Bending Spoons Shuns Big Tech Veterans to Put 25-Year-Olds in Charge

Bending Spoons弃用科技大厂老将，重用25岁年轻人挂帅 首席执行官Luca Ferrari表示，经验丰富的企业管理者动力衰退且套路陈旧，这促使公司采用量化招聘指标，并委派25岁的年轻人掌管所收购的软件资产。 Luca Ferrari深入剖析了Bending Spoons独特的运营模式，该公司利用自有资产负债表出资收购Vimeo、Evernote和AOL等数字产品并永久持有，在统一的技术平台上进行重构。他详细阐述了公司基于数据的科学招聘方法，解释了为何名校学历和非结构化面试无法有效预测工作表现，而高成长斜率与极致的主人翁意识足以证明让25岁年轻人执掌重大并购资产的合理性。Ferrari还说明了Bending Spoons为何解散执行委员会、为何将举证责任完全施加于任何试图增加组织复杂性的人，并将激进坦诚确立为企业文化根基。节目最后，Ferrari分享了支撑公司声誉的个人道德准则——绝不说谎且绝不隐瞒真相，随后由Halligan对Ferrari的运营打法进行了总结。 Bending Spoons构建了一套工业级的人才吸纳体系，单年收到80万份申请但发放的录取通知不足300份，同时将令人惋惜的人才流失率控制在仅0.6%。 (8:01) 与传统的私募股权整合或去中心化控股公司不同，Bending Spoons利用自有资产负债表出资收购以永久持有，并将收购标的完全迁移至统一的技术与运营体系。 (0:44) 量化招聘回测表明，常春藤盟校学历与高绩效的相关性较弱且往往伴随着特权意识风险，而在青少年时期活跃但在二十出头时停止的政治活动则高度预示着能动性。 (13:45) 该公司通常任命二十五岁上下、具备高成长斜率的运营人才执掌新收购的业务，比起行业资历，更看重基于第一性原理的快速解决问题能力和极致的主人翁意识。 (31:50)

Bending Spoons operates as a unified platform company that acquires digital products off its balance sheet to hold and rebuild forever, relying on extreme talent density and systematic operational leverage rather than traditional conglomerate management. Buy, Hold, and Deeply Replatform (0:44) Unlike private equity firms or decentralized holding companies like Berkshire Hathaway, Bending Spoons buys software assets directly onto its own balance sheet and has never sold a material business in its 13-year history. Luca Ferrari explains that acquired companies do not remain standalone operations; their codebases are rewritten, cloud infrastructure is re-architected, and monetization models are reworked to run on a shared, proprietary technology and data foundation. Fluid R&D, marketing, and G&A teams rotate dynamically across assets such as Vimeo or AOL, allowing roughly 20 high-caliber operators to run businesses that formerly required 200 employees. Quantitative Signals Over Elite Pedigrees Bending Spoons treats recruitment like quantitative portfolio management, testing hundreds of candidate signals against subsequent workplace performance. The firm processed around 800,000 applications in a single year while extending fewer than 300 offers, achieving a regrettable churn rate of just 0.6%. Ferrari observes that degrees from elite universities like Harvard or Stanford carry very slight performance advantages that are frequently cancelled out by candidate entitlement. Conversely, unconventional markers—such as heavy political activism at age 16 or 17 that tapers off by the early twenties—correlate strongly with drive and agency. Structured Evaluations and Reference Rankings (16:06) To eliminate the noise of unstructured interviews, Bending Spoons standardizes questions, mandates interviewers read them verbatim, and assigns independent evaluators to score recorded transcripts. Ferrari avoids standard conversational reference checks, noting that candidate-provided contacts default to politeness. Instead, evaluators force comparative rankings, asking whether an applicant was the single best or among the top five out of twenty peers. Ferrari notes that pressing for concrete rankings often causes glowing initial claims to retreat to average assessments. Prioritizing Candidate Slope Over Industry Experience (31:50) Ferrari deliberately discounts conventional industry experience, arguing that seasoned hires often copy-paste outdated playbooks, struggle to adapt to hands-on cultures, and exhibit declining motivation compared to hungrier junior operators. The company frequently appoints 25- to 27-year-olds to lead acquired units based on two core traits: rapid, first-principles reasoning and extreme ownership. The firm accepts the friction this causes with veteran employees, opting either to replace resistant staff or reassign young leaders if interpersonal friction becomes untenable. Organizational Simplicity and Radical Transparency (46:18) Bending Spoons operates without an executive leadership team, viewing formal committees as status-driven bottlenecks. Work is divided between functional leads who oversee craft and standards, and team leads who manage missions and products. To counter organizational bloat, any proposal adding complexity, policy, or headcount bears the full burden of proof, while simplifications require no justification. Ferrari couples this operational discipline with a personal commitment formed in his youth to never lie and never omit relevant information, practicing immediate, public feedback across the organization. Host Brian Halligan observes that while Ferrari's preference for raw learning trajectory over decades of experience works exceptionally well in product-led software companies, it remains an open question whether the same model translates seamlessly to complex enterprise sales motions. Ferrari maintains that organizational systems belong strictly to the incumbent leader rather than the company, leaving future successors complete freedom to redesign the operational playbook.

## Background 该视频由红杉资本（Sequoia Capital）于2026年10月8日发布在YouTube平台。视频简介介绍了Bending Spoons联合创始人兼CEO Luca Ferrari的独特经营模式：作为AOL、Vimeo和Evernote等知名软件的连续收购者，该公司秉持买入并永久持有的理念，去年从80万名求职者中仅录用不到300人，通过量化对冲式的招募体系、重用高成长斜率的年轻人才以及高度精简的运营系统，构建出远超同侪的软件重构与运营壁垒。 ## TL;DR 在本次对谈中，Luca Ferrari深入拆解了Bending Spoons独特的“平台操作系统”模式——用自有资产负债表收购成熟软件资产并进行底层重写与永久运营，核心壁垒完全建立在极致的人才密度与卓越运营之上。他直言名校背景对工作绩效几乎没有预测力，而他们像量化基金一样回测数百个信号、依靠脚本化面试与实操测试筛选顶尖人才，并大胆让20多岁的年轻业务骨干挑大梁接管大型并购业务。此外，Luca还分享了废除高管团队、推行彻底坦诚、让增加复杂度的人承担举证责任的组织哲学，以及立足米兰打破硅谷垄断的使命感；最后，他阐述了早年经历如何促使自己确立“绝不撒谎、绝不隐瞒”的高诚信底线，揭示了支撑长期商业复利的底层道德契约。 ## Atmosphere & style 整场对话洋溢着智性交锋的深度与极度求真的清醒感，完全剥离了传统企业访谈中的公关套话与行业黑话。主持人Brian与Luca的互动既像两位老练操盘手在战壕里复盘得失，又充满互相启发的反思，气氛坦诚而务实。Luca的叙事风格兼具数学量化般的严密逻辑与意大利式的直率，在谈及早年犯错与道德原则时流露出罕见的情感重量与脆弱感，给整期节目注入了极具穿透力的人性底色。 ## Bulletin - [00:44](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=44s) Bending Spoons 采用资产负债表出资且永久持有的模式，通过自研底层技术平台全面重写被收购软件的代码与基础设施，上市时估值达约 180 亿美元。 - [09:57](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=597s) 公司通过专职科学家将招聘流程量化运作，去年从 80 万份求职简历中仅录用不到 300 人，核心团队令人遗憾的流失率控制在 0.6%。 - [13:34](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=814s) 数据回测表明哈佛、斯坦福等名校光环对绩效预测力极弱且易滋生特权意识，反而是“青少年活跃参与政治、20 岁出头退出”与极高自驱力高度正相关。 - [34:36](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=2076s) 公司内部废除固定高管团队，推行成长斜率压倒经验的人才策略，常派遣 25 岁左右的高潜力员工全面接管被收购业务的研发与运营。 ## Key takeaways - [00:44](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=44s) 商业模式本质是平台型软件操作系统，依靠买入、永久持有并彻底重构底层技术栈（Replatforming）来获取远超传统私募股权的长期复利回报。 - [06:00](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=360s) 并购成功的核心几乎完全归功于投后运营而非挑选标的，高人才密度结合专有技术工具链让 20 人的团队即可支撑传统模式下 200 人的业务规模。 - [09:57](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=597s) 招聘体系犹如量化交易选股，通过回测数百个数据信号验证实际预测力，主观非结构化面试被视为无用噪音，必须由严格脚本化面试与实操测试取代。 - [19:59](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=1199s) 背调采用强制分级迫选法（如要求在共事的 20 人中进行排位）挤出推荐水分，并因合规考量在全球统一弃用盲调（blind reference）。 - [21:45](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=1305s) 决定企业成败的核心是商业战略、护城河与人才密度，具体组织汇报结构对最终经营结果的边际影响仅在正负 10% 左右。 - [25:34](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=1534s) 高层管理者无法同时并发兼顾战略与细节，应通过严格的时序拆分（Sequencing）交替进行宏观战略思考与深入一线泥潭的仆人式执行。 - [30:38](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=1838s) 大厂老兵往往面临环境过时、文化不服与动力衰退三大陷阱，成长斜率（Slope）与纯粹潜质远比过往行业经验更为关键。 - [34:36](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=2076s) 年轻业务领袖挑大梁的胜任力取决于两大核心维度：极高智商驱动的严密逻辑，以及追求成为全球顶尖的极致主人翁精神。 - [39:35](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=2375s) 人才培养的核心杠杆在于赋予远超常规水平的重大责任并置于顶级同侪之中，正规培训与主动教练仅能贡献 10% 到 20% 的边际增益。 - [42:10](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=2530s) 推行彻底坦诚（Radical Candor）的即时反馈机制，个人成长聚焦于“一年死磕一个重大核心缺陷”而非试图全面修补所有漏洞。 - [46:01](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=2761s) 彻底废除常设高管团队，改用按议题召集的务实会议机制，组织图仅被视作自动化流程的算法工具而非不可动摇的特权结构。 - [49:15](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=2955s) 组织系统不应存在圣牛，任何新任 CEO 都有权根据自身特点彻底重写管理规则，拒绝盲从行业常规认知以避免滑向平庸。 - [51:27](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=3087s) 推行“持续极度简化”价值观，任何试图增加流程、功能或人头复杂度的提议者必须自负举证责任，且决策必须具备快速回滚机制。 - [55:42](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=3342s) 扎根米兰是精准的人才与成本套利决策，利用本土优质工程教育红利与竞争真空吸纳顶尖工程师，同时践行打破硅谷单极垄断的使命感。 - [01:01:16](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=3676s) 坚持“绝不撒谎、绝不隐瞒关键事实”的非功利性道德铁律，即便短期内直面人际冲突与代价，也能在长期交易中积累巨大的信任红利。 ## Detailed walkthrough ### [00:00](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=0s) 名校光环的迷思与认知偏差 Luca Ferrari 直言不讳地指出，外界普遍认为毕业于哈佛大学（Harvard）或斯坦福大学（Stanford）的人一定是不折不扣的天才，但根据他们长期追踪的数据来看，名校毕业对实际工作绩效的预测能力极弱。名校毕业生可能确实比普通院校的学生综合素质稍高一点，但差距绝对没有外界想象的那么大。更致命的是，顶尖名校的背景往往伴随着更强的特权意识（entitlement）和自视甚高（thinking too highly of themselves），这种自满情绪反而会迅速抵消掉他们原本可能具备的天赋与潜力优势。 --- ### [00:44](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=44s) Bending Spoons 的核心飞轮：永久持有与彻底重构 主持人 Brian 介绍 Luca 掌舵的 Bending Spoons 是一家极其独特的科技公司：外表看似兼并整合软件资产的私募股权滚式投资公司（PE rollup），本质上却是一家科技平台型企业，在整个科技界几乎找不到可与之类比的对标物。 ```mermaid flowchart TD accTitle: Bending Spoons Acquisition and Integration Engine accDescr: Diagram illustrating how acquired companies are migrated onto the unified operating system and long-term hold model. subgraph Acquisition ["收购阶段 (Acquisition)"] A["评估高潜力软件业务"] --> B["资产负债表出资收购<br/>(Buy to Hold Forever)"] end subgraph Replatforming ["平台化改造 (The Platform OS)"] B --> C["底层重构与重写<br/>(Replatforming)"] C --> D["共享基础架构与技术栈<br/>(R&D / 营销 / G&A)"] C --> E["代码库重写 & 云架构优化"] end subgraph Compounding ["长期复利 (Value Compounding)"] D --> F["产品体验与商业化升级"] E --> F F --> G["超高资本回报<br/>(持续复利运作)"] end ``` 谈及商业模式，Luca 阐明了公司赖以建立的底层逻辑：如果能打造出一套系统，让运营数字技术业务的效率超越几乎所有同行，就能通过并购在极长时间维度内高效复利资本。由于被收购的业务并入这套平台系统后的表现远好于作为独立公司运营，Bending Spoons 能开出对卖家极具吸引力的收购价格，同时依然为自身股东创造高额回报。过去 13 年里，团队倾尽心力打造这套平台——汇集顶尖团队、强势文化、自主研发的专有技术及海量专有数据： - **深度整合与共享资源**：旗下拥有流动于各业务线之间的共享研发（R&D）、市场营销和行政管理（G&A）团队，能根据机会灵活调配。 - **底层技术统一化**：将收购来的产品彻底迁移到底层技术栈（Replatforming），投资人常将其比喻为一个标准化的“操作系统”（Operating System）。 - **推倒重来式的重构**：从零重塑组织架构、重写大块代码库、重构云基础设施，同时推出实用新功能、修复 Bug、提升性能并重新梳理变现与营销逻辑。 - **与传统私募股权（PE）的区别**：不设立基金周期，直接用自身资产负债表收购；从未出售过任何核心业务，也无意出售，始终坚持“买入并永久持有运营”（Buy to hold and operate forever）。 - **从独立公司到“内置应用”**：被收购对象不再作为独立实体运作，而是演变为安装在统一底层系统之上的功能型产品。 在被问及最新估值时，Luca 提到公司数月前上市（IPO）时的股权价值约为 180 亿美元。Brian 感慨这种模式让他想起年轻时在 PTC 工作、被派往日本三菱（Mitsubishi）总部的经历，感觉 Bending Spoons 更像是由核心指挥部强力驱动的三菱或三星（Samsung），而非伯克希尔·哈撒韦（Berkshire Hathaway）或 PE 基金。Luca 认同并强调，大多数连续收购者倾向于让标的公司保持现状并独立运作，而 Bending Spoons 则是自底向上全面重构，虽然极其耗时艰巨，却能换来极高的投资回报率。 --- ### [06:00](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=360s) 选品 vs 运营：80 万简历筛 300 人的操作底气 当被问及成功的秘诀有多少来自“标的选择”（Picking），多少来自“投后运营”（Operating）时，Luca 明确表示：**几乎百分之百在于运营**。虽然团队在挑选资产上也日趋成熟，但真正拉开回报差距的是运营平台的支撑。 这种运营优势的核心支柱是极致的人才密度。Luca 透露，去年公司收到了多达 800,000 份求职申请，但最终录用的人数不到 300 人。拥有顶尖的工程师、设计师、产品经理和业务负责人后，重构并高效管理一家业务变得容易得多。任何长期操盘数字化业务的人都会明白，员工人数并不能决定产出，关键在于团队的个人能力与文化建设。凭借高素质团队，Bending Spoons 用 20 个人就能跑顺别人需要 200 人才能支撑的业务。这种精简不仅直接砍下巨额成本，更能大幅加快创新步伐和商业化迭代速度。此外，过去十年中公司自主研发了一整套原生互通的专有技术工具链，覆盖数字业务运营的方方面面，形成了普通独立运作公司根本无力负担的技术“作弊码”（Cheat code）。 --- ### [08:01](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=481s) 冷启动破局：早期创始人下场的人才“飞轮” 面对当今科技界言必称“人才密度”却在扩张中不可避免被稀释的现状，Brian 追问 Bending Spoons 如何将这一理念工业化。特别是在早期默默无闻、毫无雇主品牌（Employer brand）、也没有主动求职流量的阶段，这套飞轮是如何启动的。 Luca 回顾道，只要商业策略行得通，剩下的关键全在于执行策略的人才以及界定协作规则的组织文化。在公司一穷二白的创立初期，早期团队必须学会“越级挑战”（Punching above their weight）： - **创始人全力以赴**：在公司创立的前两到三年里，Luca 至少把 50% 甚至更多的时间投入在招聘及相关人才事务上。 - **主动出击挖掘生源**：当时招聘范围主要局限在意大利本土（现已全面国际化），他和少数几位初创成员四处搜寻意大利顶尖名校的优秀毕业生名单，通过极其严苛的定向挖掘来建立最初的人才蓄水池。 ### [09:57](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=597s) 量化交易般的招聘体系与飞轮效应 Luca Ferrari 回忆起公司创立初期，他们设法拿到应届毕业生的名单并逐一追踪线索，他甚至亲自给数百人打电话，甚至开玩笑说自己当时“简直是在骚扰其中的一些人”，直到对方同意加入。随着公司发展，Bending Spoons 组建了一支规模庞大的招聘团队，配备了专门的科学家与研究人员。他们对待候选人管道的态度与选拔方式，完全就像“量化交易员在挑股票”：系统性地挖掘所有潜在信号，将其与入职后的未来绩效进行回测，验证哪些信号或信号组合真正具备预测力，并在循环中不断迭代。目前，他们在每份申请中会评估数百个不同的信号。 这种严谨的筛选机制使他们在同等雇主品牌和申请量下，能够选拔出远超行业平均水平的顶尖人才。与此同时，Bending Spoons 独特的商业模式也带来了巨大的引才优势。许多优秀的工程师、设计师和产品经理非常喜欢这种平台型机制：他们可以在同一家公司内，先花 9 个月重构 Vimeo 的视频底层架构，接着花 6 个月重新构想 AOL 的广告系统，随后再花一年时间开发底层平台技术。他们不必因跳槽而稀释个人声誉或重新建立信任，却能面对多样化的挑战，并在相同的文化与优秀同事圈子中工作。 对人才密度的极致偏执构成了公司强大的自留与飞轮机制。Luca 透露，去年公司核心团队（内部称作“Spooners”）的“令人遗憾的流失率”（regrettable churn）仅为 0.6%，尽管硅谷科技巨头对他们的团队持续不断地展开挖角。高人才密度一旦确立，团队成员便会主动拉拢身边最优秀的同学与旧友。然而，这种壁垒必须在创业初期就建立，过程也极其痛苦：为了等待卓越的人才，公司不得不长期面临人员短缺；甚至必须下定决心辞退那些称职、有贡献但并非顶级的人选。“没有人是反社会人格”，解雇表现尚可的员工并在短期内牺牲即战力以换取长期的超群战斗力，是违背人性的艰难抉择，但唯有如此才能形成良性飞轮。 - 招聘团队配备专职科学家与研究员，单份申请评估信号达数百个。 - 核心团队人员去年令人遗憾的流失率仅为 0.6%。 - 业务模式允许技术人才在同一种文化和声誉积累下跨项目轮岗（如 Vimeo 架构、AOL 商业化等）。 ```mermaid flowchart TD accTitle: 人才密度飞轮效应 accDescr: 展示从极致筛选到低流失率再到吸引顶尖人才的良性闭环 subgraph Selection["量化选拔"] A["数百个数据信号回测"] --> B["严苛门槛 & 宁缺毋滥"] end subgraph Retention["核心团队"] B --> C["极高人才密度"] C --> D["多元业务轮岗实践<br/>(Vimeo / AOL)"] D --> E["0.6% 遗憾流失率"] end subgraph Attraction["虹吸效应"] E --> F["口碑推荐与名声扩散"] F --> A end ``` --- ### [13:34](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=814s) 预测信号解码：名校光环的幻觉与反常识规律 当被问及哪些信号的决定系数（$R^2$）最高、最能预测成功时，Luca 坦言出于竞争壁垒，细节如同量化机构的模型一样保密，但他分享了几项关键洞见。像 GPA 这样的常规指标有一定帮助，但预测性并不算强；而候选人毕业于哈佛或斯坦福等顶尖名校的预测力则出奇地低。大众通常认为常春藤名校生全是天才，但数据表明，他们相较普通高校毕业生的实际能力优势微乎其微。更致命的是，顶尖名校毕业生往往容易滋生特权意识（entitled）和自命不凡的心态，抵消了原本微弱的智力优势。 在诸多非显性信号中，有一条高度相关的反常识规律：**在十几岁青少年时期积极参与政治活动，但在 20 岁出头时停止政治参与的人，其内驱力和创业精神通常极高**。Luca 解释了背后的逻辑：如果一个人 17 岁时觉得学校的条条框框不够用，主动去社会中折腾，这展现了极强的自主性（agency）、积极性和不满足于现状的健康特质；但如果到了 25、26 岁乃至 30 岁依然全身心扑在政治上，政治大概率已成为其毕生使命，这本身值得尊重，但往往意味着本职工作只能退居次席，难以将职业追求置于首位。 - 名校（哈佛、斯坦福）标签在实际表现中的正向预测力极其微弱，且常伴随自满与特权心态。 - “青少年时期参与政治、20 岁出头淡出政治”高度正相关于候选人的驱动力（drive）与创业心态。 - 候选人往往会在简历中主动列出早期的领导力经历，公司后续也会在笔试与面试中设置开放问题主动印证这一特征。 --- ### [16:40](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=1000s) 消除面试噪音：结构化问询与背调边界 针对如何获取这些关键信号，Luca 强调评估绝对不能停留在被动查阅简历信息上，而必须主动施加刺激。在 Bending Spoons，最具预测力的信号往往来自专门设计的实操测试（practical tests）。他尖锐地指出，在传统招聘环节中，**非结构化的主观面试是预测未来表现最差的工具**。如果面试官只是毫无约束地跟候选人聊天，仅凭高层级的议程和个人直觉判断“我喜不喜欢这个人”，产出的结果几乎纯属噪音。要让面试具备预测价值，就必须对其彻底结构化：给面试官下发极度精准的话术脚本，要求一字不差地依样提问，并将面试执行与后续的答案评估彻底解耦，由独立人员进行打分。 对话中，主持人 Brian 交流了 HubSpot 早期将电销团队从 10 人扩张到 1000 人时的经验：他们同样不迷信哈佛和 MIT，而是锁定佛蒙特大学、威斯康星大学等州立大学中具有竞技体育背景的人选；考核时不进行常规面试，而是安排 12 分钟的角色扮演实战卖货，随后当场给出负面反馈，观察对方在消化反馈后进行第二轮演练的学习吸收能力，最后配以“盲背调”（blind references）。 当被问及 Bending Spoons 是否也使用盲背调时，Luca 澄清道，背调是他们的最后一步环节，但他们**绝不做盲背调**。背调由候选人主动提供联系人名单，而不会私下联系未授权的知情者。这背后的原因非常务实：在意大利，绕过候选人进行盲背调是违法行为；尽管美国允许任意致电前同事，但为了在全球不同司法管辖区保持招聘方法论的一致与合规，Bending Spoons 统一采用了透明列名背调的标准。 - 非结构化面试几乎等同于随机噪音；有效面试必须采用脚本化提问，并实现“提问者”与“评估者”职责分离。 - 实操测试（practical tests）是预测表现的核心手段，重在主动诱发深层特质。 - 因意大利法律明确禁止私下进行盲调（blind references），Bending Spoons 采用跨国统一的合规背调体系，仅联络候选人提供的推荐人。 ### [19:59](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=1199s) 科学化背调：如何从“全是好评”中逼出真相 背调（Reference Check）如果只凭直觉打个电话，结果几乎必定毫无价值。候选人绝不会主动提供会批评自己的人选，因此电话那头往往是一片赞誉。Bending Spoons 的做法是将其高度结构化与科学化：他们引导候选人提供对其实际表现有充分了解的人，尤其是候选人过往几段最核心经历中的直属上司。 为了打破这种“全员好评”的虚假信号，背调提问必须极其尖锐，迫使推荐人做出艰难的分级决策。例如，他们会直接发问：“在你共事过最深入的 20 个人里，这位候选人是绝对的第一名，还是能排进前五？”紧接着追问具体细节以核实含金量。许多原本态度热情的推荐人，在严密追问下往往改口称“属于中上水平”。Luca Ferrari 指出，这立刻就戳破了泡沫——候选人听起来像个绝世天才，实际上在 20 人里可能只排到第 8 名。背调确实具备预测效力，但前提是必须投入足够的精力和结构化设计。主持人也深有同感，提到自己在做盲调（Blind Reference）时最喜欢直接抛出杀手锏问题：“这是不是为你工作过最优秀的人？”当对方迟疑着说“不是”时，真正的真实对话才算展开。 - 背调核心要求：至少覆盖候选人最重要的两段经历，且最好能直接联系到当时的直接管理者。 - 迫选分级法：不问“他表现如何”，而是问“在你看过的 20 个人里，他能排前五还是第一”，用具体的强制排序挤出水分。 --- ### [21:45](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=1305s) 组织架构是伪命题：战略与人才密度碾压管理形式 主持人谈到自己采访了 20 位 CEO，最初的设想是总结出一套标准化的管理共性，结果发现根本没有通用法则。当前的两种极端代表了管理的剧烈分化：一是 Kalshi 创始人 Tarek Mansour（TK）与联合创始人实行“混乱管理”（Management by Chaos），全公司约 200 名员工完全扁平，没有中间管理层，所有人直属两位创始人；另一端是 Databricks 的 Ali Ghodsi，拥有极度严密的管理层级，8 位直属汇报，每周开 3 次例会，每季度召开 QBR（季度业务复盘）和战略研讨会。面对这两种截然相反的模式，Luca Ferrari 认为自己处于两者之间，并直言业界严重高估了所谓“正式管理方法”对企业最终结果的影响。 在 Luca 看来，一家公司的成败几乎完全由商业战略（包括产品市场匹配度 PMF 和竞争壁垒 Moat）决定。他衡量一门好生意的标准非常苛刻：“如果管理层表现平庸，这家公司是否依然能取得成功？”排在第二位的决定因素是人才密度与企业文化（即决策逻辑、价值取向和优先级判断）。至于汇报线、实行 OKR 还是扁平管理、管理幅度（Span of Control）等具体手段，对最终产出的边际影响顶多在正负 10% 左右。无论是 200 人的扁平混乱组织，还是层级森严的巨头，只要拥有极高的人才密度和以商业结果为导向的求真文化，卓越的团队自会跨越结构上的缺陷；反之，哪怕组织结构再完美，战略与人才缺失也注定失败。 ```mermaid flowchart TD accTitle: 企业成败决定要素层级 accDescr: Luca Ferrari 认为商业战略与护城河起主导作用，人才与文化次之，组织管理架构仅产生微小的边际影响。 subgraph Core["核心决定因素（决定绝大部分成败）"] A["商业战略与壁垒<br/>(Strategy & Moats / PMF)"] B["人才密度与核心文化<br/>(Talent Density & Culture)"] end subgraph Margin["边际微调因素（影响在 ±10% 范围）"] C["汇报层级与管理跨度"] D["目标工具 (OKR vs 灵活机制)"] E["会议制度与例行流程"] end A -->|"奠定商业基本盘"| B B -->|"驱动实际业务运转"| Margin ``` - 核心商业准则：好生意的标志在于“即使由平庸的管理团队操盘，公司依然能跑赢大盘”。 - 组织结构容错率：只要人才极其出色、文化务实理性，过度混乱或过度官僚的架构漏洞都会被团队自主弥补，管理框架绝非企业胜出的第一因。 --- ### [25:34](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=1534s) 领导者还是管理者？仆人式领导与接地气的泥潭作战 针对主持人“你擅长做领导者还是做管理者”的追问，Luca 展现了强烈的自省与危机感。他承认自己习惯看到“半空的杯子”（Glass half empty），总在怀疑自己是否做得足够多、是否达到了期望。他坦承自己两方面都谈不上完美，但自认更偏向领导者，而在传统的管理细节（如规范地主持一对一沟通等）上做得很一般。他的领导力并非依靠好莱坞式的宏大演讲——“我不是威廉·华莱士（电影《勇敢的心》主角）那种一发表演说就能让人热血沸腾的领袖”——而是源于“仆人式领导”（Servant Leadership）：自己拼尽全力工作、始终与一线员工在战壕里摸爬滚打，以言行一致来赢得信任和激发团队。 关于埃莉诺·罗斯福那句“仰望星空，脚踏实地”（Head in the clouds, feet on the ground），Luca 分享了自己的微观平衡哲学。他认为自己最突出的长处在于对大局（Big picture）的把握：能极其敏锐地剔除噪音，洞察核心商业机会的逻辑走向。但他坚决反对脱离战壕的“象牙塔管理者”。如果领导者不能扎根泥潭深处接触实际业务，就既无法校准宏观战略的真实性，也无法指导或评估团队成员。对于如何同时兼顾宏观与微观，Luca 总结的心得不是在同一时刻多线程并行，而是严格进行“时序拆分”（Sequencing）——留出专属的时间段彻底放空仰望星空、构思宏观战略，其余时间则全身心深扎一线细节。 - 领导风格定位：践行仆人式领导，不靠煽情演说激励人，而是通过与员工并肩作战、以身作则建立影响力。 - 战略与细节的平衡术：战略大局与一线实操必须兼备；两者无法同时并发处理，最佳方案是按时间段交替切换（Sequencing），分别进入深度思考与泥潭执行状态。 ### [29:46](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=1786s) 执行与构想的交替，以及管理辅导的本质 Luca Ferrari 坦言，作为管理者很难在同一瞬间既当仰望星空的“架构师”，又当脚踏实地的“执行者”。现实中往往是交替进行的：一段时期他专注于宏观问题与重大机遇的设计，随后便进入浑身沾满泥巴、付出“鲜血、汗水与泪水”的执行状态。这两种状态难以并存，但可以在时间线上依次切换。 谈及对下属的辅导（Coaching），主持人提到自己在执掌数千人上市企业时，常常会觉得自己根本不够格去辅导某些高管，并询问 Luca 是否也有同感。Luca 毫不讳言自己作为 Coach 经常觉得自身能力不足。但在 Bending Spoons，团队从不假设管理者必须能解决或指导直接下属遇到的一切问题。在他们的机制里，指导是去中心化且跨越汇报关系的——任何人只要在某方面有所擅长，就应该去辅导他人。优秀的管理者应当是个聪明的问题解决者、提供情绪支撑并愿意亲历亲为的通才，但这并不意味着管理者在具体业务细节上总是比一线员工更有能力解决问题。 ### [30:38](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=1838s) 经验的诅咒：为什么大厂老兵不一定适合高增长团队 硅谷各家公司常常纠结于一个核心命题：到底应该提拔内部处于上升期但未见过大风大浪的新人，还是从外部聘请“见过世面”、经验丰富的老手？Luca 明确表示，尽管他们重视经验，但 Bending Spoons 对经验的看重程度远低于市场平均水准。他指出了倚赖“老资历”的三大根本缺陷： ```mermaid flowchart TD accTitle: 盲目迷信经验的三大缺陷 accDescr: 展示了外部经验在新环境下的三大问题，包括环境时效性失效、文化水土不服以及长期动力衰退。 subgraph Trap["盲信经验的缺陷"] A["时代变化飞速<br/>(几年前的旧解盲目套入新环境)"] B["文化水土不服<br/>(大厂政治化作风 vs 极度坦诚与卷起袖子干)"] C["动机随经验递减<br/>(20年经验打6分 vs 2年经验打满分10分)"] end Trap --> D["核心决策倾斜：押注高斜率与纯粹才华"] ``` 1. **环境变迁太快**：某人在七年前行之有效的做法，今天很可能完全不适用。老手往往会对全新的上下文视而不见，盲目生搬硬套旧方案。 2. **文化难以融合**：大厂培养出的领导风格并不契合 Bending Spoons。Bending Spoons 要求全员动手、拒绝办公室政治、崇尚“极度坦诚”（Radically Candid），且每个人都必须能够独当一面。大型科技公司出来的资深管理者往往难以承受这种文化剧变。 3. **驱动力衰退**：随着从业年限增长，人的动力往往会下滑。在现实权衡中，一个拥有 20 年经验但动力只有 6 分的人，往往远不如一个仅有 2 年经验却拥有 10 分饱满激情的人。 在劳动力市场中，才华与经验之间往往存在权衡取舍。Luca 强调，经验可以通过时间和耐心给予新人，但才华与潜质却无法后天注入。 ### [34:36](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=2076s) 25 岁接管并购企业：胜任者的两大核心特质 当 Bending Spoons 收购一家公司时，经常会发生被收购方团队错愕的场面：新东家派来接管技术甚至担任一把手的业务负责人往往只有 25 岁或 27 岁。面对被收购公司那些拥有 20 年到 30 年工龄的老员工投来的怀疑眼光，Bending Spoons 始终依靠这些年轻领袖的实际战绩来打破质疑。 虽然派遣的新人年仅二十多岁、仅有三四年的从业经历，但他们此前必然在较小规模的团队或业务上通过了验证。在挑选这些担当重任的年轻人时，Bending Spoons 最看重两项核心素质，它们构成了 80% 的成功概率： - **聪明（Smart）**：具备极强的快速学习能力，并在思考上始终维持高水准的理性逻辑。无论做诊断还是给方案，其推演过程必须严丝合缝、完全合乎逻辑。 - **极致的主人翁精神（Extreme Ownership）**：对成为本领域全球顶尖、为团队和公司做出最大贡献怀有一种近乎狂热的执念。一旦遇到障碍或能力短板，会想尽一切办法自我迭代并迅速攻克。 对于沟通技巧、亲和力等为人处世的维度，Luca 视其为基础“卫生因素”（Hygienic Factors），而非决定性的杀手锏。 ### [37:29](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=2249s) 青年领袖失效的罕见情景与应对 在极少数年轻人挑大梁却以失败告终的案例中，Luca 总结原因通常落在两端： - **情商与同理心盲区**：领导者虽然极度聪明且执行速度如飞车狂飙，却对团队的情绪感知钝化。当下属陷入困境或情绪不对时，他们没能及时察觉并在恰当的时间说出恰当的话，导致团队产生抵触情绪和阻力，最终让管理难以为继。 - **老员工的经验自傲与免疫排异**：部分老员工由于年龄和资历，心理上无法接受被一个年轻资浅的人领导，即便对方能力出众且为人谦和也无济于事。 面对第二种组织“免疫排斥反应”，Luca 直言解决路径通常只有两条：要么让那些抗拒领导的老员工调岗或离开，要么撤换这名年轻主管，具体取决于实际场景中哪种选择对业务更有利。主持人随后追问，Bending Spoons 究竟是通过系统性的培训和教练来培养这些年轻骨干，还是纯粹把他们扔进深水区让他们自我摸索。 ### [39:35](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=2375s) 让人才加速成长的两个杠杆：极端责任与顶级同侪 在人才培养上，Luca Ferrari 强调虽然公司内部有培训材料，但在他看来，90% 的培养成果取决于把新人“直接扔进深水区（throw them into the deep end of the pool）”。具体做法是赋予他们远超合理水平的重大责任，同时在他们周围围绕一群极其优秀的同事。 Luca 认为，人最核心的学习途径就是与卓越的人共事： - 身边的人总有在某些方面比你强的地方，如果你努力效仿这些优点，就能学到真本事； - 身边的人同样会在某些方面做得很糟糕，看清他们的弱点，可以避免自己踩入同样的陷阱。 责任的极限压强加上顶级的人才密度，构成了驱动员工能力飙升的最强动力。少量的正规培训和主动辅导（active coaching）固然能锦上添花，但最多只能带来 10% 到 20% 的边际增益。 ```mermaid flowchart LR accTitle: Talent Growth Drivers accDescr: Luca Ferrari's framework showing how talent grows at Bending Spoons through deep responsibility and talent density. subgraph Core["核心驱动因素 (占 90%)"] A["推入深水区<br/>赋予极端责任"] B["高密度优秀同侪<br/>观察长处与短板"] end subgraph Supplement["辅助增益 (占 10-20%)"] C["正式培训材料"] D["主动教练与辅导"] end Core --> E["人才斜率快速爆发"] Supplement --> E ``` --- ### [40:28](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=2428s) Brian 的 360 度评估：从自命不凡到发现“漏洞” 主持人 Brian Halligan 分享了自己在担任 HubSpot CEO 期间经历的一次深刻转折。当时公司决定为他进行 360 度环评，由于他的联合创始人极度内向、不愿与“其他人科动物（homo sapiens）”交谈，便别出心裁地设计了一份净推荐值（NPS）调查问卷： 1. 在 1 到 10 的打分范围内，你有多大意愿推荐 Brian 继续担任 HubSpot 的 CEO？ 2. 给出该分数的具体原因是什么？ 这份问卷发给了董事会成员、客户及员工等约 25 个人，大家写下的反馈详尽得犹如长篇小说。联合创始人随后将数据整理成直方图，并把正向反馈归类为一项项“功能（features）”，配上原汁原味的受访者原话。Brian 阅读报告的前 10 页时信心爆棚，满脑子都是“这功能我具备、那功能我也擅长，我简直是个天才 CEO”。然而翻到第 11 页，内容急转直下，变成了铺天盖地的“漏洞（bugs）”列表与直接引语，场面极其刺痛却又极具价值。 更令 Brian 意外的是，有些大家眼中的“漏洞”原本是他自以为的“功能”——比如过度热情和极度兴奋有时会给团队带来负面压力。第一年他曾试图把所有漏洞全部补齐，但后来意识到这并不现实，于是转为每年只挑选一到两个核心缺陷进行攻坚。他借此向 Luca 询问：Bending Spoons 是如何获取反馈并持续进化的？ --- ### [42:10](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=2530s) 彻底坦诚：直面问题的反馈文化与“一年死磕一个缺点” Luca 表示 Bending Spoons 高度依赖反馈机制，公司内部推行“彻底坦诚（Radical Candor）”。其核心准则是：只要你不是为了宣泄情绪，而是坚信这有助于对方进步、对团队或公司有益，就必须在发现好与坏的第一时间讲出来。员工无需过度纠结用词是否绝对精准、是否会伤害对方的感情；尽量礼貌友善固然好，但绝对诚实永远排在首位。 公司鼓励面对面的即时反馈。如果在会议中发现了问题，哪怕当着全场参会人员的面，也应该立刻提出来，不必延后；若是通过 Slack 等异步工具发现的问题，则可以直接发消息沟通。除了日常的高频反馈外，Bending Spoons 每年还会进行一次正式的结构化 360 度评估，由 6 到 10 位评估者对个人进行全方位诊断。 在个人提升上，Luca 也遵循了与 Brian 相似的路径。他早先曾试图一年改进 5 个缺点，但年中就发现人的精力和攻坚难度根本支撑不了这么多目标，因此确立了“一年只攻克一个重大提升项”的原则： - **Luca 当下的年度攻坚缺陷**：他由衷认为自己这辈子最大的特权就是能与身边这群极度优秀的人共事，但他非常不擅长当面表达赞赏。 - **自我苛责的心理根源**：Luca 对自己极为严苛，每天晚上收工时往往都会对自己感到失望，因此对别人也是同样的挑剔标准。 - **改进进展**：他正在强迫自己将内心深处的认可如实大声说出来，他评价自己目前的改进是“一半成功、一半失败”，但相比过去已经有了明显长进。 - **CEO 的表扬困境**：Brian 笑着回应称自己过去也是如此，外界常说的“每提一个批评前先给五个表扬”，在他们这里往往变成了“把所有 6 个名额全都用来挑毛病”；Brian 甚至一度需要在周五下午的日历上专门划出 15 分钟日程，强制提醒自己去赞美团队。 --- ### [46:01](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=2761s) 废除高管团队：按需组会与矩阵式组织算法 针对很多初创公司 CEO 容易陷入“信任圈极窄、仅依赖核心四人高管层”的现象，Luca 表示自己在 Bending Spoons 并不存在这种反馈，因为他们甚至直接取消了所谓的高管团队（executive team / leadership team）。 早年间 Bending Spoons 曾设立过高管团队，但后来果断废弃。Luca 解释，明确界定这群人反而会带来坏处： - 某些议题因为必须在“高管层”讨论，导致一些不需要参与讨论的人被迫卷入； - 真正精通该业务的关键一线成员，却因为没有“高管”头衔而被排除在关键会议之外。 现在公司遵循极度务实的原则：需要谁参与就叫上谁，彻底抛开职级与头衔。例如在制定战略的高层讨论中，Luca 只会邀请两类人：能为议题做出最大实质贡献的人，或者能从会议中汲取关键认知的人；“为了照顾某人的自尊心而邀请他参会”在公司是绝对不被允许的。 在组织架构上，Luca 澄清他们并非没有组织架构图，而是将其固化为一套运行“算法”的矩阵式组织： - **职能（Functions）**：负责人主抓“怎么做（the how）”，聚焦于专业硬技能的打磨； - **业务团队（Teams）**：负责人主抓“做什么（the what）”，聚焦于具体的业务目标以及推动公司成功的核心成果； - **汇报节点**：每位员工同时向一位职能主管和一位团队主管汇报，所有的线条最终汇聚到单一根节点（目前该节点就是 CEO Luca）； - **去神圣化与高流动性**：组织图只是一个用于自动化大量流程的工具，没有任何人应当受到它的限制。在 Bending Spoons，管理角色与个人贡献者（IC）之间的转换非常自由，一个人担任一段时间的 Lead 之后重新转回 IC 极为常见。 ### [49:15](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=2955s) 组织架构没有圣牛：下任 CEO 应当重写系统 Brian Halligan 提到英伟达（NVIDIA）的黄仁勋（Jensen Huang）也有类似的领导风格：黄仁勋总是在公开场合直接给出正面与负面反馈，从不搞一对一会议（1-on-1s）。当 Brian 问黄仁勋未来退休后接班人是否会继承这套系统时，黄仁勋斩钉截铁地回答：“绝不可能（Hell no）。这套体系是为我量身定制的，符合我的工作方式；换了别人来，就应该彻底重写一遍。”Brian 借此询问 Luca Ferrari，此前他也表达过类似观点，即只要他还是 CEO 就这么干，他是否认同黄仁勋的哲学。 Luca 完全赞同这一逻辑。他用执教体育队伍打比方：如果有教练声称“不管手里有什么球员，获胜的方式都只有这一种”，那他就是个傻瓜。胜负取决于你拥有的团队成员。因此，如果 Bending Spoons 未来更换 CEO，Luca 会鼓励继任者主动审视是否需要变革。如果继任者觉得现状运转良好，当然可以保留；但只要他们认为改动能带来提升，就拥有绝对的自由去重写规则。 当谈及“第一性原理”与常规认知（conventional wisdom）时，Luca 指出他们从零开始打造 Bending Spoons 时就选择质疑一切。这并非自负地认为自己无所不知，而是因为团队有着极其宏大的野心——建立有史以来最成功的公司之一。明知失败概率极高的情况下，唯一的生机就是走一条完全不同的路：“如果你去模仿别人都在做的事，你注定只能变得平庸（guaranteed to be approximately mediocre）。”因此他们从不在意所谓的主流惯例到底是什么，而是针对每一个重要问题，基于当下的具体情境和愿景去寻找最优解。Brian 追问常规认知难道全错吗，Luca 笑着打趣说，常规认知里可能也有对的，比如“待人友善”，这一点他举双手赞成。 ```mermaid flowchart TD accTitle: 组织哲学与决策路径 accDescr: 对比模仿常规认知与基于第一性原理探索最佳实践的两种发展路径 subgraph ConventionalPath ["常规认知路径"] A["盲目遵循行业惯例"] --> B["体系固化且僵化"] B --> C["走向平庸或边缘化"] end subgraph SpoonsPath ["第一性原理与领导更迭路径"] D["基于情境定制系统"] --> E["极高抱负 + 独特打法"] E --> F["下任 CEO 到任"] F --> G{"全面审视系统"} G -->|"提升效率"| H["彻底重写 / 重构系统"] G -->|"运转顺畅"| I["自主选择保留"] end ``` ### [51:27](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=3087s) 极度简化与举证责任：增加复杂度的人必须自证合理 Brian 指出当下的 AI 原生（AI-native）公司节奏极快，HubSpot 花了八年才做成的事，新公司两三年内就能搞定国际化甚至并购。他在 HubSpot 时曾是“极度简化”的信徒，甚至开会时会带一顶“NO 帽子”（No Hat），因为 HubSpot 更有可能“撑死而非饿死”（die of overeating and starvation）。然而如今的初创团队似乎人人戴着“YES 帽子”四处出击且成果颇丰。 Luca 坚定地认为，“极度简化（radical simplifier）”依然是制胜战略。复杂度直接意味着更高的成本、更慢的节奏和更差的产出。尽管今天的技术、AI 与基础设施让人们能以同样资源完成远超 1800 年代或 1990 年代的工作量，但“少做胜过多做”的底层法则并未失效。很多公司看似同时推进四件事也成功了，但这并不代表这是最优解；如果砍掉最小、最不重要的第四件事，集中精力打磨最重要的核心，核心业务可能会再增长 35%，整体净回报反而更高。 在 Bending Spoons，他们甚至设立了一项考核维度的企业价值观——“持续极度简化（relentless simplification）”。团队制定了极其严苛的规则： - **举证责任倒置**：任何人在系统中提出增加复杂度的建议——无论是流程中增加一个步骤、产品增加一个功能，还是团队增加一名人头——举证责任（burden of proof）完全由发起提议的人承担；反对增加复杂度的人无需提供任何证据。 - **快速回滚机制**：推进该项新增举措必须具备清晰无可辩驳的证据，或者该决策必须是可逆的（revertible）。推动者必须在事后快速复盘该举措是否取得明确成功，否则必须迅速将其撤销回滚。 - **对习惯性复杂度保持警惕**：人类会对长期存在的复杂度习以为常并陷入“盲区”。公司极力鼓励团队成员对早已存在的陈旧流程和制度发起挑战，哪怕这种质疑在当下会被当成麻烦，也必须给予掌声与赞赏。 - Luca 曾于 2010 年创立过一家 AI 初创公司，后于 2013 年关闭清算，这让他对底层技术红利深信不疑，但技术突破并不等同于可以随意堆砌业务复杂度。 ### [55:42](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=3342s) 为何选择米兰？“不酷”本身就是一种酷与使命感 Brian 提到许多硅谷信徒认为离开硅谷就建不成伟大公司，硅谷虽是最好的起步地，但极其昂贵、缺乏忠诚度，规模化极其困难。他问 Luca 为什么把总部设在意大利米兰，以及远离硅谷是否会觉得自己不够“酷”。Luca 幽默地回应：“我一点都不在乎酷不酷。一个跑去买 AOL 的人，怎么可能在乎酷不酷呢？”随后他又调侃道，或许在某种意义上，“不酷本身就是一种极致的酷（the uncool is cool）”。 Luca 坦言自己没在旧金山创业过，但他推断如果一家公司是全宇宙最受追捧、最具潜力且资金最充裕的顶尖项目，旧金山确实是能以最快速度吸纳最多顶尖人才的胜地；但如果只是一个处于中间梯队的项目，那在硅谷规模化就会步履维艰。尽管 Bending Spoons 如今已是高度国际化的企业（在意大利境外招聘的人数远超本土，并正计划在未来一年内开设美国办公室），但当年与联合创始人在丹麦读完工程学位、关停第一家 AI 公司后，团队是蓄意选择回到米兰起步的。这一决策基于两大考量： 1. **商业与人才套利（极其现实的考量）**：意大利拥有 6,000 万人口和极高水平的基础工程教育，但本土却极少有优秀的现代科技企业。许多顶级工程师要么屈就于平庸平淡的传统公司，要么为了前途被迫背井离乡离开家庭。Bending Spoons 带着极大的野心与充满吸引力的愿景回到米兰，为这些人才提供了两全其美的方案——既能在本土拥有一份激动人心的前沿科技工作，又能与家人生活在一起。这让团队得以用相同甚至更低的成本，组建起一支在国际上具备极强战斗力的卓越团队。 2. **使命感（Missionary）驱动**：Luca 和联合创始人深信，全球的知识、机会与财富如果能更均衡地分布，世界将会变得更加美好，而不应全部过度聚集在极少数几个地区。 - 意大利 6,000 万人口的工程师教育红利与本土科技公司真空，构成了 Bending Spoons 早期最大的人才护城河。 - 公司正进一步加大全球招聘力度，米兰本土以外员工比例持续扩大，并正筹备进入美国开设办事处。 ### [58:54](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=3534s) 拒绝硅谷垄断：在非典型土壤证明雄心 Luca Ferrari 谈到了自己对全球科技创新生态分布的深层思考。他坦言，世界上存在一两个像硅谷这样“略高一筹”的超级创新枢纽并非坏事，全球经济要实现极致繁荣或许确实需要这种集聚效应。然而，一旦这种差距拉大到不可逾越的鸿沟，导致世界上所有至关重要的事情全部集中在一两个地方发生，就会带来严重的系统性问题——长远来看，这甚至对生活在硅谷本地的人同样弊大于利。 正是出于这种反思，Bending Spoons 团队立志在米兰扎根，贡献属于自己的“1%的改变”。他们希望向世界证明，即便在意大利这样看似不可能诞生顶级科技巨头的非典型土壤上，同样能够孕育出抱负宏大、击败全球同行的世界级软件公司。