”与“外挂叠加”的思维分野 在总结双方合作经验时，安永（EY）指出，当前这套推动智能代理（Agentic）企业转型的治理与实施框架不仅适用于能源行业，更是在各行各业落地 AI 战略的通用范式。成功的 AI 转型必须明确锚定业务成果，在清晰的演进路线图上设立可衡量的里程碑，并彻底重构人机协作（Humans in the loop）的工作流形态。 更深层次的差异在于领导者的转型思维方式是“内置原生”（Built-in）还是“外挂叠加”（Bolted-on）。虽然“外挂叠加”式地在现有遗留流程上打补丁、自动化特定任务仍能产生一定效益，但真正能够实现指数级价值跃迁的企业，无一不是采取了“内置原生”的思维——将 AI、智能治理与信任机制从底层直接植入运营模型与业务价值流的核心，从而彻底颠覆传统的价值创造方式。 - **跨行业通用框架**：以成果为导向定义转型路线图，在核心业务流程中紧密嵌入人类监督与智能代理的协同。 - **内置原生 vs 外挂叠加**： - *外挂叠加（Bolted-on）*：仅在既有旧流程上做局部任务自动化，收益空间有限。 - *内置原生（Built-in）*：将 AI 与治理体系深度融入运营底座，彻底重塑业务流并解锁根本性的商业价值。 ## Notable quotes > [00:04](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=4s) 大概 500 位最要好的朋友 — Audi Rowe > [00:04](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=4s) 如何跳出零散、孤立的技术用例（use cases），转而聚焦于业务成果（outcomes）并重塑实现路径？ — Audi Rowe > [01:31](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=91s) 可能和我们的电网一样古老，甚至比电网还要老 — Cindy Lee > [03:13](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=193s) 拿着 AI 这把锤子到处找钉子 — Audi Rowe > [05:43](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=343s) 朝同一个方向划桨 — Cindy Lee > [05:43](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=343s) 有时需要将原有流程彻底颠倒过来 — Cindy Lee > [14:00](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=840s) 不增人头，重塑产能 — Cindy Lee > [14:00](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=840s) 如何花好手头的每一块钱，从而为股东争取赚回七块钱 — Cindy Lee > [15:51](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=951s) 如何从业务中硬生生砍掉 1 亿美元成本 — Cindy Lee > [15:51](https://www.youtube.com/watch?v=qiYw-5Ii1jE&t=951s) 不要害怕变革。会计和财务人员可能会对变革感到焦虑，但请主动拥抱它，并把整个团队带在身边——你绝不可能在孤岛中完成转型。 — Cindy Lee ## Mentioned Audi Rowe, Cindy Lee, 安永 (EY), 杜克能源 (Duke Energy), Ai4, FDI (Finance Data Infrastructure), SOX (萨班斯-奥克斯利法案), GL (总账系统), 布宜诺斯艾利斯, 印度 ## Video description Audi Rowe - Americas AI Experience, Strategy, and Transformation Consulting Leader, EY Cindy Lee - Senior Vice President of Finance Transformation, Duke Energy Organizations everywhere are investing in AI, yet many struggle to move beyond pilots and deliver measurable business value at scale. In this fireside chat, EY's Audi Rowe and Duke Energy's Cindy Lee will explore how Duke is transforming finance into a catalyst for enterprise-wide AI adoption by embedding AI, governance and trust into the core of its operating model. Through Duke's real-world transformation journey, attendees will learn how to move from isolated AI initiatives to an agentic enterprise by reimagining value streams, building a trusted AI foundation and delivering outcomes that accelerate growth, improve decision-making and unlock lasting business value. [Source](https://www.youtube.com/watch?v=qiYw-5Ii1jE)

[0:04] Audi Rowe: 大概 500 位最要好的朋友 [0:04] Audi Rowe: 如何跳出零散、孤立的技术用例（use cases），转而聚焦于业务成果（outcomes）并重塑实现路径？ [1:31] Cindy Lee: 可能和我们的电网一样古老，甚至比电网还要老 [3:13] Audi Rowe: 拿着 AI 这把锤子到处找钉子 [5:43] Cindy Lee: 朝同一个方向划桨 [5:43] Cindy Lee: 有时需要将原有流程彻底颠倒过来 [14:00] Cindy Lee: 不增人头，重塑产能 [14:00] Cindy Lee: 如何花好手头的每一块钱，从而为股东争取赚回七块钱 [15:51] Cindy Lee: 如何从业务中硬生生砍掉 1 亿美元成本 [15:51] Cindy Lee: 不要害怕变革。会计和财务人员可能会对变革感到焦虑，但请主动拥抱它，并把整个团队带在身边——你绝不可能在孤岛中完成转型。

[17:40] Cindy Lee: lean into it [18:02] Audi Rowe: bolting on [18:02] Audi Rowe: built-in

[0:04] All right. Well, we were told that we  were preparing for a fireside chat with [0:09] 500 of our best friends, apparently. So,  thanks for coming. We appreciate it. We did [0:16] prepare uh a conversation to have today around  uh a very special journey that that Duke's been [0:23] on uh with Cindy's leadership. Uh so, we're  going to do just that. Um my name is Audi Row. [0:29] I'm a partner at EY and I focus on AI driving uh  experience transformation and strategy. And uh one [0:36] of the things that we've been reflecting on is  is really that question around value and how do [0:41] you capture value with AI? How do you start to  think about pivoting away from use cases and how [0:47] do you really focus on outcomes and reimagining  how to get there? And so instead of going through [0:53] a hundred sides, uh the probably the the best way  to do it is actually to bring it to life through a [0:59] through a a great example. So I'm very fortunate  to have uh Cindy Lee joining uh me on stage [1:06] today. Cindy has had an incredible career at Duke  Energy. You may know Duke. Uh Duke is one of the [1:12] the leading energy providers. They serve over 10  million customers today. Uh and Cindy's uh been [1:17] with Duke for 24 years. You've held multiple  roles. Uh you've been the chief accounting officer [1:23] and controller and you are currently leading  the finance transformation. That's right. And [1:27] so we're going to talk a little bit through your  experience. Thank you for joining today. Thank you [1:31] for having me. Yeah. So maybe we'll start with uh  a big number. So Duke has said that they're going [1:37] to invest $200 billion in really reimagining the  energy infrastructure in the next decade. And if [1:45] we think about the energy in infrastructure that's  been built over the past 80 years to accomplish [1:51] that in 10 years is is quite significant. Uh and  you're at the helm of that and uh finance is kind [1:57] of where where you're starting. So talk to me a  little bit about um how we start with with finance [2:03] on that type of ambitious journey. Well, when  you're you're reimagining the grid and putting all [2:08] kinds of new equipment in, you also have to think  about how we're going to finance that. And that's [2:13] what part of our journey and we need to be able  to do that with insight-led decisions for our for [2:20] our leaders to be able to pick the best projects.  