Thrive Embeds Engineers in Buyouts to Drive 40% AI Efficiency Gains

Thrive在并购企业派驻工程师，推动AI效率提升40% 永久所有权结构与派驻工程团队带来了可衡量的AI回报，在第三方软件试点折戟之处取得成效，原因在于股权共享使员工利益与自动化保持一致，而非导向裁员。 Anuj Mehndiratta 与 Jim Siders 详细介绍了 Thrive Holdings 的永久资本与所有者-运营者模式，该模式将前置软件团队直接派驻到所收购的运营企业中，而非追求传统的短期私募股权转手套现。他们结合具体案例——例如与 OpenAI 合作将报税准备时间缩短 40%，以及在 Shield 旗下的 19 家运营公司中实现 30% 至 70% 的 IT 工单自动化——阐释了其“做多人类”（Long Humans）的理念，即通过自动化消除繁重的行政杂务，将员工提升至高价值的咨询顾问角色。最后，他们建议企业领导者避免因开展数十个浅层 AI 试点而陷入停滞，而应将资源集中于解决具体、关乎生存的运营瓶颈，并实施严格的 KPI 追踪。 Thrive Holdings 借鉴伯克希尔·哈撒韦和丹纳赫的“所有者-运营者”模式，依托永久资本架构与派驻工程团队，深入被投企业内部重塑业务工作流，而非仅做外部软件交付。 (0:04) 针对贡献美国 GDP 约 46%–47% 却长期缺乏优质技术服务的中小微企业市场，Thrive 向 Shield Technology Partners 投资近 3 亿美元，致力于赋能并倍增现有员工效能，打破传统私募 3 至 6 年的快进快出周期。 (4:33) 践行“做多人类”（Long Humans）理念，Thrive 与 OpenAI 合作构建税务 AI，将报税全流程耗时缩减 40%，把专业人员从繁杂的案头工作中解放，转型为高价值战略顾问。 (9:21) 在 Shield 旗下的 19 家运营企业中，自主 AI 系统已能承接并解决 30% 至 70% 的 IT 运维工单，使工程师团队得以全面转型拓展企业级 AI 咨询服务。 (12:15) 针对企业 AI 落地，领导者应避免铺开上百个缺乏量化指标的碎片化试点，而应集中资源攻克单一关键业务痛点，以严格的硬性 KPI（如工单分流率和 NPS）衡量实际回报。 (18:08)

Thrive Capital and Shield Technology Partners argue that capturing the transformative value of artificial intelligence requires an owner-operator model backed by permanent capital, where embedded software engineers redesign core workflows alongside practitioners rather than deploying superficial point solutions. Permanent Capital and the Owner-Operator Structure (4:33) Drawing inspiration from industrial holding companies like Berkshire Hathaway and Danaher, Thrive Holdings operates as a permanent capital vehicle with indefinite holding periods rather than relying on traditional private equity roll-ups that flip assets every three to six years. Following Thrive's early investment in OpenAI—made when OpenAI generated $50 million in API revenue at a $29 billion valuation—Thrive committed nearly $300 million to Shield Technology Partners to acquire and transform businesses in mid-market sectors such as IT services and accounting. Workflow Automation and the 'Long Humans' Thesis (9:21) Rather than deploying superficial chat interfaces, Thrive and Shield focus on deep process re-engineering where software acts like an 'Iron Man suit' to augment rather than replace staff. In a collaboration with OpenAI on Thrive's accounting platform, mapping the full five-stage tax return lifecycle reduced end-to-end filing time by 40%, reallocating administrative hours into strategic advisory work. Similarly, across Shield's fleet of 19 operating companies, AI systems autonomously resolve between 30% and 70% of inbound help desk tickets, allowing IT technicians to focus on complex technical challenges and emerging enterprise AI advisory services. Change Management Through Aligned Incentives (15:04) External software vendors and consultancies often encounter internal resistance because automated tools are viewed as direct threats to employee jobs. Jim Siders notes that successful transformation must be conducted with, through, and within an organization by identifying internal champions and using new tools to actively retrain the existing workforce. By uniting business operators and software engineers under a single corporate umbrella with shared equity incentives, efficiency gains are channeled into capacity creation and value-chain expansion rather than workforce downsizing. Avoiding Tool Sprawl Through Targeted Execution (18:08) Anuj Mehndiratta and Jim Siders caution leaders against succumbing to the 'hammer looking for nails' trap by launching dozens of uncoordinated, demo-driven AI experiments. Siders compares enterprise software adoption to hiring a landscaper: customers evaluate the final condition of the yard, not the brand of rakes or tractors used. To overcome the cold start problem and achieve tangible returns, Mehndiratta advises organizations to restrict focus to a single core operational workflow, commit dedicated resources, and rigorously track baseline business metrics such as ticket deflection, Net Promoter Score, and capacity growth. The discussion emphasizes that enterprise AI value stems from deep operational integration rather than rapid technological rollouts, though it leaves open how easily the capital-intensive owner-operator model can be replicated outside well-funded holding structures.

Thrive Capital的Anuj Mehndiratta与Shield Technology Partners的Jim Siders提出，企业AI转型的关键在于超越表层软件采购，采用长期的“所有者兼运营者”控股架构；通过将自研工程团队深度嵌入被控业务以重塑工作流，在赋能核心人力而非裁员的前提下捕获实际转型价值。 永久资本架构重构企业AI落地机制 (0:04) Anuj Mehndiratta追溯Thrive Holdings的模式起源于对OpenAI的早期投资。在OpenAI年API营收仅5000万美元、估值290亿美元且仍受微软深度控制与非营利架构约束时，Thrive即大额注资，并预判AI将带来类似早期电商崛起时的系统性颠覆——单纯架设网站无法成为电商，企业必须彻底重构第三方物流（3PL）和搜索引擎优化（SEO）等底层流程。为此，Thrive借鉴伯克希尔·哈撒韦（Berkshire Hathaway）和丹纳赫（Danaher）的工业控股公司模式，建立永久资本载体，采取长期控制型所有权架构，以规避传统风险投资和私募股权基金三至六年的退出周期对企业价值的破坏。 Jim Siders在Palantir任职13年后转而执掌获Thrive注资近3亿美元的Shield Technology Partners。他指出，占美国GDP约46%至47%的中小型与中端市场企业长期被追求大型订单的传统SaaS厂商忽视。Thrive Holdings模式为旗下19家IT服务运营公司配备了专属的软件工程团队，前沿技术团队无需寻找产品市场契合点即可在自有业务场景中直接构建工具，通过充当人才倍增器使老牌企业转变为顶尖平台，而非依靠削减人员压缩成本。 “做多人性”与高价值咨询转型实践 (9:21) 在具体落地场景中，Anuj强调“业务本身即是产品”，坚决反对脱离实际运营的炫技演示。在Thrive与OpenAI联合为其会计平台研发税务AI（Tax AI）的过程中，工程团队将税表填报拆解为文档收集、分类摄入、自动处理与提取、财务分析洞察以及最终申报与咨询五大环节。技术人员与一线会计师深度协同，精准界定AI代理与人工介入的边界，最终将完成单个税表的端到端总耗时削减了40%，并直接将后台行政工时转化为前台客户顾问服务。Anuj援引Thrive的“做多人性”（Long Humans）理念指出，历史上的数字化往往只是将纸笔记录搬上屏幕，而现代AI代理应承担机械性流程，释放人类在战略、信任和人际关系上的核心判断力。 这一逻辑在IT服务中同样得到验证。Jim透露，Shield旗下19家运营企业已利用AI系统实现客户服务台30%至70%常规工单的自动化解决。企业并未借此裁撤IT人员，而是将释放出的人力全面转向复杂技术攻坚及新兴的AI战略咨询。Jim表示，中端市场企业对智能化升级的咨询需求呈现爆发式增长，高阶服务已成为重要的增量业务支撑。 内部利益一体化化解组织转型阻力 (15:04) 针对技术引入可能导致的组织摩擦与抵触情绪，Jim结合其在Palantir推动系统落地的经验强调，AI转型绝不能作为外部技术手段机械强加给组织，而必须由内而外、与员工共同推进。Anuj对比指出，外部技术顾问或软件供应商因直接威胁员工岗位，往往成为企业内部“最被痛恨的第三方”；而Thrive将业务运营者、一线实践人员与软件工程师统一置于同一控股实体下，并建立全员共享股权机制，使技术红利直接转化为内部员工赋能而非裁员风险。 这种激励机制的完全对齐在根本上改变了员工的心理预期。各平台企业不再将自动化视为生存威胁，而是转入主动出击模式，积极探索如何向价值链上游延伸、开辟全新客户服务与利润空间。组织通过识别各部门的早期拥抱者带动变革，并利用AI工具开展跨岗位再培训，从而在企业内部培育出积极推进“创造性颠覆”的内生文化。 规避工具导向陷阱与聚焦核心指标 (18:08) 面对企业在庞大AI投入面前因缺乏方向而陷入的冷启动瘫痪，Anuj援引Y Combinator的原则敦促管理层“在考虑规模化之前先做不可规模化的事”。他指出，即便Thrive拥有顶尖工程师并紧密合作OpenAI，也严格将AI试点限制在极少数核心场景，避免陷入铺设上百个散乱项目的陷阱。领导者应挑选单一关键业务职能集中配置资源，并设定严格的量化指标——例如生产力增幅、客户净推荐值（NPS）或实现30%至40%的常规工单分流，以确凿的业务成果打破市场对AI万亿投资回报的普遍怀疑。 Jim则以园艺业务作比，警告管理者避免陷入“拿着锤子找钉子”的工具导向思维。客户雇佣园艺工是为了拥有更美观的庭院，而非关注铁锹或拖拉机的具体型号。他认为，“前沿部署工程”的精髓在于对业务问题领域的深刻体察，企业领导者最宝贵的资产是对自身痛点与价值定义的直觉洞察，唯有将这种业务直觉与目标明确、指标严苛的技术执行相结合，才能真正赢得转型。 Thrive Holdings与Shield展示了通过永久资本整合与内部前沿工程部署推动AI转型的清晰框架，证明了中端市场企业在不裁员前提下实现降本增效的可行性。然而，该模式高度依赖充裕的长期资本掌控力、稀缺的顶级工程团队供给以及与前沿模型厂商的独家协同，缺乏此类控股与人才资源的广大独立企业如何在外部采购模式下达成相似的组织协同与利益对齐，仍是尚未解决的核心挑战。