Luca 幽默地承认，作为意大利人，选择米兰当然包含了本土文化理解更深、起步更顺手的便利因素，但他们从不后悔这个选择。“你永远无法对人生进行 A/B 测试（You can never AB test life）”，虽然无法断言留在硅谷是否会发展得更好，但他确信扎根米兰是团队做出的最佳抉择。 - **避免单极垄断**：科技界如果将所有资源、人才与注意力彻底倾斜给极少数超级枢纽，将损害全球整体活力与创新多样性。 - **以身作则的雄心**：留在意大利不仅是出于文化熟悉度，更是为了用行动证明在硅谷之外也能建立世界一流、具备百亿美元乃至更高野心的顶尖科技企业。 --- ### [01:01:16](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=3676s) 从不撒谎、绝不隐瞒：极高诚信背后的道德契约 主持人 Brian Halligan 抛出了最后一个问题：在采访前，他私下致电了三位被 Bending Spoons 收购公司的创始人。令人吃惊的是，三位创始人在完全没有串通的情况下，不约而同给出了完全一致的评价——Luca Ferrari 拥有“异乎寻常的极高诚信度（exceptionally high integrity）”。Brian 感慨，在历访过的大量科技公司 CEO 中，很少能听到被收购方给出如此毫无保留的评价，他好奇这种品质究竟源自何种行事准则。 Luca 听后十分触动，坦言这番评价“让他这一整天都充满了意义”，比夸赞他的外表更让他感到珍贵。他直截了当地指出，这种诚信绝非出于功利主义的商业算计（not a point of expedience），而纯粹是一项极度私人的道德信条。Luca 极为坦诚地揭开了自己年轻时的心理创伤：在二十岁出头时，他曾背叛过当时的女友。那一刻铺天盖地的羞耻感，让他意识到有时候犯下大错比理性论证更能让人看清该如何为人处世。 从那场刻骨铭心的羞耻感中，Luca 给自己定下了一条终身铁律：**不仅绝不撒谎，而且绝不对当事人隐瞒任何其理应知晓的关键信息，哪怕这样做在短期内会对自己造成巨大的伤害**。他将这条原则彻底贯彻到了生活的方方面面与商业交易中。在并购谈判中，这种透明度虽然可能在短期内增加摩擦，但长期来看却积累了无法估量的信任红利——创始人愿意主动找上门来卖公司，因为他们深信“你的承诺字字千金（Your word means something）”。 ```mermaid flowchart TD accTitle: Luca Ferrari 诚信原则的形成与商业复利循环 accDescr: 展示早年错误反思如何转化为不撒谎、不隐瞒的准则，并在长期建立声誉与交易信任。 subgraph Origin["道德觉醒与原则确立"] A["年轻时的重大错误与羞耻感"] --> B["确立核心道德契约：<br/>绝不撒谎、绝不隐瞒关键事实"] end subgraph Practice["日常与商业践行"] B --> C["短期承受代价：<br/>承担冲突与不利后果"] C --> D["长期口碑积累：<br/>言出必行、透明可信"] end subgraph Compounding["商业复利回报"] D --> E["收购交易飞轮：<br/>被收购创始人高度信任，主动寻求交易"] end ``` - **道德动机而非功利手段**：高诚信是个人生命追求，旨在成为一个“如实告知真相、值得他人托付”的人，而非刻意打造的公关人设。 - **不撒谎与不隐瞒的并重**：许多人通过选择性沉默来规避谎言，但 Luca 要求自己把对方应知的所有信息全盘托出，即便会导致即时冲突或个人利益受损。 - **交易中的信任复利**：在残酷的商业并购中，“言出必行”的声誉让 Bending Spoons 获得了不可替代的卖家信任。 --- ### [01:04:02](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=3842s) Brian 的复盘总结：极度求真、陡峭斜率与人才闭环 在采访尾声及后续的个人复盘中，主持人 Brian 对 Luca Ferrari 和 Bending Spoons 的独特运作方式表达了极高的敬意，并系统梳理了这次对话的核心启发： 1. **诚信与直面冲突的勇气**：Brian 将 Luca 的“绝不隐瞒”与 Databricks 联合创始人兼 CEO Ali Ghodsi 的观点联系起来——作为 CEO，最致命的缺陷之一就是回避冲突（conflict averse）。只要坚持不隐瞒任何重要信息，就注定要不断面对人际冲突，而真正卓越的领导者必须具备承受这种张力的定力与勇气。 2. **人才飞轮与量化招聘**：Bending Spoons 将雇主品牌建设与漏斗获客当作严谨的营销增长体系来做。这让 Brian 想起了 HubSpot 早期将营销骨干 Katie Burke（后担任 HR 负责人，现任法律 AI 独角兽 Harvey 的 COO）转去做人力资源的经历——把招聘变成一套自我强化的飞轮机制。更绝的是，Luca 组建专门的数据科学团队，追踪成千上万个指标来量化评估候选人，同时主动大幅降低传统面试的主观权重，因为如今的求职者早已被训练得极其擅长应试与表演。 3. **成长斜率（Slope）压倒一切经验**：即便公司体量已经非常庞大，Luca 依然坚定地选择“成长斜率”而非行业经验。除了几位联合创始人，Bending Spoons 许多核心管理层和被收购企业的负责人年仅 20 多岁。这种全年轻班底在产品驱动增长（PLG）模式下运转顺畅，展现出了强大的爆发力。 4. **绝不降低标准的痛苦代价**：多数公司在规模扩张后都会被迫妥协并降低人才门槛，但 Bending Spoons 始终坚守极度严苛的准入标准。这种坚持虽然导致许多员工不得不承受长期高强度的辛苦与痛苦，却保全了极致的人才密度。 5. **战略、团队与运营的三位一体**：尽管 Luca 在对谈中曾提出“足够好的战略可以容忍平庸的团队”这一烧脑命题，但 Brian 总结认为，商业世界最终并不存在任何捷径——既需要洞察客户与竞争差异的非凡战略，也需要拥有陡峭学习曲线的极高密度团队，同时必须具备 Bending Spoons 那样将软件彻底重构的顶级执行与运营能力。 ## Notable quotes > [00:00](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=0s) Harvard and Stanford barely predict anything. — Luca Ferrari > [00:44](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=44s) Buy to hold and operate forever — Luca Ferrari > [25:34](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=1534s) Head in the clouds, feet on the ground — Luca Ferrari > [39:35](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=2375s) throw them into the deep end of the pool — Luca Ferrari > [49:15](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=2955s) Hell no. This system was built for me, for the way I work. If someone else comes in, they should rewrite the whole thing. — Jensen Huang > [49:15](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=2955s) If you copy what everybody else is doing, you are guaranteed to be approximately mediocre. — Luca Ferrari > [51:27](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=3087s) die of overeating and starvation — Brian Halligan > [55:42](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=3342s) the uncool is cool — Luca Ferrari > [58:54](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=3534s) You can never AB test life — Luca Ferrari > [01:01:16](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=3676s) exceptionally high integrity — Brian Halligan > [01:01:16](https://www.youtube.com/watch?v=hZxmbhsbvyw&t=3676s) Your word means something — Luca Ferrari ## Mentioned Luca Ferrari, Brian Halligan, Bending Spoons, AOL, Vimeo, Evernote, Airtable, HubSpot, Databricks, Ali Ghodsi, Kalshi, Tarek Mansour (TK), NVIDIA, Jensen Huang, PTC, Mitsubishi, Samsung, Berkshire Hathaway, Katie Burke, Harvey, Harvard University, Stanford University, MIT, University of Vermont, University of Wisconsin ## Video description Luca Ferrari is co-founder and CEO of Bending Spoons, a serial acquirer of AOL, Vimeo and Evernote, Airtable, etc, that buys to hold forever, and whose only real edge is that it operates software companies better than almost anyone else. Bending Spoons got 800,000 job applications last year and hired fewer than 300 people. Luca runs hiring the way a quant runs a portfolio — hundreds of signals tested against future performance — and he's blunt about what the data says. Harvard and Stanford barely predict anything. Interviews are the worst signal in the process unless they're scripted. The best candidates were politically active as teenagers and stopped in their 20s. He hires for slope over experience, which is why a 25-year-old routinely shows up to run a company Bending Spoons just bought, and why he thinks experienced hires from big tech lose motivation and struggle to adapt. He also explains why Bending Spoons has no executive team, why the burden of proof always sits with whoever wants to add complexity, and why he stayed in Milan on purpose. I called three founders whose companies Luca acquired before this interview. All three, unprompted, said the same thing: exceptionally high integrity. His answer about where that comes from is the most honest closing I've had on this show. Chapters: 0:00 – Introduction 1:13 – The Bending Spoons model: buy to hold forever, then rebuild from the ground up 8:07 – 800,000 applicants, fewer than 300 hires: industrializing talent density 13:45 – Hiring like a quant trader: hundreds of signals, scripted interviews 21:45 – Why strategy and talent trump org structure 25:34 – Leader or manager? Head in the clouds, feet on the ground, but never both at once 30:38 – Slope over experience: why 25-year-olds run the companies Bending Spoons buys 41:08 – Radical candor, the 360 that found Brian's bugs, and picking one flaw a year 47:00 – No exec team, no sacred org chart: whoever is CEO should rewrite the system 51:27 – Question everything, add nothing: the burden of proof is on whoever adds complexity 57:08 – Why Milan? The uncool is cool; the world needs more than one Silicon Valley 1:01:16 – Never lie, never omit: the high-integrity code 1:04:02 – Brian's takeaways [Source](https://www.youtube.com/watch?v=hZxmbhsbvyw)

[0:00] Luca Ferrari: Harvard and Stanford barely predict anything. [0:44] Luca Ferrari: Buy to hold and operate forever [25:34] Luca Ferrari: Head in the clouds, feet on the ground [39:35] Luca Ferrari: throw them into the deep end of the pool [49:15] Jensen Huang: Hell no. This system was built for me, for the way I work. If someone else comes in, they should rewrite the whole thing. [49:15] Luca Ferrari: If you copy what everybody else is doing, you are guaranteed to be approximately mediocre. [51:27] Brian Halligan: die of overeating and starvation [55:42] Luca Ferrari: the uncool is cool [58:54] Luca Ferrari: You can never AB test life [61:16] Brian Halligan: exceptionally high integrity [61:16] Luca Ferrari: Your word means something

[19:59] Luca Ferrari: It sounded like this person was a genius, and you're telling me it's maybe, you know, ranked eight out of 20. [25:34] Luca Ferrari: William Wallace kind of leader. [42:10] Luca Ferrari: I appreciate their skills and talents and their contribution tremendously. I'm awful at saying that aloud. I'm very critical of myself too... generally at night when I call it a day, generally I'm disappointed in myself. [46:18] Luca Ferrari: whoever the CEO is at that point in time [49:15] Jensen Huang: Oh, no. Hell no. This is kind of built for me specifically. It's the way I work. Somebody else should come in and completely rewrite it. [50:22] Luca Ferrari: I don't know, like being nice to people. I guess that I agree. [51:40] Brian Halligan: die of overeating than starvation. [55:42] Luca Ferrari: Someone who buys a company like AOL cannot possibly care about cool... or maybe I do care, but I think that the uncool is cool. [58:54] Luca Ferrari: make a 1% change by showing you can build a world-beating company, massive ambitions in an unlikely place. [59:47] Luca Ferrari: I was so ashamed of myself that I made a promise to myself to never lie, but also never omit something important that people should know, even if it's to my detriment.