200 B billion dollars seems like a lot of money, [2:26] but it is scarce. It's a still a scarce  resource across our six states. And so, um, [2:33] we want to make sure that we reimagine our finance  processes because they're probably just as old, if [2:39] not older than our grid is. [laughter] Very fair.  So that's where we're starting on the journey. [2:44] I like it. And when we've talked about it, we were  um we've been able to catch up recently and we [2:48] were just having lunch together. We talked a lot  about um that you didn't just start with shiny [2:53] technology. You're not starting with AI or you're  not starting with a hammer and looking for a nail, [2:57] so to speak. You really were very intentional  and I know that trust was a big part of [3:02] of of starting on that journey. So maybe for the  room share a little bit more about that journey [3:07] and how you thought about framing it especially  where trust was a was a big part of that. [3:13] Yeah. So uh we've had a long-standing relationship  with EY now specifically starting back in 2018 [3:20] with a tax managed service um partnership and we  um we basically uh managed service all of our tax [3:29] except for four people. So, it was a pretty big  lift and we've gotten not only the value from the [3:35] cost savings taken out, but we've also partnered  with EY to get a lot of tax tax optimization [3:42] because you can imagine we have a lot of property  tax. Um, and that team has delivered actual cash [3:47] savings in the hundreds of millions of dollars.  So, that's been major and they continue to do so. [3:52] So that built the trust because EY knew our  systems, they knew our company, they knew our [3:57] culture really because there's a lot of folks  right um in Charlotte with us. So then about [4:02] four years ago we started a a journey to look  for a controllership managed service. Um and we [4:10] again chose EY and one of the major reasons for  that was because the trust that we have with EY [4:16] you know our systems you know our people you know  our culture and we did a um managed service down [4:21] in Buenus Aries and we have folks in India but it  was really more than that. it had a transformation [4:27] construct in the contract and so we've been on a  journey since then to really standardize practices [4:34] um uh make hundreds of improvements over the  couple past couple of years and now we're [4:41] pulling in uh that FDI layer uh there's a tool  called FDI um it's a data lake type layer that [4:49] allows us to bring agentic in so we can bring  our wellcontrolled uh general ledger and other [4:55] financial data into the data lake and um begin  to use agentic on it faster than we could ever do [5:01] by ourselves at Duke Energy. So and and in in  that journey what you what when we've had these [5:07] when we were kind of going through it what what  what struck me was the mindset you took in in [5:11] approaching from a controllership perspective. So  you really focused on outcomes uh versus really [5:17] automating what you do today. And in and in in  that way, you really adopted more of a built-in [5:23] mindset with AI versus kind of that bolt-on  mindset. And you kept humans at the center. I [5:28] know as a leader, your people and your culture,  you just noted that is a huge aspect of of how [5:33] you lead, but it was being really intentional  about how do you reimagine those processes [5:39] and how do you think about the role of the human  working with where the technology is at today. [5:43] Is that fair? Yeah, that that is very fair. You  can't just as as folks have ch even just going [5:49] over to the cloud for a GL system. You can't just  take your old processes and put them in. We you [5:54] have to really reimagine reimagine them. And we're  also standardizing them at the same time because [5:59] we have six different utilities. We brought them  all onto the same GL, but they they don't always [6:05] have the same um processes. And so it's really  important to get that base layer uh improved and [6:12] all rowing in the same direction. And um before  you start adding on the data intelligence layers [6:19] um reimagining those processes, you know,  turning them upside down sometimes. Um and [6:26] finance is a good place to start because it really  does we really do have well-controlled governed [6:30] data. It's just the the processes around it that  then we can start to automate and and utilize [6:37] agentic AI and agents more often. Yeah. And  when we were talking through it, this is I [6:41] think the only side. So I I think we have one more  slide. I think it says thank you. So I'm already [6:46] I'm already