## Background 该内容由 Ai4 于 2026 年 8 月 21 日发布在 YouTube 平台上。视频记录了由《华盛顿邮报》旗下 Washington Post Intelligence 副主编 Yun-Hee Kim 主持的深度圆桌对话，Thrive Capital 合伙人 Anuj Mehndiratta 与 Shield Technology Partners 首席执行官 Jim Siders 共同探讨企业如何通过“所有者-运营者”模式与长期主义视角，由内而外推动企业级 AI 的落地并弥合技术与实际业务成效之间的鸿沟。 ## TL;DR 对话深入剖析了 Thrive Holdings 借鉴伯克希尔·哈撒韦与丹纳赫模式建立的控股型“所有者-运营者”架构，通过永久资本和全员利益对齐，打破传统 VC 与 PE 的周期局限。团队将顶尖的前线部署工程师直接扎根于 IT 服务及会计等实体业务一线，与员工共创垂直应用，实现了报税耗时缩短 40% 以及服务台工单 30% 至 70% 的自动化拦截。他们旗帜鲜明地倡导“做多人类”与“钢铁侠战衣”理念，主张利用 AI 接管非判断性重复工作，从而释放专业人才向高附加值咨询与战略服务跃升。最后，嘉宾呼吁企业领导者摆脱单纯以技术或工具为导向的冷启动陷阱，凭借对业务原生问题空间的直觉敏锐度，聚焦单一核心职能进行穷尽式 ROI 度量。 ## Atmosphere & style 整体对话基调务实沉稳、充满战略洞察与实战厚重感，完全脱离了浮夸的技术 Demo 秀和空泛炒作。Anuj 以资本操盘手的高维视角阐述顶层组织哲学，而深谙一线业务的 Jim 则以生动接地气的比喻（如园艺工具与钢铁侠战衣）带来强烈的实操体感。两人的交流默契互补，从宏观资本范式层层抽丝剥茧至微观的组织变革阵痛，令整场对谈兼具思想深度与雷厉风行的执行迫切感。 ## Key takeaways - [00:04](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=4s) Thrive Capital 早期在 OpenAI API 收入仅 5000 万美元时以 290 亿美元估值重注重购，源于敏锐预见全球正迈向“智能极度充沛”的技术拐点。 - [00:04](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=4s) 如同历次技术范式转移，仅靠采购软件无法实现组织转型，企业必须重构底层运营与工作流才能承接 AI 浪潮。 - [00:04](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=4s) 借鉴伯克希尔与丹纳赫的“所有者-运营者”模式，Thrive 通过全员持股让技术专家与一线业务团队利益完全绑定，协同推动端到端变革。 - [04:34](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=274s) 占美国 GDP 约 46% 至 47% 的中小企业长期被传统企业软件与 VC 忽视，而永久资本载体能够摆脱传统 PE 短期套现裁员的周期局限。 - [04:34](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=274s) 工程师以“前线部署”方式深扎业务一线，既免去了寻找 PMF 与客户获取的焦虑，也为一线业务提供了随调随用的顶尖技术战力。 - [09:22](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=562s) 坚持“业务即产品”理念，拒绝脱离实际的演示 Demo，以重塑完整业务闭环为目标（如 Tax AI 助力报税耗时缩减 40%）。 - [09:22](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=562s) 践行“做多人类”（Long Humans）理念，AI 负责自动化处理非判断性底层杂务，释放人类聚焦于决策判断、战略咨询与客户信任。 - [12:16](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=736s) 推行“钢铁侠战衣”赋能模式，Shield 旗下 IT 服务台实现 30% 至 70% 工单全自动解决，推动传统技术支持人员向高附加值咨询顾问转型。 - [15:05](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=905s) 变革管理不能依赖外部供应商强加，必须在“同一屋檐下”由内而外推进，消除员工对被技术替代的抵触情绪。 - [17:13](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=1033s) 利益高度一致孕育出“创造性颠覆”文化，员工主动拥抱革新，共同探索企业全新的业务线、收入池与利润池。 - [18:08](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=1088s) 破解企业 AI 冷启动难题的关键在于：切忌全面铺开，先做不可规模化的攻坚，锁定单一业务职能并对其所有核心指标进行穷尽式度量。 - [20:34](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=1234s) 企业领导者应抵制“以工具为先”的技术自嗨，充分发挥对生存级业务问题空间的直觉与嗅觉，以此牵引技术落地产生实效。 ## Detailed walkthrough ### [00:04](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=4s) Thrive Holdings 的创立初衷与所有者-运营者模式 论坛开场探讨了企业如何“由内而外”（inside out）推动 AI 落地应用。Thrive Capital 合伙人 Anuj Mehndiratta 率先回顾了 Thrive Holdings 的诞生渊源。Thrive Capital 作为 OpenAI 最早的外部主要股东之一，在 OpenAI 的 API 年收入仅有 5000 万美元时，便以 290 亿美元的估值领投。当时外界普遍认为他们“彻底疯了”（totally lost our minds），因为 OpenAI 当时仍是非营利架构，估值倍数极高且微软掌握着强大的控制权。然而，Thrive 敏锐预见到技术拐点已经到来，全球正在进入“智能真正充沛”（intelligence truly abundant）的指数增长曲线。 回顾互联网和云计算等历次重大技术范式转移，企业往往能看到浪潮，但真正置身浪潮之上并完成自我改造却极其艰难。以电子商务的兴起为例，许多传统公司以为“转型电商”就是搭建一个网页，结果大量企业依然倒闭，因为它们忽视了背后的 SEO（搜索引擎优化）、3PL（第三方物流）以及围绕新技术必须对运营模式进行的根本重构。Thrive 意识到，面对极具颠覆性的 AI 浪潮，传统企业需要的不仅是一款新软件，而是一个能与它们并肩作战、从底层改变业务运作方式的深度合作伙伴。 为了从根本上推动组织行为变革，Thrive Holdings 确立了以“所有者-运营者”（Owner-Operator）为核心的控股公司模式，这一模式深受伯克希尔·哈撒韦（Berkshire Hathaway）和丹纳赫（Danaher）的启发。Thrive 认为，唯有通过股权所有权实现“利益深度一致”（aligned incentives），让技术专家、AI 领袖与业务运营团队持有共同的股权，大家才能真正合力探索可能性的边界。Thrive Holdings 作为控股方与企业管理者紧密协作，将自研技术团队直接派驻到 IT 服务、会计服务及建筑环境（built environment）等实体业务一线，从工作流层面去推动企业整体进化。 ```mermaid flowchart TD accTitle: Thrive Holdings 所有者-运营者落地模型 accDescription: 展示 Thrive Holdings 如何通过永久资本和技术前线部署改造传统业务板块 subgraph CapitalModel["资本与组织架构"] A["永久资本载体<br/>(Permanent Capital)"] --> B["控股型运作<br/>(Owner-Operator)"] B --> C["全员利益绑定<br/>(Shared Equity)"] end subgraph TechDeployment["技术部署与赋能"] C --> D["前线部署工程师团队<br/>(Forward-Deployed)"] D --> E["深入真实业务工作流<br/>(IT / 会计 / 建筑环境)"] end subgraph BusinessTransformation["业务价值重构"] E --> F["非判断性流程自动化<br/>(AI Agents)"] F --> G["专业人员能力倍增<br/>(Iron Man Suit 模式)"] G --> H["向高附加值咨询转型<br/>(Long Humans 理念)"] end ``` - **OpenAI 早期投资战绩**：在 OpenAI API 收入仅 5000 万美元、估值 290 亿美元时重注入局。 - **借鉴标杆企业**：以 Berkshire Hathaway 和 Danaher 的长期控股运营范式为蓝本。 - **模式核心逻辑**：拒绝单一技术采购思维，依靠全员持股的利益一致性推动业务流程端到端重塑。 --- ### [04:34](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=274s) 永久资本与前线部署：Shield 与 Thrive Holdings 的独特合作 Shield Technology Partners 首席执行官 Jim Siders（此前在 Palantir 工作了 13 年）分享了 Shield 与 Thrive Holdings 建立深度合作的原因。Thrive 近期向 Shield 投资了近 3 亿美元。Jim 指出，这种合作模式在当今市场上是独一无二的“瓶中闪电”（lightning in a bottle），打破了传统软件初创公司与传统私募股权（PE）的局限。 普通的风险投资（VC）模式要求软件初创公司追求高增长与高毛利的大型企业（Enterprise）合同，从而忽略了占美国 GDP 约 46% 至 47% 的中小企业（SMB）与中型企业市场。而传统的私募股权模式虽然擅长以低价收购资产，但往往受限于 3 至 6 年的基金退出周期，其打法大多是套用统一模板压缩成本、裁员并整合转售，这种“短期协同效应”在很大程度上限制了长期转型的野心，甚至可能破坏企业内在价值。 与之不同，Thrive Holdings 采用的是“永久资本载体”（Permanent Capital Vehicle），秉持无限期持有（indefinite hold）的理念。其目标不是通过压榨利润率来转卖企业，而是为优秀员工赋能，实现能力的“成倍放大”（force multiplier）。在技术研发层面，这种模式让一流的高速软件工程团队拥有了一批现成的、无需额外获客成本的“受众客户”。工程师以“前线部署”（forward-deployed）的方式直接扎根一线，不再为寻找产品市场契合度（PMF）而焦虑，而是专注于用 0.1 版本的技术去解决实际问题；对于一线业务人员而言，他们相当于拥有了一支专属的顶尖软件工程团队，随时能掌控技术产品的演进路线。 - **中小企业市场空白**：美国多达 46%-47% 的 GDP 来自中小企业，这部分市场对 AI 转型极度渴望，却因客单价问题被传统大型企业级软件商忽视。 - **资本性质差异**：抛弃 PE 基金 3 到 6 年的低价买入、降本裁员、转手套现逻辑，转为永久资本的长期无限期持有。 - **技术交付优势**：一线企业拥有随时响应自身需求的“世界级软件工程师专属团队”，技术团队则获得无需获客成本的真实业务试验场。 --- ### [09:22](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=562s) 业务即产品：与 OpenAI 共建 Tax AI 及“做多人类”理念 谈及具体的技术实践，Anuj 阐述了 Thrive Holdings 的核心指导原则——“业务即产品”（the business is the product）。在科技界，许多人热衷于打造炫目夺目的技术 Demo 来博取眼球，而 Thrive 关注的唯一核心是技术如何切实改变员工的日常工作方式以及如何带来业务成果。 在与 OpenAI 联合为会计业务开发 Tax AI 的过程中，团队没有只停留在单点工具开发，而是梳理了完成一份报税单的完整端到端流程：从前期的客户文件收集、分类、处理、数据提取，到深入的数据分析、提供专业见解、运行申报流程，再到最后向客户提供咨询反馈。通过由自身技术栈掌控整个流程，并由注册会计师（CPA）等一线从业者基于真实业务数据共同定义产品，他们打造出的 Tax AI 将报税所需时间缩短了 40%，同时使会计师与客户面对面沟通交流的时间大幅增加。 Anuj 强调，Thrive Holdings 专门发表过一篇题为《做多人类》（Long Humans）的文章。当外界普遍认为 AI 将取代所有人类工作时，Thrive 却持有截然相反的见解。过去的技术往往只是把纸笔时代的手工劳动数字化，而今天 AI 和智能体（Agents）可以接管繁琐的、无须人类复杂判断的底层工作，从而把人类重新释放回面向客户的前沿阵地，去专注于人类独有的判断力、战略制定、信任建立和前瞻性工作。 - **流程时间锐减**：与 OpenAI 联合研发的 Tax AI 帮助会计业务将单笔税务申报耗时缩短了 40%。 - **做多人类（Long Humans）论点**：AI 接管机械化、重复性数据处理工作，人类员工则集中精力于需要判断力、情感信任和战略咨询的高阶工作。 --- ### [12:16](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=736s) 从 IT 服务到更广阔领域：为专业人员穿上“钢铁侠战衣” 针对 AI 对更多传统行业的重塑，Jim Siders 认为目前在会计、IT 服务或合规等领域的技术应用仅仅触及了皮毛。业务需求在历次技术变革中并不会消失，而是会沿着价值链和质量链不断向上攀升。例如，生成式 AI 的爆发不仅没有减少对 IT 专业人员的需求，反而因为技术复杂度的提升，使得中小企业对高质量、高阶 IT 支持的需求迎来了爆发式增长。 Jim 将这种赋能模式形象地比喻为给优秀专业人员穿上“钢铁侠战衣”（Iron Man suit）。技术的目标不是把员工自动化淘汰掉，而是帮他们剥离那些容易引起人为失误的杂务与琐事。以 Shield 旗下的 19 家业务子公司为例，除了一家主营 ERP 集成的企业仍在改造外，所有 Help Desk（服务台）业务已有 30% 至 70% 的进线工单（tickets）完全由 AI 自动化处理。 Shield 并未借此裁员，而是推动这批 IT 人员向更复杂的价值链高端跃升，填补市场长期以来未被满足的高阶需求。Jim 透露，当他们尝试向客户推出 AI 咨询顾问服务时，客户强烈的反响“简直让人惊掉下巴”（takes your head off）——中小企业对实现自身数字化转型的渴望与紧迫感空前强烈。通过自动化剥离基础工作，团队得以腾出精力和产能，去承接利润更丰厚、影响更深远的复杂企业服务。 - **Shield 业务规模与落地现状**：旗下管理 19 家企业，Help Desk 服务台中 30% 至 70% 的进线工单已由 AI 自动解决。 - **钢铁侠战衣模式**：自动化消除易出错的低效干扰项，将传统支持人员直接升级为 AI 咨询顾问。 - **市场需求反馈**：中小企业客户对获取 AI 转型指导服务表现出近乎“迫切”（desperation）的极高热情。 --- ### [15:05](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=905s) 组织变革管理：从“强加技术”到“一体化协同推进” 在讨论如何引导基层员工主动拥抱转型而非被动接受变革时，Jim 结合在 Palantir 积累的经验指出，技术变革绝不能是一场单纯的技术运动，而必须是一场“技术与实体运营深度融合的行动”。传统优秀组织内部蕴藏着大量潜在人才，如果管理者将技术视为强加于员工之上的外力，变革必然受阻；唯有“与他们一起、通过他们、由内而外”（with them, through them, and within them）协同推进，变革管理才能真正奏效。 在任何行业中，固步自封的人固然存在，但始终会有一批渴望尝试新事物的早期采用者（early adopters）。利用 AI 辅助员工再培训（retraining），能够迅速激活并重塑组织的人才梯队。 Anuj 对此高度认同，并指出这正是 Thrive Holdings 坚持将技术研发与实体运营置于“同一个屋檐下”（under one roof）的关键所在。如果由外部第三方软件供应商介入推动自动化，该供应商往往会被一线员工视为“抢走饭碗的公敌”；但若在控股体系内部一体化推进，由熟悉具体流程的一线人员与自己的技术团队共同开发 AI 工具，技术就变成了为员工创造产能、拓展新业务可能性的核心助力。 - **变革管理原则**：技术不能“施加于”业务之上，必须与一线人员共同构建（inseparable unified problem space）。 - **消除对立情绪**：内部一体化协同避免了第三方供应商带来的“砸饭碗”敌对心态，将自动化带来的富余工时顺畅转化为拓展新业务的能力空间。 ### [17:13](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=1033s) 创造性颠覆文化与利益绑定的组织解锁 Anuj Mehndiratta 深入阐述了 Thrive Holdings 内部所建立的核心企业文化。当组织开始思考如何推动员工向价值链更高端迈进、以及如何通过利益一致化来创造增量与业务增长时，整个对话的性质就发生了根本性的转变。选择在各个平台上与他们深度合作的企业，往往由具有远见、明确表态“我渴望成为下一轮演进的一部分，绝不想被时代淘汰”的领导者掌舵。这种心态迅速辐射到基层员工，激发了整个团队探索新可能的热情：员工们不再焦虑于被替代，而是兴奋地思考还能为客户交付哪些额外价值、如何解锁全新的收入池（Revenue Pools）与利润池（Profit Pools），以及未来应该开辟哪些新的业务线。 这种模式在 Thrive Holdings 孕育出了一种“创造性破坏”（Creative Destruction）与“创造性颠覆”（Creative Disruption）的鲜明文化。Anuj 指出，这种看似松散自由的文化特质之所以能够真正运转并发挥威力，正是因为他们把所有的能力、业务和激励机制整合在了同一个屋檐之下，实现了利益的高度对齐。这种深度的组织重塑与协同，成为了推动传统业务实现 AI 跃迁的“超级解锁器”（Big Unlock）。 ```mermaid flowchart TD accTitle: 组织利益对齐与价值跃迁循环 subgraph Align["统一组织与激励"] A["同一屋檐下的业务整合"] --> B["利益完全对齐"] end subgraph Culture["文化与组织演进"] B --> C["创造性颠覆文化"] C --> D["主动拥抱技术革新"] end subgraph Value["价值链向上跃迁"] D --> E["释放新收入与利润池"] E --> F["开辟全新业务与服务线"] end ``` --- ### [18:08](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=1088s) 破解 AI 冷启动：聚焦单一职能与严密衡量 ROI 主持人 Yun-Hee Kim 抛出收尾问题：在场的企业领导者当下最应该采取的一项行动是什么？Anuj 借用了 Y Combinator 早期的经典箴言——“在考虑规模化之前，先去做那些不可规模化的事”（Do the unscalable thing before you start to think about scale）。他指出，AI 的强大往往是一把双刃剑：正因为大家都觉得 AI 无所不能、可以在企业各个角落开花结果，反而导致了极度棘手的“冷启动问题”（Cold Start Problem）。 尽管 Thrive 拥有顶尖的 AI 团队、OpenAI 的深度支持以及极具前瞻性的领导层，但他们在公司内部推行的 AI 项目实际上“屈指可数”（A handful），绝不是盲目铺开上百个项目。Anuj 强烈建议所有企业领导者：**挑选一个业务职能，并对所有指标进行穷尽式衡量**。深入该职能、集中资源攻坚，并锚定一个清晰可量化的核心指标——无论是运营效率、生产力、客户净推荐值（NPS），还是全新的服务线。当前行业内普遍焦虑“数万亿美元投资投下去了，ROI 究竟在哪里”，其根源就在于缺乏聚焦与清晰的衡量框架。AI 的价值绝不在于屏幕上展示的光鲜 Demo，而在于真实业务结果。 - **拒绝全面铺开**：不要试图在所有服务线和业务能力上同时推进 AI，集中精力于少数关键点突破。 - **业务结果大于技术展示**：真正的成功指标是“30% 到 40% 的工单被 AI Agent 成功分流拦截（Deflected），客户问题以更快速度解决，同时人类员工得以腾出精力为更多客户提供更高价值服务”。 - **度量驱动成效**：衡量框架的缺失是导致当下企业 AI 落地受挫、ROI 模糊的罪魁祸首。 --- ### [20:34](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=1234s) 领导者的最大利器：对“原始问题空间”的敏锐嗅觉 Jim Siders 随后从自身视角对这一建议进行了提炼与补充。他警告领导者切忌陷入“手里拿着锤子，满世界找钉子”的误区。他用了一个生动的生活比喻：“没有人去雇佣一家园艺绿化公司，是为了看他们的耙子和铁锹用得多精妙；人们想要的，只是自家的院子变得更漂亮。”如果客户一上来就去纠结绿化公司用什么型号的拖拉机或铲子，显然是荒谬的。这种以工具为导向（Tool-forward / Solutions-forward）的思维模式在技术创新周期的早期很常见，但对业务推进具有毁灭性。 Jim 认为，企业领导者手中最大的底牌与天赋，是对“狂野真实的问题空间”（Wild Problem Space）有着近乎本能的深刻理解，而这正是所谓的“前向部署工程师”（Forward Deployed Engineers）方法论最核心的追求。每位领导者心中都有必须直面和对抗的生存级痛点（Existential Problem）。领导者能够凭直觉感知到价值是否真正产生，因此应该将这种对问题空间的直觉作为领导力工具，用来塑造 Anuj 所强调的高度专注。 - **抵制“工具导向”陷阱**：不要沉迷于最新的技术或工具模型本身，紧盯最终要解决的业务结果与客户诉求。 - **发挥问题直觉（Intuitive Taste）**：利用业务负责人对生存级业务痛点的透彻理解，引导技术资源聚焦于最能产生实际价值的核心场景。 ## Notable quotes > [00:04](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=4s) 企业面对极具颠覆性的 AI 浪潮，需要的不仅是一款新软件，而是一个能与它们并肩作战、从底层改变业务运作方式的深度合作伙伴。 — Anuj Mehndiratta > [00:04](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=4s) 唯有通过股权所有权实现利益深度一致，让技术专家、AI 领袖与业务运营团队持有共同的股权，大家才能真正合力探索可能性的边界。 — Anuj Mehndiratta > [04:34](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=274s) 这种合作模式在当今市场上是独一无二的“瓶中闪电”，打破了传统软件初创公司与传统私募股权的局限。 — Jim Siders > [09:22](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=562s) 业务即产品（the business is the product）。在科技界，许多人热衷于打造炫目夺目的技术 Demo，而我们关注的唯一核心是技术如何切实改变员工的日常工作方式以及带来业务成果。 — Anuj Mehndiratta > [09:22](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=562s) 今天 AI 和智能体可以接管繁琐的、无须人类复杂判断的底层工作，从而把人类重新释放回面向客户的前沿阵地，去专注于人类独有的判断力、战略制定和信任建立。 — Anuj Mehndiratta > [12:16](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=736s) 技术的目标不是把员工自动化淘汰掉，而是给优秀专业人员穿上“钢铁侠战衣”，帮他们剥离那些容易引起人为失误的杂务与琐事。 — Jim Siders > [15:05](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=905s) 技术变革绝不能是一场单纯的技术运动，而必须是一场技术与实体运营深度融合的行动，唯有“与他们一起、通过他们、由内而外”协同推进，变革管理才能真正奏效。 — Jim Siders > [18:08](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=1088s) 在考虑规模化之前，先去做那些不可规模化的事（Do the unscalable thing before you start to think about scale）。挑选一个业务职能，并对所有指标进行穷尽式衡量。 — Anuj Mehndiratta > [18:08](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=1088s) 当前行业内普遍焦虑数万亿美元投资投下去了 ROI 究竟在哪里，其根源就在于缺乏聚焦与清晰的衡量框架。AI 的价值绝不在于屏幕上的光鲜 Demo，而在于真实业务结果。 — Anuj Mehndiratta > [20:34](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=1234s) 没有人去雇佣一家园艺绿化公司是为了看他们的耙子和铁锹用得多精妙，人们想要的只是自家的院子变得更漂亮。 — Jim Siders > [20:34](https://www.youtube.com/watch?v=KKdfzCZ15dE&t=1234s) 企业领导者手中最大的底牌与天赋，是对狂野真实的问题空间有着近乎本能的深刻理解，应该将这种对问题空间的直觉作为领导力工具。 — Jim Siders ## Mentioned Anuj Mehndiratta, Jim Siders, Yun-Hee Kim, Thrive Capital, Thrive Holdings, Shield Technology Partners, OpenAI, Microsoft, Berkshire Hathaway, Danaher, Palantir, Y Combinator, Washington Post Intelligence, Tax AI ## Video description Anuj Mehndiratta - Partner, Thrive Capital Jim Siders - CEO, Shield Technology Partners Yun-Hee Kim - Deputy Editor, Washington Post Intelligence, The Washington Post How can businesses drive enterprise AI adoption from the inside out? Join Anuj Mehndiratta, part of the founding team at Thrive Holdings, and Jim Siders, CEO of Shield Technology Partners, to discuss how Holdings is adopting an owner-operator model and long-term approach to bridge the gap between AI capabilities and impact. [Source](https://www.youtube.com/watch?v=KKdfzCZ15dE)

[0:04] Anuj Mehndiratta: 企业面对极具颠覆性的 AI 浪潮，需要的不仅是一款新软件，而是一个能与它们并肩作战、从底层改变业务运作方式的深度合作伙伴。 [0:04] Anuj Mehndiratta: 唯有通过股权所有权实现利益深度一致，让技术专家、AI 领袖与业务运营团队持有共同的股权，大家才能真正合力探索可能性的边界。 [4:34] Jim Siders: 这种合作模式在当今市场上是独一无二的“瓶中闪电”，打破了传统软件初创公司与传统私募股权的局限。 [9:22] Anuj Mehndiratta: 业务即产品（the business is the product）。在科技界，许多人热衷于打造炫目夺目的技术 Demo，而我们关注的唯一核心是技术如何切实改变员工的日常工作方式以及带来业务成果。 [9:22] Anuj Mehndiratta: 今天 AI 和智能体可以接管繁琐的、无须人类复杂判断的底层工作，从而把人类重新释放回面向客户的前沿阵地，去专注于人类独有的判断力、战略制定和信任建立。 [12:16] Jim Siders: 技术的目标不是把员工自动化淘汰掉，而是给优秀专业人员穿上“钢铁侠战衣”，帮他们剥离那些容易引起人为失误的杂务与琐事。 [15:05] Jim Siders: 技术变革绝不能是一场单纯的技术运动，而必须是一场技术与实体运营深度融合的行动，唯有“与他们一起、通过他们、由内而外”协同推进，变革管理才能真正奏效。 [18:08] Anuj Mehndiratta: 在考虑规模化之前，先去做那些不可规模化的事（Do the unscalable thing before you start to think about scale）。挑选一个业务职能，并对所有指标进行穷尽式衡量。 [18:08] Anuj Mehndiratta: 当前行业内普遍焦虑数万亿美元投资投下去了 ROI 究竟在哪里，其根源就在于缺乏聚焦与清晰的衡量框架。AI 的价值绝不在于屏幕上的光鲜 Demo，而在于真实业务结果。 [20:34] Jim Siders: 没有人去雇佣一家园艺绿化公司是为了看他们的耙子和铁锹用得多精妙，人们想要的只是自家的院子变得更漂亮。 [20:34] Jim Siders: 企业领导者手中最大的底牌与天赋，是对狂野真实的问题空间有着近乎本能的深刻理解，应该将这种对问题空间的直觉作为领导力工具。

[4:33] Jim Siders: lightning in a bottle [4:33] Jim Siders: force multiply excellent people [4:33] Jim Siders: private zoo of world-class software engineers [9:21] Anuj Mehndiratta: the business is the product [9:21] Anuj Mehndiratta: Long Humans [12:15] Jim Siders: Iron Man suit [12:15] Jim Siders: you touch it and the response just takes your head off [15:04] Anuj Mehndiratta: the most hated third party in the world [18:08] Anuj Mehndiratta: do the unscalable thing before you start to think about scale [20:34] Jim Siders: hammer that's looking for nails

[0:04] Thank you everyone for joining this fireside  conversation. In this session, we'll talk about [0:08] how businesses can enable AI adoption from inside  out. Um so, Anuj Mondrata is part of the founding [0:16] team at Thrive Holdings, and Jim Sellers is CEO of  Shield Technology Partners. Welcome. And thank you [0:24] all for joining us this afternoon. So, let's  talk about the business model and AI change. [0:29] Thrive Holdings has kind of a unique business  model to deploy AI adoption within workplaces. [0:35] Can you first explain to those of us who might not  be so familiar with this model, what do you mean [0:40] by owner-operator model and why is it different?  Yeah. Well, thank you for being here. Um I think [0:49] that where it's maybe helpful to start is to share  a little bit about where Thrive Holdings actually [0:55] came from and the origin story. Uh like mentioned,  my name's Anuj. I'm a partner at Thrive, [1:02] and we felt very lucky at Thrive Capital to be  one of the largest outside shareholders in OpenAI. [1:08] And we actually invested in the company uh  when it was doing 50 million of API revenue. [1:14] We invested at $29 billion valuation. And people  looked at us like we totally lost our minds. [1:20] Um and the reason why it was it was a nonprofit at  the time. It was like I said, the the multiple was [1:27] a bit crazy. Microsoft very much controlled  the company given the current structure. [1:32] And people looked at us like, you know, AI's  been this big promise, like what's changed? [1:36] And I think what we saw was that the  technology had reached a point where [1:41] we were going to go on a curve to making  intelligence truly abundant across the world. [1:46] And I think a lot of our frameworks come from  having now studied previous technology paradigm [1:51] shifts. Things like the internet, things like  cloud. And what we learned was that businesses [1:58] often times can recognize a technology wave,  but actually putting themselves on that wave [2:03] is an entirely different challenge. And  what does that actually look like? Well, [2:08] I think it's often helpful to look at historical  analogies. So, if you study e-commerce, [2:12] what ended up happening? There were a bunch  of businesses that said, "Okay, I've got to [2:16] become an e-commerce company. That's the way to go  sell my products. I'm going to sell them online." [2:20] And what is selling online entail? It entails  something much broader than just putting up a [2:25] website. A lot of companies put up a website.  