[0:00] I I think people would be surprised by [0:02] how poorly graduating from Harvard or [0:05] Stanford predicts performance. Like [0:08] people think, "Oh, you must be a [0:09] genius." No, actually, yeah, you're [0:11] probably a little bit stronger than, you [0:13] know, people graduating in other [0:14] universities, but but not so much [0:16] stronger. Um and and often we find that [0:20] people tend to lose some of that [0:22] potential raw talent edge by being a bit [0:25] more entitled and thinking too highly of [0:27] themselves. [0:37] >> [music] [0:44] >> Okay, today we have Luca Ferrari on. I [0:47] wanted him to have him on for a long [0:48] time because he runs Bending Spoons, [0:51] which is this fascinating company. It's [0:53] like a little bit of a private equity [0:55] rollup company, but it's really a tech [0:58] platform company and it's unique. I [1:01] can't find any other company that sort [1:03] of rhymes with it. He is very unique in [1:05] the way he thinks and the way he builds [1:06] and the way he CEOs. Hope you like it. [1:09] Okay, Luca, welcome to Long Strange [1:11] Trip. Thanks for coming on. [1:12] >> Thank you for having me. My pleasure. [1:14] Okay, I think by this point most of the [1:17] listeners have heard of bending spoons, [1:20] but I doubt they understand the model [1:24] and what the magic is. So, if you [1:26] wouldn't mind educating us like two [1:28] three minutes on how it works and what's [1:30] special about it. [1:31] >> Yeah, sure. So we I mean the entire [1:33] thesis uh on which the company is built [1:36] is that [snorts] [1:38] if we can build a machine a system [1:40] whereby we can operate digital [1:43] technology businesses better than almost [1:45] anybody else then we are pretty much [1:48] guaranteed to be able to uh compound [1:50] capital efficiently for a very long time [1:52] through acquisitions. And the reason is [1:54] that uh if a lot of businesses are [1:57] basically better off once they're [1:59] integrated into the system what we call [2:00] our platform than being operated as [2:03] standalone companies by someone else [2:04] then we should be able to offer prices [2:06] to people that are appealing and so they [2:09] will sell their businesses to us and we [2:11] still deliver a high return for [2:12] shareholders. So that's the you know the [2:15] the thinking behind it. Of course the [2:17] difficult part is to to build you know [2:19] to build that system that platform so [2:20] that we can operate these businesses [2:22] better than almost anybody else and [2:23] that's what we have uh uh poured our [2:26] hearts and minds for uh into over the [2:28] past uh well 13 years technically. Um [2:32] and so we have tried to have an amazing [2:35] team of people strong culture uh [2:37] powerful proprietary technologies uh [2:39] lots of proprietary data uh and what we [2:42] do we buy these businesses we integrate [2:44] them very deeply into this platform. So [2:46] we have a shared R&D, marketing, GNA [2:49] team that that can fluidly move across [2:51] all of our businesses. We basically [2:54] replatform the acquired businesses so [2:57] that they based on the same [2:58] technological foundation. We some [3:00] investor at some point called it an [3:02] operating system. I think it's a pretty [3:03] apt comparison or metaphor. Um uh and [3:07] then we transform them pretty deeply. So [3:09] we can rebuild the org uh rewrite big [3:11] chunks of the codebase, rearchitect [3:13] cloud infrastructure. uh we try to [3:15] launch a you know useful new features uh [3:18] fix bugs uh improve performance rethink [3:20] monetization and marketing so the you [3:23] know there is a serial acquire component [3:25] uh that you will find in private equity [3:27] you'll find in companies like uh [3:29] Bergkshire Haway or transdime [3:32] the big difference certainly visav [3:34] private equity is that we are not a fund [3:36] we never you know we buy off our balance [3:38] sheet we've never sold a material [3:39] business we don't intend to sell one in [3:42] the future um so we buy to hold and [3:44] operate forever. And the other very big [3:46] difference is definitely visav private [3:48] equity but also certainly you know the [3:50] bers of the world is that we transform [3:53] these businesses pretty deeply and we [3:54] integrate them all together. Um so [3:57] they're much simplified a bit like uh [3:59] they go from being a company uh to being [4:02] a product installed on the same [4:03] foundation basically. So that's the [4:05] model. I've been doing it for for 13 [4:06] years at this point. [4:07] >> Yeah, it's going remarkably well. You [4:09] went public. What's what's roughly the [4:11] market cap? Um, at the time of IPO a few [4:14] months ago, [4:16] uh, $18 billion, I think, equity value. [4:19] >> Congratulations. I'm a huge fan of the [4:21] model. It's really unique. I was [4:23] comparing in my heads at Berkshire [4:24] Hathaway, but it reminds me when I was [4:27] in my 20s, I worked for a US tank [4:29] company called PTC, and they moved me to [4:31] Japan to set up uh the Japanese [4:34] operations when I was very young. And my [4:38] office was inside of Mitsubishi [4:40] headquarters. And so I dealt with [4:42] Mitsubishi headquarters people, but like [4:44] all the different divisions. It feels [4:46] more like Mitsubishi or Samsung than it [4:49] does like some big PE firm or like [4:52] Birkshere. Is that how do you how how [4:54] would you describe it? [4:56] >> Yeah, I don't I don't know those [4:57] companies that well, so I can't say for [4:59] sure. But uh um again the the [5:01] distinctive features are we we buy to [5:03] hold and operate forever. So I think [5:05] that kind of strikes out P firms. Um and [5:09] uh and then again this very deep [5:10] integration. Most companies that acquire [5:13] other companies serially in my [5:15] experience try to run them somewhat [5:17] separately for the most part. At least [5:18] that's what I've seen. We take the [5:21] completely opposite approach where we [5:23] again we we make sure they have the same [5:25] technological foundation broadly [5:27] speaking the same core team that moves [5:29] fluidly across to go after R&D [5:32] opportunities and we draw when uh those [5:34] are exhausted. Um and then we we a lot [5:37] of serial acquirers keep these companies [5:39] you know broadly speaking [5:41] the same way they were before we we [5:44] change them completely from the ground [5:46] up and it's very timeconuming and [5:48] challenging but the the the upside you [5:50] know the bright side of that uh is is we [5:52] generally deliver very high returns so I [5:54] don't know I think uh I don't really [5:57] know of any company that does it this [5:58] way out there but uh [6:00] >> it's kind of an end of one I agree it's [6:01] kind of an end of one is the magic Luca [6:04] in the picking or in the operating and [6:08] what percentage would you give it? [6:10] >> No, I mean it's almost completely in the [6:12] operating. I'd say the platform I [6:14] describe we couldn't do what we what we [6:16] do. So, I I'd like to think we've gotten [6:18] pretty good at, you know, selecting the [6:20] right businesses, but it's it's almost [6:23] it's it's relatively easy to deliver [6:25] high returns if you if you have access [6:27] to our platform because, for example, we [6:30] have we have been enjoying exceptional [6:33] uh an inflow of exceptional talent. We [6:35] got 800,000 applications. Last year, we [6:37] had hired fewer than 300 people. So when [6:40] you have amazing engineers, designers, [6:42] product managers, general managers, it's [6:43] much easier to take a business and [6:45] rethink it from the, you know, from the [6:46] ground up and run it much more [6:48] successfully. Anyone who's run a [6:50] certainly a digital business, probably [6:51] any business, but certainly a digital [6:53] business for a sufficiently long period [6:55] of time will tell you that headcount [6:57] matters [6:59] not a lot, but it's it's just how good [7:01] the people are in the culture you've [7:02] established. And so we when you have [7:04] access to this caliber of people, you [7:05] can run a business with 20 people that [7:07] someone else would need 200 people for. [7:08] And so that alone will give you both [7:10] cost efficiencies and opportunity to [7:12] innovate faster, improve monetization [7:14] faster. Uh so that's an element for [7:16] instance of of our advantage visav these [7:19] companies being run as standalone or [7:21] businesses being run as standalone [7:22] companies. Uh we have invested a lot [7:25] over the past decade in building [7:27] powerful proprietary technologies to do [7:29] everything pretty much I mean across the [7:30] board that you need to to run a a [7:32] digital business. they're natively [7:34] integrated with one another and and they [7:36] just makes everything so much more [7:38] effective and faster. If I were to run [7:40] one of these businesses on a standalone [7:42] basis, I don't know that I would be well [7:44] certainly wouldn't have the resources to [7:45] build this these technologies. Wouldn't [7:46] it wouldn't even know where to start. Uh [7:49] and so, you know, that's also kind of a [7:51] cheat code. So, definitely the the [7:53] operational part is far more important [7:56] uh than uh than the picking. But we try [7:57] to pick well obviously. Why why do it [7:59] poorly if you can do it well? And Luca, [8:01] let's just I want to click on the [8:02] hiring. So I spend most of my time with [8:04] CEOs these days and everyone talks about [8:08] being have high talent density, being a [8:11] talent magnet, and a lot of them pull it [8:14] off early, but as it scales it, it kind [8:17] of inevitably waters down. You seem to [8:20] have almost industrialized this talent [8:23] density idea and you have some very [8:26] unique ways of pulling in applications, [8:29] pulling in interest, the top of the [8:31] funnel of qualifying the interest. Um, [8:35] let's just talk about the top of the [8:37] funnel and talk about them the [8:38] qualification process and talk about it [8:41] like roll the clock back when you were [8:43] nobody's. No one had ever heard of you. [8:44] You had no employer bl brand. You had no [8:47] inbound interest. How did you get the [8:49] talent flywheel kind of rolling? [8:52] >> Yeah, so we we we basically believe that [8:56] if you have a strategy that works or can [8:58] work, then almost all that matters is is [9:01] the talent you you deploy toward the [9:04] execution of that strategy and the [9:05] culture uh meaning like the the rules of [9:07] the game, how people interact with with [9:09] one another, what they choose to do and [9:10] how. Uh that you know that those are the [9:13] areas we've obsessed over uh pretty much [9:16] since the beginning. Now, you're right. [9:18] I mean, you start with nothing. And so, [9:21] that's what founders or early team [9:23] members need to to do well. That's, you [9:26] know, basically punching above their [9:27] weight. Uh, I remember early on spending [9:32] at least 50% of my time, probably more [9:34] for at least the first two, three years, [9:37] um, on on recruiting among other, you [9:39] know, talent related things. And I I and [9:43] a couple others managed to find the the [9:47] lists of of of the best graduates or of [9:50] some in that yeah at the time we hired [9:52] pretty much only in Italy. Now we're [9:54] international but some of the you know [9:56] the the best Italian universities we [9:57] managed to find the lists of graduates [9:59] and we kind of tracked them down and I [10:01] remember calling hundreds of them poorly [10:04] harassing a few [laughter] until they [10:06] accepted [10:07] to join us. Um and later later we built [10:10] a pretty substantial talent team. These [10:13] days we we have you know scientists, [10:16] researchers uh and we treat [10:20] the selection of of of of talent within [10:22] our pipeline uh probably in in similarly [10:27] to how a um a quantitative trader would [10:32] uh picking stocks. So we we we look for [10:34] all signals and then test signals [10:36] against future performance to determine [10:38] whether they're predictive or what [10:40] combinations of signals uh are uh and [10:44] keep iterating. It's a very rigorous and [10:46] sophisticated uh system. At this point [10:48] we we assess hundreds of signals uh in [10:51] each uh in each application. Uh so that [10:54] you know that helps you select [10:56] potentially dramatically better given [10:58] the same level of employer brand and [11:00] ability to attract applications and you [11:02] know half the game is also attracting [11:04] very good applications because if you if [11:05] only geniuses you know apply then it's [11:07] easier to to to have to build a good [11:09] team. Um and so there I think we we have [11:12] a kind of kind of an unfair advantage [11:14] with our model. We found that a lot of [11:16] engineers, designers, product managers [11:18] absolutely love the idea of being able [11:19] to work at a company where they can say [11:23] spend nine months helping rebuild say [11:26] Vimeo's uh video infrastructure and then [11:30] six months rethinking advertising on AOL [11:33] and then maybe a year building a [11:35] platform technology all with the same [11:37] employer broadly speaking with the same [11:38] colleagues with the same culture. the [11:40] track record they're building does not [11:42] get, you know, diluted as you when you [11:44] change employer, obviously a lot of that [11:45] is lost. You have to reestablish [11:47] credibility. Um, and so that's a that's [11:50] a very powerful uh point of attraction. [11:52] Uh, and the other one is because we've [11:54] been so [11:56] uh systematic and uncompromising on [11:58] having extremely high levels of talent [12:00] density pretty much from the beginning, [12:02] of course, compatibly with with our [12:04] appeal to candidates at each point in [12:06] time. [12:08] Once that's in place, uh it's a powerful [12:12] means of retaining excellent performance [12:15] and attracting more of those because you [12:17] know when people find that they love [12:19] their colleagues, they never leave. We [12:21] have essentially zero regrettable churn. [12:23] We had6% [12:26] regrettable churn of spooners, the [12:27] members of our core team last year. Um [12:30] and you know, trust me, they [12:32] [clears throat] get there are poaching [12:34] attempts from big tech all the time. So [12:36] it's not for a lack of opportunities. Um [12:39] and equally these people will tell their [12:41] you know former fellow students or you [12:43] know whatever oh it's amazing here you [12:45] should apply. So it makes your life [12:47] easier but it's very difficult to get to [12:49] a very high level of talent density late [12:50] in a company's history. You need to [12:53] establish it from the beginning and just [12:54] be