robbing you of the next slide, but  this is this is the slide that we wanted to kind [6:51] of give the mindset that you use or the framework  that you use with the team um in kind of working [6:57] through this journey that you've gone on. So I  think if you start at the bottom, you talked a [7:01] lot about how do you think about getting your data  right? How do you think about the systems and the [7:06] landscape that exists and thinking about where  from an AI native platform you're building that [7:11] that that new technology and you're bringing that  into your existing landscape. Uh and then moving [7:17] up you thought about h what is the intelligence  you referenced FDI the tool that we have and [7:22] how are you starting to build and compound on that  knowledge. You're we're going to talk a little bit [7:26] about trust. I know we have a few more questions  here. So we I wanted to put up this framework so [7:30] everybody could kind of see this was the structure  I think as a leader that you worked through [7:34] different elements of these layers as you've gone  on your agentic journey. Is that has that is that [7:40] a fair representation of That's right. Yeah.  And you have to start at the foundation and and [7:44] build the trust as you go along. But I see that  responsible AI and that's important to finance [7:50] folks as you know. We're very distressful at  times. We want things well controlled. Well, well, [7:55] let's actually jump there because you work in a  very regulated industry, right? And there's a lot [8:00] of discussion right now around how do you build  in a responsible way? How do you have confidence [8:05] and if you are reimagining processes, how do you  have confidence that agents are actually executing [8:10] on what's expected? So when we think about trust  and we think about the human role, again back to [8:15] your leadership principles as well, how do you  think about trust and and humans working together [8:21] from a from a from a responsible perspective or  especially within the industry you work in? Well, [8:26] and certainly the trust is built with EY already,  right? We we know you know our systems and [8:34] ensuring that the controls are there  in place. we have governance around um [8:39] tightness that we're not going to have any cyber  issues um and making sure that we then pull the [8:45] humans into the loop with all of the processes  and I've found that in this journey bringing the [8:51] the humans I should say the team along with us  um is the probably the most important part of [8:57] building that trust because it doesn't happen on  day one. It it happens over time as you go through [9:02] this journey and we're on a journey. We talked  about that at lunch that this isn't a oneanddone [9:07] kind of thing. You're gonna evolve as you go over  time. Um so we've built that um trust with making [9:13] sure that we get our FDI socks certified with our  um with our auditors and our internal audit team. [9:20] Um we've brought humans in the loop along the  way on all of the little um pieces of value that [9:26] we're adding um as we go through the process. um  making sure that the people understand the change [9:34] and making sure that they feel like they're  involved. I think that's really important. Um and [9:39] having a seat at the table and being able to say,  "Hey, we can we do it this way?" But I think the [9:44] other thing that was really helping build trust  is when we showed um one of the examples I'll give [9:49] you, I don't think we talked about this at lunch,  but we were um the team was putting together a [9:53] manual journal entry in our FDI tool, which  allows these agents to build it. And we had a um [9:59] it had probably 13 different pieces of support um  data support data files that had to be brought in. [10:06] And an Excel tab that had like 20 an Excel  file that had like 20 tabs with all this data [10:12] um used to take about 5 hours to pull together.  And the agent went through and pulled together [10:19] the journal entry, showed all of the formulas that  somebody would have been looking for in Excel to [10:24] make sure all the tick marks were there. It showed  how it got all of the information that it needed [10:30] and that built the trust because the uh  manager, the accounting manager could follow, [10:36] how it was built just like they may have a little  different than they may have, but it was it was a [10:41] lot quicker and that was really building trust  and getting people excited in the process. Hey, [10:46] we can cut a journal entry from four hours down  to a half an hour. That makes meaningful change. [10:52] It does and I think that is a good example of  where that human is elevating to a different role. [10:56] Right. And you talked we did talk about lunch  where you also just mentioned an aspect of uh [11:01] the culture and bringing folk along and you talked  about going to the different centers that you've [11:05] that you've um that you visited in different  countries to stand up. Is there any learnings you [11:10] had from a change? I know you're on this journey.  