A lot of them died anyways because they didn't [2:30] think about SEO. They didn't think about 3PL. They  didn't think about all the changes that had to be [2:35] made to the actual way in which the business was  run and operated around that technology to kind of [2:41] go and be on this wave of the internet. And when  we reflected on the power of this technology and [2:47] the way things were going to change, we said, "A  lot of businesses are going to need a partner to [2:52] really understand it and really understand all the  ways the business needs to fundamentally change." [2:57] And so, that was really the genesis of Thrive  Holdings. It actually came from our investment [3:01] in OpenEye. It was recognizing this technology.  It was having a front seat at the table [3:04] and saying, "Man, this is going to  be amazing, but also it's going to be [3:07] challenging. It's going to be difficult.  It's going to be massively disruptive." [3:11] And so, we designed Thrive Holdings  to go and capture a lot of that value. [3:15] You know, you mentioned owner-operator. We said,  "Well, how do you actually go drive change? How do [3:19] you go drive behavior shift?" Well, you need to  have aligned incentives. Aligned incentives can [3:24] come through ownership. And what does that mean?  Well, if businesses, AI leaders, technologists, [3:31] people like Jim, people like myself all have  shared equity ownership in an outcome, well, [3:37] now we're all going to bring our ideas together  to actually go and advance what's possible. [3:42] And so, we designed the company as a holding  company, really inspired by Berkshire Hathaway, [3:46] Danaher, others. People who have defined  this methodology of going and owning [3:51] and operating companies and changing them from  within. What does that mean? We are control [3:56] owners in Shield. We work very closely with Jim,  who's someone that I've known for many years now, [4:01] and I feel very lucky to work with. We deploy  technology teams, who again have shared equity and [4:07] shared interests in actually working hand-in-hand  with our businesses in IT services, in accounting [4:12] services, now in the built environment across  many different platforms that we're building. [4:16] Understanding the workflows, understanding the  way things need to change, and really going and [4:21] furthering the way the business needs to change,  not just the technology should be adopted. [4:26] And so, that was really the genesis of the  company, and I think that's really the model [4:29] that we've tried to design. So, Jim, can you  talk about how that partnership work? Obviously, [4:34] it's different from an investor model, and Thrive  just to point out to the audience has invested [4:40] close to $300 million in Shield of late. What  does the partnership kind of look like inside? [4:49] Inside it's uh By the way, thank you for having me  here, and I'm I'm grateful for the opportunity. Um [4:56] it's unlike any existing model, in my opinion.  You know, I I spent a long time at Palantir, [5:03] 13 years before this at Palantir, and left there  when the company became kind of the center of the [5:09] world in a lot of ways. Certainly has showed  up in the previous talk and several others. [5:14] Um and the reason that I left to to come do  it is actually because of this partnership, [5:18] because I I I feel it's kind of lightning in a  bottle. I mean, I maybe easier to describe in [5:24] terms of what it's not or what it what it rhymes  with or what first principles we came from. So, um [5:30] you know, that's it's it's easy to form a software  startup, for example, to try to address some of [5:35] these issues, right? There's a lot of talks that  are happening here from really good companies that [5:39] are building software to go bring AI to various  different spaces, and that's and that's valid. [5:44] In the space that I'm most familiar  with, IT services and IT operations. [5:50] In the small and medium business space, um that's  not a lucrative space for most companies who are [5:56] seeking operational change or like trying to  address operational things with a software [6:01] product, right? Like enterprise deals are just  frankly more lucrative. Um so your venture backers [6:07] and you know the growth curve that your investors  expect, the time frames that they're working on, [6:11] they expect that you're going to be going after  these these more and more lucrative, more and [6:14] more high margin deals, which basically means that  there's like up to you know what, 46, 47% of GDP [6:19] in the United States running through small and  medium businesses and small enterprises that you [6:23] can't really tap, you can't touch. So that that's  an unaddressed opportunity. It is an unaddressed [6:28] need, certainly for AI and transformation, the  the disciplines that we're all interested in. [6:33] Similarly, um in private equity or other ways of  building holding companies, um there's some really [6:40] good recognition of fundamentals. They're really  good at finding and buying affordably and well, [6:45] good companies, running them in a credible  way. But largely they were trying to buy the [6:49] same company over and over again, run the same  playbook as cheaply as possible in all of them, [6:55] and sell all of them all together in three, five,  six years, something like that. Again, driven by [7:00] the return that the investors need in a specific  time frame, which limits your ambition, right? At [7:05] the very at best it limits the ambition. At worst  it may actually be value destructive in some ways. [7:09] It may be more about recognizing synergies  in a short time frame than about actually [7:14] accomplishing this bigger goal of again, trying  to