completely obsessed with it and never [12:57] giving up even when it means hiring way [13:00] fewer people than you'd like to uh for [13:02] prolonged period periods of time. [13:04] parting ways with people who are being [13:05] helpful just because you think you have [13:07] better people. It's quite painful both [13:10] because you know we are nobody's a [13:12] sociopath here and as in most companies [13:14] you don't like to to to to to tell [13:17] someone who's a perfectly good [13:18] contributor that they should look for [13:20] for another job. Um and also you're [13:23] you're losing on on good capacity in the [13:26] short term to look for amazing capacity [13:28] down the line. So those are difficult [13:29] things to do, but if you do them [13:30] consistently, I think you you develop a [13:32] very powerful flywheel. [13:34] >> Sounds like you get kind of a data [13:35] science type team that is looking at all [13:38] the inbound and is looking for signals. [13:40] What are the signals that are most [13:42] strongly correlated with success? The R [13:44] squar's high. And then once you get [13:47] beyond that, I assume there's a human [13:49] interview. What are the questions that [13:50] tease out what you're looking for that [13:52] maybe aren't obvious? [13:54] >> Yeah. So we are fairly fairly um [13:57] secretive about uh you know the details. [14:00] I guess you got a similar answer from my [14:03] quantitative trader again that was my my [14:06] metaphor earlier. But no I can share [14:08] some some of this um there there are [14:10] some obvious signals [14:12] uh in people's applications. You know [14:14] the their GPA for example it's not super [14:17] strong but it's helpful. uh the [14:19] university graduated from. I I think [14:21] people would be surprised by how poorly [14:24] graduating from Harvard or Stanford [14:27] predicts performance. Like people think, [14:30] oh, you must be a genius. No, actually, [14:31] yeah, you're probably a little bit [14:33] stronger than, you know, people [14:34] graduating in other universities, but [14:36] but not so much stronger. Um and and [14:39] often we find that people tend to lose [14:42] some of that potential raw talent edge [14:45] by being a bit more entitled and [14:47] thinking too highly of themselves. Uh [14:49] but it's still predictive. I mean all [14:50] else been equal. Someone from Harvard [14:52] will do a little bit better than someone [14:53] from a less well-known university. So [14:55] some of those are probably somewhat [14:56] obvious. Um there are many that are not. [15:00] Um one I I can share is you know [15:02] relatively predictive. I was surprised [15:04] when this came up. We found that if [15:06] someone is politically active in as a as [15:09] a as a as a teenager [15:11] um but but ceases to be in their early [15:16] 20s that that tends to correlate highly [15:18] with with with someone's drive and and [15:21] and entrepreneurial um [15:25] appro mindset um and and the way we have [15:28] explained it is which of which at least [15:31] of Dennis are highly valued qualities. [15:33] The way we've explained it is is is that [15:35] if you are 17 years old and uh school is [15:39] not enough, you want to um you feel like [15:42] you need to do more. That shows agency, [15:44] that shows productivity, that that shows [15:46] a level of uh not settling for the [15:49] status quo, that's generally healthy. [15:51] But if you still do that when you're 25, [15:52] 26, 30, it means it's probably a life [15:56] mission, which is perfectly fine, but [15:58] likely incompatible with your job, your [16:00] career being, you know, top priority. [16:02] your job is probably probably takes a [16:04] backseat to to your political [16:06] aspirations. [16:06] >> How do you find that signal? Do you ask [16:08] them that question in an interview? Do [16:10] you look through their old Instagram? [16:12] Like how does that signal even arise in [16:14] the process? [16:15] >> Thankfully, people think they need to [16:18] include leadership experiences in their [16:19] CVs because that gets you a job at you [16:22] know consulting company or something. So [16:24] most people we find would include [16:26] something significant in that area if [16:27] they did it. But in any case, we we do [16:30] ask later in the process whether it's an [16:32] interview or written assessments uh [16:35] questions that would very clearly lend [16:37] themselves to to people sharing this [16:38] sort of thing. So we and and with this [16:40] I'll get to the other thing you can do [16:42] which is you shouldn't be passive in [16:44] accepting whatever signals are present [16:46] in someone's application. You can be [16:49] very proactive. We find that a lot of [16:51] the best signals in our case come from [16:53] the the practical tests, the things we [16:55] do or ask candidates to go through. Um [16:59] and so you you can elicit signals uh [17:04] that way. Um interviews are of all the [17:08] typical elements in a selection process, [17:11] at least in our experience, they are the [17:13] worst uh predictors. uh but you can make [17:16] them decent if you uh give interviewers [17:20] extremely precise questions they need to [17:23] ask and you tell them to ask them [17:24] exactly in the same way and then you you [17:27] make sure that the answers are recorded [17:29] and assessed separately. So you what's [17:33] completely useless pretty much is having [17:35] someone talk to someone else with [17:37] essentially no guidelines or just a high [17:39] level agenda and then tell you I like [17:42] the guy, I didn't like the guy. that's [17:44] noise pretty much. Um, so if you [17:46] structure it a lot and you try to [17:48] separate the actual conducting of the [17:50] interview from the uh evaluation of the [17:52] responses, then it can be somewhat [17:55] predictive. [17:56] >> Can I tell you about HubSpot unlocked on [17:59] the sales side [18:01] a good process for doing this where we [18:04] scaled our you know inside sales or from [18:07] 10 to a thousand where lots of we're [18:10] kind of like you. We we had a better uh [18:13] hard it was harder to get into HubSpot [18:14] than Harvard. So we had lots of inbound [18:17] and we would look at all the resumes and [18:19] we would look for people who went to a a [18:21] state school, you know, went to [18:23] University of Vermont, University of [18:25] Wisconsin. We weren't looking for [18:26] Harvard and MIT and we look for person [18:29] who had done a sport um who was athletic [18:32] and competitive. And we didn't think too [18:35] much about it. We didn't interview them [18:37] uh in a standard way. We just said if [18:40] they look kind of like this, let's give [18:41] them a half an hour. And we give them a [18:43] half an hour with a person. And what [18:45] that person would do was set up a role [18:48] play. So I would be the HubSpot [18:49] employee, you'd be the potential. I'd [18:52] set it up and then I'd have you sell to [18:54] me and I give you 12 minutes to sell to [18:56] me. So selling me whatever and I'm [18:57] hearing, I'm giving you objections. We [19:00] finish it. [19:01] Then I would give you feedback, Luca, on [19:03] how you did. I give you a minute to [19:05] think about it. And then we do the exact [19:07] same thing again. And if the person [19:09] absorbed that feedback and did a [19:12] halfdecent job at the uh you know at the [19:15] pitch, we would do a couple blind [19:17] references. Boom, we hired them. That [19:20] actually really ended up scaling for us. [19:22] Uh we didn't do the normal interviews [19:24] that really worked on the sales side [19:26] because I kind of agree with you. The [19:27] other thing I wanted to ask you about, [19:28] but it's just blind references like [19:31] where is that in your signal? Yeah, we [19:34] do that as the very last step. Well, so [19:37] no, we don't do actually blind [19:38] references. We do references. We don't [19:40] do blind references because they're [19:41] illegal in Italy. So, you can't uh ask [19:44] uh and so we figured, okay, let's just [19:45] have a a method that works across [19:47] jurisdictions. Even though I know in the [19:49] US you can call people and ask, oh, how [19:51] was this team member? But, uh we do [19:53] reference is the it's the the candidate [19:56] who presents a few names. uh we try to [19:59] guide it toward okay give me people who [20:02] you know had plenty of exposure. We we [20:05] try to get at least one person for at [20:06] least you know the couple of biggest [20:08] experiences someone had so they and uh [20:10] you know and ideally their manager if [20:12] possible but ultimately it's their call [20:15] and uh and we we we structure those [20:17] interviews with references quite quite a [20:19] lot as well and we find that um they can [20:21] be predictive but again you need to be [20:24] scientific about it. If you leave it to [20:25] someone picking up the phone and calling [20:27] then almost always it's positive. You [20:29] know very rarely do people give you [20:31] names of people who will [20:33] will criticize them and so you need to [20:35] be very specific in your questions and [20:38] ask to make uh hard grading decisions. [20:41] You know, for example, we ask them, [20:43] okay, of the 20 people you have worked [20:45] with the most, say something like this. [20:47] I'm just uh would you say this person is [20:50] is the best or top five? And and then [20:53] you you try to ask a few follow-up [20:55] questions to make sure there's a little [20:56] bit of substance behind that response. [20:58] And and you find that people who are [21:00] enthusiastic, often they will tell you, [21:02] well, it's above average. And you're [21:03] like, you know, hold on a sec. It [21:06] sounded like this person was a genius, [21:07] and you're telling me it's maybe, you [21:09] know, ranked eight out of 20. It's which [21:11] is nice, but you know, um, so you you [21:14] can make it predictive, but you you need [21:16] to put in the effort and the structure [21:17] before you get there. [21:18] >> Okay. I love that advice. Um, I get a [21:21] lot of blind references when people do [21:22] that to me. It's I think it's brilliant [21:24] when they Is that the best person you've [21:26] ever that's ever worked for you? Well, [21:29] no. [laughter] [21:30] Then they say, "Why?" And okay, that [21:32] opens things up. I want to talk about [21:34] you. I You're the 20th CEO I've [21:37] interviewed. It's been really fun. And [21:39] honestly, my thesis when I started this [21:41] podcast was I'm gonna figure out kind of [21:44] the common thread what everyone's doing [21:47] and I'm going to write it down. I'm [21:48] going to kind of standardize it. The [21:51] truth is there's no [laughter] not even [21:54] close to a common thread. Everyone does [21:57] it differently and there is no playbook [21:59] and the playbooks are changing extra [22:01] fast right now. In my last two [22:03] interviews are emblematic of it. I [22:06] interviewed TK from Kshi and he and his [22:10] co-founder have I forget 200 employees. [22:13] There's no levels of management. They [22:14] all report to those two. And he referred [22:17] to it in the podcast as management by [22:19] chaos. My last interview was with Aliot [22:22] from Data Bricks and he's got eight [22:25] direct reports. It's, you know, very [22:28] structured. His direct report team, he [22:31] has three meetings a week with his [22:33] direct reports. He has a QBR uh once a [22:37] quarter and then he has offsite strategy [22:38] session once a quarter. So like polar [22:41] opposites both doing extremely well. [22:44] Where do you fit in that spectrum? [22:47] >> Probably somewhere in between. I I think [22:49] these things are generally [22:51] uh like is more um formal approaches to [22:56] management. I think they are they're [22:57] greatly overestimated uh in the impact. [23:00] I think if you you know in in a [23:04] company's outcome I think almost all of [23:06] the outcome depends on essentially your [23:09] strategy what people sometimes call [23:10] product market fit but also the modes [23:12] you have and you know that that's [23:14] overwhelmingly what drives it I always [23:16] ask myself if to me a business is a good [23:18] business if it would do well even if [23:20] management was mediocre maybe not awful [23:22] but mediocre so that's you know that's [23:24] the first thing and the second thing [23:25] that matters to me is talent levels and [23:27] culture which culture is really like you [23:29] know How do you make decisions? What [23:31] matters? What doesn't? Uh what do we [23:33] value? What do we not value? Then [23:34] everything else, you know, reporting [23:36] lines, uh whether you have OKRs or you [23:39] keep it uh fluid and and informal [23:42] span of control. I'm sure it it does [23:45] matter on the margin, you know, making [23:46] things a bit better, a bit worse. And [23:48] I'm sure different solutions would be [23:49] optimal depending on the people involved [23:51] because uh ultimately a a manager a CEO [23:55] is [23:56] some sort some sort of a leader or [23:58] should be and and and the same strategy [24:01] the same approach to to working with [24:03] others depends on the human capital you [24:05] have different people react different on [24:06] the culture you're trying to to to to [24:08] foster but I think that within reason [24:10] that doesn't make a huge difference [24:12] maybe like a plus minus 10% uh kind of [24:15] impact so I think you're much better of [24:17] obsessing over well first and foremost [24:19] the strategy but then once that's in [24:20] place the quality of the people and the [24:22] culture you establish um so if if you [24:25] know if the 200% organization chaotic [24:29] super high I I I don't know them but [24:31] assuming you know super high talent [24:33] density a culture of uh uh business [24:35] orientation rationality hard work you [24:40] know even if a little extra structure I [24:43] theoretically could be helpful I think [24:45] you know the talent and the culture will [24:46] mostly trump any such gap or close any [24:50] such gap and vice versa you know maybe [24:52] in the other extreme a little bit more [24:54] fluidity and flexibility with ADL but if [24:56] you have very good people and and a [24:58] great culture of let's get it done for [24:59] the benefit of the business people will [25:01] not get stuck because there is a little [25:03] extra structure so extra structure so I [25:06] don't I haven't thought a lot about [25:08] these things I I'm convinced they're not [25:10] the most important thing to be honest in [25:11] in excelling in business [25:13] >> fascinating when I was a CEO go. [25:17] I tried to get better and evolve as the [25:19] business scaled. U and I got a lot of [25:21] feedback and one of the pieces of [25:24] feedback I got a couple times was Brian, [25:26] you're a very