We also talked about you're on this journey now. [11:14] You're leading this from a finance transformation  perspective. So is there any learnings you've had [11:18] from that change management per aspect of how  the humans are now starting to see and build [11:23] confidence in the technology? You just referenced  the ability to to see and have that audit trail. [11:28] How are you seeing your teams work with this  technology and and build confidence in it? Yeah, [11:33] I think being involved in the process but seeing  seeing the controls, seeing the change is really [11:42] is allowing them to trust it. You can't just  tell people, but you and you have to bring people [11:47] along who are excited about it and who are they're  like itching to to get started on on their thing, [11:52] but you also have to bring along the people who  are skeptical so that they can see that they're [11:57] still in the loop and that their job is going to  change and evolve that they can think more about [12:02] the strategic things they need to think about or  maybe get some work life balance back in their [12:06] life. You know, rather than working 60 hours,  maybe 45 or something like that. Very fair. [12:12] And one of the things we were also talking about  was those milestones because we talked about it [12:15] as a journey, right? And we were talking about  what you've accomplished right now. Um, new, [12:21] you know, you referenced where it started from  2018 and as you're on this journey as you look [12:25] forward of the the very large investment that  you're going to make, some of those milestones [12:29] that you've already hit, but also thinking about  and knowing where you're going to measure the next [12:34] and having an having a perspective. How do you  think about your milestones as you look forward [12:38] and and kind of being able to attest? we were  talking about um month close and things like that. [12:44] Yeah, you definitely have to show the  CFO the value of the investment. And so, [12:48] yes, taking out a we've taken out a day from our  close, our aspiration is take out a couple more. [12:54] Um with a $200 billion capital plan, cash is  very important and our FFO to debt metrics [13:01] are very important. uh our cash flow comes in  late at Duke Energy and so we're really working [13:07] little there's many blueprints within that  cash flow to get that uh information to him [13:13] much quicker so he can sit on it before he has  to talk to the investor public about it. So to me [13:19] those quicker insights are a lot of  value. There could be some cost savings um [13:25] and there will be cost savings but I think the  other thing to really lean into is the insights [13:31] that you're going to get. you're going to get them  quicker, they're going to be stronger. Um, and um, [13:37] you're really going to be quicker to  to market on that. You're going to [13:40] be able to uh, help someone pick the the  best capital project for our customers. [13:46] Um, that's important because affordability is  really important, too. We can't just be wasting [13:51] any of that 200 billion. It's you have to make  sure that it's affordable for our customers. So, [13:56] over the term, it makes sense. And and one other  you just you just touched on one thing that just [14:00] made me remember when we were talking about this  $200 billion spend. You were also talking about [14:06] um the ability to once you hit these milestones as  you're thinking about the way that you're showing [14:12] your team the successes along the way. You've just  you've just referenced that. but also being able [14:18] to um we know ROI is very important for the team  but being able to think about growth differently [14:24] especially when you think about where that 200  billion will go and what that means in terms of [14:29] the way that where the human is elevating. So  you talked a little bit about how you're thinking [14:33] about that also from a personnel perspective.  So can you touch on that from from how you're [14:38] thinking looking forward when that investment  is being made? What does that mean to your to [14:43] the workforce that you're bringing in or  now maybe Not because you're you're using [14:48] new technologies. How are you thinking about  that? Well, and you hit on that. I mean, there's [14:52] there's some natural thought to like, oh, I've got  to add all of these accountants to help support [14:56] this $200 billion uh capital plan, but really  the goal is to to do that without new people and [15:03] to just repurpose the the time and the capacity  that our folks have. I mean, if you have some um [15:09] different talent along the edges, you can you  can do that, but I believe that there'll also [15:14] be some reskilling. So you'll move people where  they're needed most. Stop doing the things that [15:19] don't matter and move people towards where they're  needed most. They'll also just have more time to [15:24] think. The accounting and finance profession, we  do a lot of data gathering, report writing without [15:31] actually sitting back and think. We can have our  agents do that in much