tap this huge percentage of the economy that's [7:20] craving transformation and not able to get it cuz  they can't afford it cuz they don't have these [7:23] giant deal sizes. So the partnership with Thrive  Holdings is very different um in a few ways. So [7:30] um so Anoush mentioned that um it's a per-  it's a it's a holding company, it's not a fund. [7:36] It's a permanent capital vehicle, and that means  that we have an indefinite hold. The goal is not [7:41] to sell all of the companies after smashing  a bunch of margins out of them and firing a [7:45] bunch of people. In fact, the goal is exactly the  opposite. It's to It's to force multiply excellent [7:51] people in the companies that we buy and turn them  into world-class versions of themselves. That's [7:55] actually the ambition of the partnership. That's  unlike anything that's out there that I know of. [8:01] The other thing is the technology delivery  model. So, you do in fact need some of the first [8:06] principles of high-velocity world-class software  companies, right? These are the These companies [8:12] are better at innovation than anybody else  on Earth. They're better at developing talent [8:17] than anywhere else on Earth. Um and so, you  actually do need to bring those things to bear. [8:21] What if you could have as a high-velocity software  company a captive audience of customers that you [8:27] don't need to go acquire and where you can  go to the work in a forward-deployed fashion, [8:33] not not worrying about product-market fit, but in  fact worrying about whether you're addressing the [8:37] outcomes, whether you're creating durable value  with version 0.1 of the thing that you're doing. [8:43] And if you're an IT company or an end customer,  you have basically a private zoo of world-class [8:49] software engineers who are just there working  for you. You have by definition, you have your [8:54] thumb on the pulse of their product roadmap  at all times. It's a win-win for everybody. [9:00] And our bet is that the the the first principles  of this partnership are going to let us address [9:06] this untapped segment of the economy, frankly,  create a lot of value in every step of the chain [9:12] for everybody all the way through and it and it's  the foundation of that thesis is actually the [9:17] things that are unique about the partnership. So,  let's drill down on some real-life examples. So, [9:22] you mentioned Open AI, but Thrive and Open  AI co-developed Tax AI for accountants. [9:28] What was the big takeaway from that collaboration?  Yeah, it's a great question. I think that [9:35] we have a few principles that have kind of guided  even the way we've thought about technology here, [9:40] which are one, the business is the product.  What do I mean by that? I think oftentimes in [9:45] technology, you can actually say, "Oh, I want to  go build the sexiest demo that I can put up on [9:50] the screen and people will kind of ooh and ah and  say, 'Man, that's really cool.'" And I think that [9:55] oftentimes is the motivator behind a lot of the  work that goes into building great software. Now, [10:00] I'm not saying that we don't admire great  demos and great technology, but in our view, [10:05] what we've done is because we are focused  on how do we actually go and drive outcomes, [10:10] we are really focused on how is technology  changing the way the people work within our [10:15] companies. So, when we thought about tech tax  as part of our accounting business, we said, [10:20] "What is the actual full process end-to-end of  completing a tax return?" From document collection [10:25] to document classification to document  processing to data extraction to analysis to [10:30] providing insights to actually running the return  to then going and giving feedback to the customer, [10:36] and what are the parts of that process where you  need a human? What are the parts where you don't? [10:40] And where agents can actually go and do  some of the work on the people's behalf. [10:44] And what we learned was that by owning the entire  end-to-end process with our technology stack, by [10:51] actually defining the product with practitioners  like accountants, by using real-world data, [10:57] we were able to actually partner with OpenAI,  our own engineers, the accountants that exist at [11:01] our business to build a product that actually  reduced the amount of time that it takes to [11:06] complete a tax return by 40% and increased the  amount of time that our accountants are spending [11:13] with actual customers. Now, if you think  about that, 100%. And so, I think our view [11:20] is that a lot of people are kind of of taking this  lens that AI is going to replace all human work. [11:25] Our whole thesis, we actually published a piece  on this at Thrive Holdings is called Long Humans, [11:29] where people are saying, "Hmm, how is that  consistent with what you're talking about?" Well, [11:34] for a long time technology has promised to  actually go do work on our behalf. It's never [11:37] happened. It's actually just digitized work that  was previously done maybe manually or on pen and [11:42] paper and pencil. Now, you can actually put humans  back in the customer-facing part of the work, [11:48] let them focus on judgment, let them focus  on strategy, let them focus on trust and [11:52] relationships, and forward-thinking, proactive  work, and let AI and the agents go do the work [11:59] that really doesn't require the human judgment  that I think we all have inside of us. And so, [12:03] our view is we want to amplify that. We've  done that now in tax with our accountants. [12:08] We've done that now at Shield with our IT  service professionals in a pretty amazing way. [12:12] And so, I can let you maybe Jim share some of  those