less time. And then we can [15:36] really think through how we're going to work with  our business partners to spend that next dollar [15:41] to earn earn that next seven dollars hopefully on  that dollar that we spend um for our shareholders. [15:48] That's awesome. All right. So So just kind of in  wrapping then because I think you just touched on [15:53] the mindset that you're using not just about  efficiency. have a lot of conversations with [15:57] with sea level executives right now around  that AI transformation journey. And too often [16:02] I feel like we're having a conversation around  I need to take $100 million out of my business [16:06] versus also leaning into what you just touched on  from an innovation and a growth perspective and [16:11] really thinking about where you're harnessing  these technologies to help you really grow um [16:17] without necessarily always thinking about growth  in the same way maybe historically. So as we [16:21] think about um where folks may be on that are in  the audience on their journey, what would be a [16:27] a a takeaway uh for anyone that may be starting  on their their AI transformation journey? Yeah, [16:33] I mean cost is something to consider? It certainly  is, but I would think about those insight driven [16:39] decisions that speed and agility that you can  get to something before it fails too fast and [16:44] costs too much. Um that I mean that always can be  the case. and then speed to decisions so that the [16:53] that the leaders can take action quicker. Um I  would also say that avoided cost that we talked [17:00] about a um a minute ago is extremely important  because again we don't want to add the next [17:06] dollar to our customers for for costs that  are not necessary. Right? So I think there's [17:12] more. The value is going to really be in the our  ability as a finance team to thoroughly support [17:18] all of our functional leaders in this buildout  and choosing the best project with the best dollar [17:26] to spend. And we can we can do that by getting  information to them much quicker than we do today. [17:31] I like it. you had you had one saying you  wanted to say to everybody when we were we [17:34] were talking through what would be the comment  that you'd make to everybody um in terms of uh [17:40] your encouragement to them? Yeah, I would say  don't be afraid of be afraid of change. Are I've [17:46] noticed our teams are coming along on that journey  and they they can we as accountants can be and [17:51] finance people can be nervous about change. Don't  be afraid of the change. Lean into it but bring [17:56] the team along with you. You can't do it in a  silo. I love it. Um, you know, when we when we've [18:02] we were on this journey, we're very fortunate  that we we could end be on this journey with Duke. [18:06] When we're when we're on this journey with clients  right now, I think some of the things we hope you [18:09] take away, I know this was a very quick fireside  chat with 20 people as this was pitched to us. Um, [18:15] the the the things that I think stand out to us,  this this framework is a way that we're working [18:21] with clients across industries. I know we've just  talked a lot about energy and we've talked about [18:25] that space, but this is a framework that we're  using across industries. is it's a it's a way [18:30] to think about what are the elements that need to  come together to be able to successfully navigate [18:35] AI transformation. How do you really think about  creating value with AI by focusing on those [18:41] outcomes by being clear on those milestones that  you're defining along that road map and really [18:46] reimagining how you're how you're doing work today  with humans um and agents working together, humans [18:52] in the loop. And so um it was a framework that we  wanted to to at least tease with everyone. It's [18:57] also a mindset. I mean you you've embodied it. Uh  you're you're an exceptional showcase of a leader [19:01] that's taken that mindset of building in versus  bolting on. And I do I do see a lot there's still [19:08] a lot of opportunity for automation. So there will  be value realized when you when you think about [19:13] looking at what's done today and taking a more of  a bolt-on. You're automating tasks that are done. [19:18] But I but I do think if I look across the  trends that we're seeing with with clients [19:23] that those that are adopting that built-in  mindset that are really uh embracing doing [19:28] very different things are those that are starting  to realize value really fundamentally differently [19:33] and I think you would you would you would you're  a great testament and there's lots more to come. [19:38] There is. Yeah. Awesome. Well, thank you all very  much for spending time. I know this is just a a [19:42] small part of your afternoon um but we appreciate  spending time. We do have a um an area in the [19:48] exhibit hall uh for uh for EY. It's our we we  showcase this method and we can kind of go deeper. [19:54] So if you all have questions, I think we're we're  heading there next. Yes. Thank you all. Thank you.