examples as well. What What are What are [12:16] the other areas that AI could disrupt beyond  kind of taxes and IT services? Well, I mean, [12:22] just in those spaces, we've only just scratched  the surface is the thing I would point out. I [12:26] mean, there's there's a lot of examples of work  that in in my thesis, and I believe in Anuj's, [12:34] that the work will persist throughout multiple  waves of transformation, but the it needs to climb [12:41] higher and higher up the value chain and up the  quality chain. You know, like the thing I'm most [12:44] familiar with is IT, because that's what I do all  day. You know, the the onset of generative AI in [12:54] our society broadly has not, in fact, reduced the  need for IT professionals. In fact, in the small [13:01] and medium business space that I'm working in the  most, in small and and mid-market enterprises, [13:05] in fact, it's increased the need for for for  high-quality and more complex support, right? So, [13:12] with the the IT professionals that we have,  or with tax professionals, or with regulatory [13:19] professionals who work on audit prep for technical  compliance, for example, or building compliance, [13:25] right, in the built environment, Um if you  could have each one of them be like force [13:30] multiplied essentially. Like if we had an Iron  Man suit for the best of these professionals, [13:35] rather than seeking to automate them away,  what we do is automate away the things that [13:39] are a distraction for them where they're a source  of human error and let them focus more of their [13:43] time on high-quality work that's potentially  more lucrative or certainly more value creating [13:48] in some way um for the customer. So, like in  our case, and just as a as a concrete example, [13:55] I mean most of our businesses we have we  have 19 businesses in the Shield Fleet now. [13:59] And uh most of our businesses uh save one that  does mostly ERP integrations, which we're working [14:06] on still. All of our help desk businesses um have  between 30 and 70% of the tickets inbound to them [14:13] being done by an AI now. But we're not using that  as an excuse to fire a bunch of IT professionals [14:20] and squeeze out margins. In fact, what we're  doing is pushing those IT professionals farther [14:24] up in value and complexity chain. So, they're  actually addressing unmet needs that there's no [14:29] current story for in the industry among all of our  customer base. And it turns out that customers are [14:34] incredibly grateful. Like we we thought that  there might be a little bit of demand for AI [14:39] advisory services and it's like you you touch it  and the response just takes your head off. The the [14:45] amount of passion behind it and like desperation  to transform their business and get get access to [14:50] this. And it's and it's by automating away this  basic work, getting increasingly good at doing it [14:55] quickly and with increasingly high quality um that  we get access to this harder and more lucrative [15:01] and more important problem space. You talked  about um the fact that humans could do more [15:05] lucrative work, but how do you bring employees  into that transition rather than have like change [15:11] come upon them? Well, there's in the businesses  that I'm in this is something that I did a lot of [15:18] in my previous world, you know, at Palantir both  internally and with our customers in the Fortune [15:23] 500 and these giant important enterprises  and institutions of the world. You know, [15:29] one of the things that we don't talk about in the  technical disciplines and I and by that I include [15:33] CPAs and and regulatory professionals is the  ability for AI to assist and and lever retraining. [15:42] You know, I I think there's there's a lot of  latent talent in good organizations. And if you [15:48] approach operating these companies and building  technology with them as a unified problem space [15:56] that they're inseparable. Like you can't  deal with you can't do things to them. [16:00] You have to do it with them and through them  and within them. Then you can't do something [16:05] that's purely a technical motion. It must be  a technical and operational physical motion. [16:10] That gives you access to a bunch of really  interesting stuff that there there's a lot [16:14] of academics on this like like how well can  people be retrained? Well, you know what look [16:18] like there are some people um who are set in a  way of doing things whatever the industry is. [16:24] But in every case there are going to be enough  that that are early adopters that are interested [16:28] in doing the thing that a change management motion  does in fact work and it's something I've spent a [16:32] lot of time on over the years throughout the last  few waves uh of of technological transformation. I [16:38] it's something I really believe in. Yeah, and you  should. Yeah, I the only thing I would add is that [16:41] this is why I think doing this all under one  roof really matters. And because I think that [16:47] for us technology here is creating capacity. Okay.  Well, in some ways you can imagine if this was a [16:53] third party coming in, what are they going to do?  That they all of a sudden that's going to be the [16:57] most hated third party in the world because that  third party is going and taking people's jobs. [17:02] But if you do this in a connected motion where  it's the same people that are doing the work to [17:08] actually go and really automate work, apply AI,  and then they're actually talking about okay, [17:13] how do we push people further up the value  chain? How do we have aligned interests in [17:17] going and creating more and growing? It actually  changes the nature of the conversation. So, the [17:22] businesses that are choosing to partner with us  across all of our platforms are the ones that are [17:26] really being led by people who are saying,  "I want to be part of this next evolution. I [17:30] don't want to be left behind." And so, they're the  ones that are excited about adopting technology. [17:34] They're the ones that are excited as an employee  base to go and figure out what is the other work [17:38] that I could be doing for my customers. How  do I unlock new revenue pools? How do I unlock [17:42] new profit pools? How do I go figure out what the  next service lines are that we should be creating? [17:47] And so, there is a culture of creative  destruction, creative disruption, and I think that [17:52] really is what we've been able to create at Thrive  Holdings. And so, it's a very loose feeling. It is [17:57] a cultural attribute, but it happens because  we have done all these things under one roof [18:02] with aligned incentives. And I think that's  really been the big unlock for us. Well said. [18:08] What is one action a leader in this room  should take now with AI? Final question. [18:15] Um I can start. I think that there was this  amazing line from uh YC back in the day, [18:22] which is that uh you do the unscalable  thing before you start to think about scale. [18:28] And I think with AI, everyone kind  of views it as like, "Oh my god, [18:32] I've got this like incredible tool that should be  able to do so much across my business." And it's [18:37] almost because it is so powerful that it makes  the cold start problem really, really hard. And [18:42] what I would say is that what we've learned is we  have very high-powered AI teams. We've got OpenAI. [18:49] We've got amazing leaders. All of them thinking  about how do we go and change our businesses? If [18:54] you asked us how many initiatives we have around  AI at our company, we would say it's a handful. [18:59] It's not a hundred. And so, what I would say to  everybody is pick one thing. Pick one function [19:06] and measure everything. And really go deep on that  function, apply resources to it, have a measurable [19:12] outcome around it, and if you're able to do that,  you're able to actually go and see what it takes [19:17] to really go and harness the value of AI because  it's not just about building a product or buying [19:22] some piece of software and adopting it. It's  about what is the metric that I care to move. [19:27] Is it one around efficiency? Is it one around  productivity? Is it one around customer NPS? [19:31] Is it one around a new service line? I I It can be  kind of whatever is best suited for your business. [19:37] But I do think focus and that clear execution  and that ability to go and measure results, [19:43] everyone's asking this question of like, "Where's  the ROI? We're investing trillions of dollars. [19:47] Where's it all going to go?" The reason that it's  not becoming clear to people is because we haven't [19:51] figured out the measurement frameworks, we haven't  been focused. I think that's where enterprises are [19:55] falling short and so those would kind of be the  things that I would really recommend. It's also [19:59] what we're doing ourselves. We're not going across  all of our service lines, all of our capabilities. [20:03] We're focusing on certain areas. We're really  diving into those things and then we're really [20:07] making sure that we're measuring outcomes and  results because that's what actually matters. [20:12] It's not about the fancy demo that you can show  on the screen. It's about, "Oh my goodness, [20:16] we've got 30 40% of our tickets being deflected  by AI agents. Our customers are getting those [20:21] problems solved faster. And by the way, our  humans are doing more work with more customers." [20:26] And that is a really really powerful thing.  And so that's where we're seeing the value [20:30] and I think that's because of the way  in which we focus our efforts. Jim. [20:34] Uh a similar answer but in Jim speak. I would  say there's a great danger for a leader in [20:39] being a hammer that's looking for nails. You know,  nobody goes to a landscaping company looking for [20:43] excellent rake and shovel execution, right? What  they're looking for is their yard to look better. [20:48] Um and and you know your yard, you know the  problem space. So you have an an intuition, [20:53] an intuitive taste about what value is  going to be when it when it's created. [20:58] So you go looking for the thing that's going  to accomplish the value at the price you want, [21:02] right? Like So you would never dream of like, "Let  me see what type of tractors and shovels you use. [21:08] That would be crazy, right? This wheel this  solutions forward tool forward way of thinking [21:12] is natural at this part of an innovation cycle,  but it's ultimately destructive to to progress. [21:19] The gift that every leader has in their  organization is what forward deployed engineers, [21:24] which by the way is often misused, I I  hate to say, but what what true true true [21:29] forward deployed methodology is seeking, which is  intimate familiarity with a wild problem space. [21:36] If you have an existential problem, and every one  of us does as a leader, you have a great gift, [21:41] right? Because you have you know a thing that you  can tilt against, and you will know viscerally [21:46] whether value is being created in that space. So,  use that that intuition about the problem space [21:52] as a leadership tool to create exactly the type  of productive focus that Anoush is talking about. [21:57] That would be my advice. Great. Well, I'm  afraid we're out of time. Thank you so much, [22:01] Jim and Anoush. Thank you, guys. Thank  you. Audience, great to Oh, thank you.