good leader and a very bad [25:27] manager. [laughter] [25:29] [clears throat] [25:30] Are you where are you on that? [25:32] >> I don't know. You know, I I I get a lot [25:36] of feedback. Um, I don't think anyone [25:40] uh kind of rated me on those two [25:43] [laughter] [25:43] against those two archetypes or job [25:45] descriptions. [25:46] >> How about a little self-reflection here? [25:50] >> I don't think Yeah, I I'm not sure if [25:53] I'm a a good leader or a good manager. [25:57] uh to be honest um [26:00] I generally feel that I I'm pretty [26:03] privileged that I work with very good [26:04] people and my I tend to be more [26:08] self-conscious and and and question [26:10] whether I'm I'm doing enough and I'm [26:13] living up to expectations. Um so I'm I [26:17] see I tend to see the h the glass half [26:19] empty. So my instinctual reaction would [26:22] be I'm I'm probably not terrible at [26:24] either. I would be disingenuous if I [26:26] told you I think I suck. I but I I I [26:29] would probably rate myself something [26:30] like okay at at both but but not [26:33] necessarily amazing. I probably a better [26:35] leader than manager if I understand what [26:37] you're hinting at. I think I've [26:38] generally done okay at setting a a [26:43] vision that made sense and a compelling [26:45] vision and and communicating that [26:47] reasonably well. [26:49] If I've succeeded as a leader has been [26:50] mostly through what I would call servant [26:52] leadership, like be just working super [26:55] hard and and helping people as much as I [26:56] could. I don't think I'm the kind of [26:59] William Wallace kind of leader, someone [27:00] making a a grand speech and everybody, [27:03] you know, get their hot and and get and [27:05] get their hearts pounding and everybody, [27:07] you know, gets to work. Um, but when you [27:09] work with someone who's walks the talk [27:11] and is in the trenches and does their [27:14] best, I think you tend to be inspired. [27:16] So maybe that kind of person. I try to [27:17] be that kind of person. I don't think [27:19] I'm a great manager. I don't think I'm [27:20] the best at running one-on- ones or [27:22] anything like that. [27:23] >> Okay. Uh, one of the things I used to [27:25] say to myself is actually I got this [27:28] quote from Eleanor Roosevelt of all [27:29] people. Um, [27:32] I wanted to keep my head in the clouds [27:34] but my feet on the ground. And by that I [27:37] meant I I wanted to be okay a JV William [27:41] Wallace painting a picture of the [27:42] future, getting everyone excited about [27:43] where we're going, looking over the [27:45] horizon. But being in the ground, it's [27:46] like, "Holy crap, there are a lot of [27:48] problems and we have to deal with all [27:49] these freaking problems." Uh, and my [27:52] glass was very half empty and I spent a [27:56] lot of times a little negative looking [27:58] at the problems. Where are you in the [28:01] Eleanor Roosevelt quote? [28:02] >> I want to give myself a a little pat on [28:04] the shoulder. I think if there's one [28:06] thing I tend to do reasonably well is [28:10] being able to see the big picture. Uh, [28:12] so you know, I guess head in the cloud [28:14] kind of thing. uh maybe I'm not the best [28:16] at say articulating it again with the [28:18] grand speech but I think I'm pretty good [28:20] at you know in the substance at [28:21] understanding what matters what doesn't [28:23] where the opportunity is what logically [28:25] we should be doing and what's just noise [28:27] uh and I I think I've historically been [28:30] very hands-on in the details in the [28:32] trenches um [28:35] I think it's a it's a good um quality [28:38] that you want to have as an entrepreneur [28:40] and CEO I don't think it's a it's not [28:43] super widespread in the population But [28:45] it's quite commonly found I think in [28:47] entrepreneurs and CEOs and I think it's [28:49] quite powerful because if you if you [28:50] don't have the big picture you can't [28:52] really set you know a directional [28:54] strategy that's that's the winning [28:55] strategy uh unless someone else bring it [28:57] brings it to you which is always dice [28:59] but you you can't count on that [29:00] happening it's you know hope is not a [29:02] strategy as they say um but if you're [29:04] not in the details in the trenches feet [29:07] firmly planted in the on the ground then [29:10] I think you're unlikely to be able to [29:12] inform that big picture properly. you're [29:14] unlikely to be inspiring to your [29:16] colleagues. Very few people like working [29:17] with with people who are in an ivory [29:19] tower. They don't get their hands dirty. [29:21] It doesn't really scale. I think it's [29:23] not good. Um you're you're not in a [29:26] position to assess others, to coach [29:28] others because you don't really know [29:29] anything about the work at some point. [29:31] So I I do believe you need both to be [29:33] successful. Um um and you know finding [29:37] some sort of equilibrium. I find it's [29:38] easier to sequence those activities and [29:40] trying to run them in parallel. So I try [29:42] to have times where I'm very much in the [29:44] clouds thinking okay what are the big [29:46] problems or big big opportunities how [29:48] should you go about them and be more of [29:50] an architect and then long stretches [29:52] where I just get into execution mode and [29:54] I'm you know mud uh head to toe just [29:59] getting stuff done you know blood sweat [30:01] and tears um I think it's quite [30:03] difficult to be both at the same time [30:04] but you can be both in sequence [30:06] >> let's talk about coaching for a second [30:10] similar to me I had ever been CEO of a [30:12] company with thousands of employees in [30:14] public and I was figuring out as I went [30:16] along and sometimes I just didn't feel [30:19] qualified to coach some of my [30:20] executives. Do you ever have that [30:22] feeling? And kind of and also related to [30:24] that so many of the CEOs I'm working [30:26] with like they have a very talented [30:29] person internally. They're on their way [30:30] up. Should they bet on that person or [30:33] find someone who's seen the movie before [30:35] and has a little bit of experience? like [30:37] that's [snorts] happening everywhere [30:39] across Silicon Valley right now. Should [30:41] we hire the bet they were done that [30:42] person or should we should should we [30:43] make the person bet on the person who's [30:45] on their way up and really hasn't seen [30:47] anything close to this movie? [30:49] >> So on coaching I I'm certainly in uh [30:52] inadequate as a coach at often with [30:55] people I work with. Um I think at bend [30:58] influence we work on uh of the [31:01] assumption that our our managers should [31:04] always be capable of solving uh or [31:07] advising on on a solution across all [31:10] problems that are experienced by by the [31:12] direct report. Uh we we just think that [31:14] any one person may be able to coach any [31:16] any other person on some things and they [31:18] should do so regardless of reporting [31:20] lines. Uh I think a good uh a good [31:22] manager uh would generally be someone [31:24] very smart, a good problem solver. So at [31:26] least they can provide kind of [31:28] generalist support. They should be able [31:30] to they should be emotionally supported [31:33] and uh and ready to get their hands [31:34] dirty. But ultimately, you know, being a [31:37] someone's manager doesn't mean you're [31:38] always better positioned to solve a [31:40] problem for them. You know, like it's a [31:41] there are so many skills needed in a [31:43] business like like yours like like like [31:45] ours that obviously you need to divide [31:48] and conquer. In any case, um I I've [31:51] never been a big fan of uh I mean we do [31:53] value experience, but I think we value [31:55] experience a lot less than most people. [31:57] Um I I believe that yes, the being there [32:00] done that helps, but there are a few big [32:04] butts. The first one is the world [32:06] changes pretty fast. So I'm not exactly [32:08] certain that someone who did something [32:09] seven years ago, uh that's still that [32:12] still applies all that that well. uh and [32:15] there is a huge risk that people will be [32:17] blind as to the new context because they [32:19] did it that way and they will just try [32:21] to paste whatever solution they applied [32:23] before. So maybe you know you buy some [32:25] good but you also buy a lot of bad. I'm [32:27] not sure which of the two is stronger [32:28] for a company like Benoons [32:31] also we have such a as far as I can tell [32:34] distinctive culture that we found that [32:37] every time we hired someone almost every [32:38] time we hired someone very experienced [32:40] they struggled to adapt. we're so [32:42] hands-on uh no no politics, radically [32:45] candid uh everybody needs to uh to carry [32:49] their own weight and and it doesn't [32:51] translate well. Often people who have [32:53] been successful at large tech companies, [32:55] they tend to be to have a different [32:56] style to leadership which may very well [32:58] be optimal at those companies. I don't [33:00] know but it's not at at bending. So we [33:02] also need to account for the cost of of [33:04] that cultural change which is sometimes [33:06] doesn't even happen and if it does it [33:08] takes a long time. Um [33:11] and last but not least, I find that uh [33:14] generally people lose motivation as they [33:17] gain experience. And so you need to [33:19] consider whether it's better to have [33:21] someone with 20 years of experience but [33:22] I don't know six out of 10 motivation or [33:25] someone with two years of experience but [33:26] 10 out of 10 motivation. I find often [33:30] the latter is better. It depends on the [33:32] situation. So we're generally we we're [33:34] not big on on experience. We have hired [33:36] experienced people occasionally. We [33:37] prefer to hire. Has that changed as [33:39] you've scaled? Have you changed your [33:40] mind at all on that or you're like [33:42] actually no? [33:43] >> Yeah. Interesting. [33:44] >> No. And I may be wrong. Look, uh we have [33:46] tried. It's not that we never hired [33:48] experienced people. We have hired some [33:49] experienced people with great success, [33:51] you know. It's not that we we were [33:53] completely close-minded uh regarding [33:55] that. [33:56] Um, no, we we keep investing in uh [33:59] students or new graduates or people with [34:01] a couple years of experience where we [34:02] think that we can really help them mold [34:05] their their skill set and and [34:07] professional culture in the way that we [34:09] believe will best uh uh serve them and [34:12] the company in the long run. Um, and and [34:16] also there is a trade-off between talent [34:17] and experience in a competitive labor [34:20] market and and you know experience we [34:22] can provide it to you. We just need to [34:23] be a little bit patient. uh talent we [34:26] can't and so we we we focus more on on [34:28] talent than experience. So all [34:30] considered we're happy with this. We we [34:32] wouldn't change it necessarily. Uh but [34:34] we we have encountered [34:36] difficult situations sometimes where [34:38] maybe we have you know a 25year-old [34:41] lead engineer or or something showing up [34:43] at an acquired company and and [34:45] essentially being the person in charge [34:47] on on our end and and people with 20 [34:50] years of experience thinking does this [34:52] guy even know what what he's doing? and [34:53] and I think our track record [34:55] consistently shows yeah it does or she [34:59] does but uh I can see how it's a more [35:03] difficult sale than if we had a [35:05] 50-year-old person with 30 years of [35:07] experience come in and say the same [35:08] things. It's much easier to assume that [35:10] person is competent. You know I' [35:12] >> I've spoken to a couple of the founders [35:13] of companies you acquired and one of [35:15] them said the person that replaced them [35:17] was 25 and the other said the person who [35:20] replaced them was 27. [35:22] Um, they didn't seem bothered by it, but [35:24] I thought that was kind of interesting. [35:26] What I'm curious about is like, let's [35:28] just say that 25-year-old that takes [35:29] over, I don't know, mirror or whatever [35:32] company you're buying. [35:34] What are you looking for in that person [35:36] who's going to run that show? Is there a [35:39] certain quality you're looking for? I [35:41] know you rotate them around, but let's [35:43] say it's the first first go at it. What [35:45] are you looking for? [35:46] >> Yeah. and and we we wouldn't put someone [35:48] in that position who we haven't tested [35:50] in a smaller version of it to be clear. [35:52] So it's generally someone who led a team [35:55] or a smaller business before. Uh but yes [35:58] it could be someone with just three four [36:01] years of of experience very very uh [36:03] junior. Um well on a high level we we [36:07] look for smart and what we call extreme [36:10] ownership. Being smart for us, I mean [36:13] pretty obviously is you learn very [36:15] quickly and we have evidence of that and [36:17] you're capable consistently capable of [36:19] high quality reasoning. So when you [36:22] propose a solution or you you diagnose a [36:24] problem consistently what you say or [36:27] write makes a lot of sense. There's many [36:29] of all the all the answers but your [36:31] analysis always rational, accurate, [36:34] logically sound. Um, we think that works [36:39] across the board and and in almost any [36:41] position. Two, uh, we want you to have [36:44] extra extreme ownership, which means you [36:46] are absolutely [36:48] you care tremendously about being or [36:51] becoming the best in the world at your [36:52] job, bringing the the greatest possible [36:55] contribution to your team and the [36:56] company. It's you're almost fanatically [36:58] about that. So when when you find an [37:00] obstacle or you you you identify a [37:02] shortcoming in your skill set, you'll be [37:04] hellbent on on overcoming that. But then [37:06] of course we look for more specific [37:08] skills. But that's really like we find [37:10] that that that's 80% of the of when when [37:13] those are in place in big ways, those [37:15] qualities that you're very very likely [37:17] to succeed assuming you're you know [37:21] you can communicate a little bit, you're [37:22] you're a nice you're not an [37:24] those you know hygienic factors. But the [37:26] the core killer features are those two. [37:29] >> And when it doesn't work, so you put in [37:30] the 25-year-old to run it and like I [37:33] don't know, there's an immune system [37:34] rejection or they just weren't up to it. [37:36] What is it that that falls down? [37:40] >> Uh I think when uh that that's happened [37:42] very rarely. I can only think of one or [37:44] two cases, but when that's happened, I [37:47] think it's been primarily [37:49] um one of two things. one uh the person [37:53] while very very good um so it could also [37:57] be that we misjudged those qualities but [37:59] so far it hasn't been because any of any [38:01] such uh we haven't seen any such case at [38:03] least for for our general managers what [38:05] what we've seen is either the person we [38:08] we basically we didn't pay enough [38:10] attention at a person's ability to to to [38:14] to lead others call it communication a [38:17] certain level of common sense and [38:19] empathy and so they were incredibly [38:21] smart problem solvers. [38:24] They moved 100 miles an hour, but but [38:27] they failed to notice when someone was a [38:29] little bit off, had an issue. They [38:31] didn't say the right thing at the right [38:32] time, and that basically created a level [38:35] of hostility or at least friction that [38:38] ultimately made them uh untenable as [38:41] leaders. [38:42] The the other case was where uh enough [38:46] people in the org [38:48] for you know all sorts of reasons felt [38:52] more competent [38:54] and although this person was great and [38:56] super nice and empathetic and [38:58] communicative these people couldn't get [39:00] over the fact that oh I'm so much more [39:03] experienced I cannot take even if you [39:06] know like I cannot be led by this person [39:08] even if just formally that's been very [39:10] rare but I can remember once one such [39:12] case where that occurred and look and [39:14] there we could do one of two things. We [39:16] either move or part ways with you know [39:19] these people who reject leadership or we [39:21] change the leader and you know it [39:22] depends on the context which of the two [39:25] makes more sense. [39:25] >> How do you develop these people? Like do [39:27] you actually train them, develop them, [39:29] coach them or it's like they're smart, [39:32] they're going to figure it out, you kind [39:34] of throw them into it and they they're [39:35] they're good. Yeah, we have training [39:37] materials, but broadly I think 90% of of [39:40] the outcome depends on you throw them [39:42] into the deep end of the pool. You know, [39:44] you give them way more responsibility [39:46] that seems reasonable and uh and then [39:49] you surround them with uh uh amazing [39:52] colleagues. We find that we learn [39:53] primarily by working with with amazing [39:56] people. Almost invariably some of the [39:58] people around you will be better than [40:00] you at some things. And so if you are if [40:02] you try to emulate those good traits, [40:04] you'll learn from those. and also you [40:06] know they will suck at some things and [40:07] if you notice the weaknesses you can [40:08] avoid you know falling into the same uh [40:11] trap yourself. Uh so the the the level [40:14] of responsibility and and and the level [40:16] of talent around you are the two things [40:18] that we find drive growth the most. Um [40:21] but yeah I mean a little bit of formal [40:22] training here and there and and active [40:24] coaching certainly adds that 10 20% [40:28] extra for sure. I can tell you the one [40:30] thing that helped me on my journey. We [40:33] did we decided we're going to do a 360 [40:35] degree review for me, which was a good [40:36] idea. And my co-founder volunteered, [40:39] which is a strange thing because he's [40:40] really introverted. And he's like, "How [40:43] do I do this without talking to other [40:45] homo sapiens?" And so he did a net [40:47] promoter survey. First question is, on a [40:50] scale of 1 to 10, how likely you to [40:52] recommend Brian as the CEO of HubSpot? [40:55] And number two is why did you give that [40:57] number? And then sent it out to like 25 [40:59] people, the board, customers, employees, [41:02] and like people wrote freaking novels. I [41:04] mean, they really wrote long novels. And [41:07] he spent some time, correlated it, [41:08] showed me the histogram, very helpful. [41:11] And then he did something clever like [41:13] for each he had feature for each feature [41:16] be like, "Here's a feature." And he [41:18] would pull out direct quotes on what [41:20] people said about the feature. So it [41:21] really hit home to me that, "Okay, [41:23] that's a feature." In the first 10 [41:25] pages, I'm going through the report and [41:26] I'm like feature feature. I got this. I [41:29] am fantastic at the CO. I have this [41:31] absolutely nailed. Page 11 is where the [41:35] bug started. And there [41:37] >> that's painful. [laughter] [41:40] >> And the direct quotes in there, it was [41:42] incredibly [41:44] helpful [41:46] uh to to help me get better. One of my [41:48] ahas, by the way, was some of the [41:50] features were bugs. You know, people [41:52] didn't like that. I was really [41:53] passionate and sometime I get really [41:55] excited about things and so the first [41:57] year I tried to fix all my bugs and then [42:00] second, third, fourth year I'm like I'm [42:01] going to pick one or two of the bugs [42:03] that I'm going to work on. How do you [42:05] personally go about getting feedback and [42:07] getting better and how do you do that [42:09] with your folks? [42:10] >> Yeah. So we we we're big on feedback as [42:13] a bit like you're just described. Um [42:17] uh we have something called radical [42:18] cander internally where basically the [42:20] idea is you should always tell people [42:22] when you see something good or bad uh as [42:24] long as you don't share that because [42:25] you're venting but because you believe [42:27] there's a chance that they could improve [42:29] or the team could benefit or the company [42:30] could benefit you should share. Don't [42:33] worry about uh the accuracy of what [42:35] you're saying the wording whether you [42:36] hurt feelings. Obviously try to be nice [42:38] if you can but that's you know secondary [42:40] to being honest. Um and so it's of [42:42] course we're not perfect. Not everybody [42:44] does it not all the time, but we try to [42:45] have this environment where uh opinions [42:49] and and and and observations flow very [42:52] quickly and and abundantly. So that's a [42:54] very good source of um of learning. Uh [42:58] >> face to face. You you make everyone do [43:00] it face to face. [43:00] >> Yeah, we try. But our recommendation is [43:02] to if you if it happens in person, say [43:05] it immediately. Even in front of [43:06] everybody in a meeting. Don't delay. [43:08] Just say it. Again, respectfully, [43:10] empathetically, but just straight up. uh [43:13] if it's if if it's something that you [43:14] notice through asynchronous [43:16] communications such as on Slack, then [43:18] you can send them a message or whatever, [43:19] like it's really secondary as long as [43:21] you do it and you're fully candid. Um so [43:24] that helps a lot. And then we have uh [43:26] once a year we we have something a [43:29] little bit like what you described where [43:30] we'd give more structured feedback based [43:32] on a bunch of assessors, not 25, but [43:35] more like six to 10 depending on the [43:37] person. Um [43:39] additionally, and and so that applies to [43:41] me as well. Um, I try to also once a [43:44] year identify at least one major point [43:47] of improvement. I I I went through [43:49] exactly the same thing where I I had [43:51] five and then I figured uh halfway [43:53] through the year it's too many. If these [43:55] are challenging to improve, it's it five [43:58] is too many. So, I try to have one per [43:59] year that I that I try to to get a lot [44:02] better at. Um, [44:03] >> what's your one right now, Luka? I [44:05] happen to be someone who I told you [44:09] earlier that [44:11] I don't remember exactly the words I [44:13] used but the the message I was trying to [44:15] convey is probably the greatest [44:17] privilege I feel I have in my life right [44:19] now and I've had it for a while is the [44:21] quality of the people I get to work [44:22] with. I don't mean it and I I I really [44:24] mean it. It's I really believe it to be [44:25] the case. Um and um as I I I appreciate [44:31] their skills and talents and their [44:32] contribution tremendously. I'm awful at [44:36] saying that aloud. I'm very critical of [44:38] myself too. To be clear, I generally at [44:42] night when I call it a day, generally [44:45] I'm disappointed in myself. Okay. So, [44:47] it's not I'm I'm the same way toward [44:50] myself as I am to others. Maybe more [44:52] more more critical of myself, I'd like [44:54] to think. But so, anyway, the point is [44:56] even people I I hold in the highest [45:00] esteem, uh I rarely tell them how [45:02] amazing they are. And so I I decided [45:04] that I need to be better at at telling [45:06] them how much I appreciate them again in [45:08] a fully honest way. You know, it's not [45:09] like this is not being hypocritical, [45:11] just voicing the what I know [45:14] unconsciously, but don't really linger [45:16] on. Uh so I've tried to do that a lot [45:18] more. I think it's been a half a [45:20] success, half a failure, but certainly [45:22] an improvement over past levels. [45:24] >> I had the exact same feedback all pretty [45:27] much every year. And I would and I'm [45:30] like I'm and there there's a there's a [45:31] saying in in some old school saying that [45:34] you should give five pieces of positive [45:36] feedback to every one kind of you know [45:40] correction. I was definitely the [45:42] opposite [laughter] [45:44] point that out. [45:45] >> Let's use all six slots for corrections. [45:48] >> I was pretty close to that. And so I [45:50] worked on that. I got slightly better. [45:51] And there's a point where I would put [45:53] like a 15 minute slot on Friday [45:55] afternoon like give compliments. uh to [45:59] force myself to do it. Uh so I hear you. [46:01] Most CEOs have that same a lot of CEOs [46:04] also get feedback from their teams that [46:06] their trust circle is just small. They [46:09] trust four people and they run [46:12] everything through those four people and [46:13] everyone else feels like they're very [46:15] arms length. Do you get that? [46:18] >> No, actually I haven't gotten this [46:19] feedback. We are we're quite distributed [46:21] in that way. There's no we don't even [46:23] have um we we don't have a an executive [46:26] team or a leadership team. There's no [46:28] clear I mean people could reconstruct it [46:30] and say okay I think whoever leads [46:32] product should be in there and whatever. [46:34] But we we never we did have it many [46:36] years ago. We dropped it because we [46:37] found the the precise [46:40] uh definition of that group didn't help [46:42] at all because it would lead to some [46:44] people being involved in in discussions [46:46] who shouldn't be involved just because [46:48] they're part of the executive team and [46:50] then some people missing in discussions [46:52] because you know we needed to have a [46:54] discussion at the executive team level. [46:55] So now we're just okay you involve [46:57] whoever needs to be involved no matter [46:59] let's say the title or or the standing u [47:02] the executive team we don't need an [47:03] executive team. Yeah. Interesting. So, [47:06] how is it organized? Like if [snorts] [47:08] you do, let's just say you're doing [47:10] strategy offsite where you want to think [47:12] big things and do some naval gazing. [47:14] Who's there? [47:15] >> I would invite people I think can [47:17] contribute the most or take away the [47:19] most. Sometimes invite someone who in [47:21] general I'm and I think this is not just [47:23] me but the company. We we hate meetings [47:26] and we hate big meetings. Um [47:29] but there are cases where inviting [47:31] someone who won't contribute to the [47:32] meeting could still be a good thing if [47:34] you think they can learn something. [47:36] Inviting someone just so they feel good [47:37] about themselves is never an option [47:38] though. Um it has to be useful. Um I [47:42] would try to invite the people who I [47:43] think could could contribute the most. I [47:45] would have to think uh about who those [47:49] people are. Uh some names come to mind [47:51] but also depending on the precise say [47:54] part of the strategy or angle like we [47:56] buy different people. Luca, you it [47:57] doesn't sound like you have an org [47:59] chart. You're more like Kali. There's no [48:01] I report to Luca. [48:03] >> No, we do. We do actually. It's an [48:05] algorithm. So, we we have two it's a [48:08] basically a matrix organization with [48:10] functions and teams. U functions being [48:12] in charge of uh the how things are done, [48:16] you know, the the hard skills. I'm [48:18] oversimplifying and teams being in [48:19] charge of what we do. So, like the [48:21] objectives, what we're trying to [48:22] accomplish to make Ben successful. Uh [48:25] and then every person has a functional [48:26] lead and a team lead. Uh and it all all [48:30] flows to a single node which currently [48:32] would would be me is me. Um but you know [48:35] it's basically whoever the CEO is at [48:37] that point in time. Um so there is a [48:40] pretty specific form of org and we use [48:42] it to automate a lot of stuff but uh um [48:45] we we we remind people that it's only [48:48] uh useful to the extent that it's [48:50] useful. It's in no way they should feel [48:53] limited by it or it's al quite fluid. [48:55] We've had many cases where someone is a [48:56] lead for a while then goes back to being [48:58] an individual contributor. Um yeah, [49:01] >> you said something interesting whoever [49:03] is the CEO at that point in time. [49:07] Um [49:09] I had a chance to chat with Jess Wong a [49:11] year ago and he's got a unique style [49:13] where he's got 60 direct reports and [49:15] similar to you he does gives direct [49:17] feedback positive and negative in [49:19] public. He doesn't do one-on- ones. [49:21] Um, and I asked him like when if and [49:24] when you ever retire, will the person [49:27] who replaces you inherit that and use [49:30] that system? And he said, "Oh, no. Hell [49:32] no. This is kind of built for me [49:34] specifically. It's the way I work. [49:36] Somebody else should come in and [49:37] completely rewrite it." You said, "As [49:40] long as you're CEO," what did you mean [49:42] by that? And do you agree with Jensen? [49:45] >> Yeah, absolutely. I mean I I think look [49:47] as I said earlier it's it's almost like [49:49] if you if you're coaching or or you know [49:51] running a sports team I think you be a [49:54] fool saying well the way to win is this [49:57] and that and that's the only way to win [49:59] or the best way to win no matter the the [50:01] the people you have on the team right so [50:02] the players uh uh and so yes u I mean if [50:06] uh if and when we we change our CEO I [50:09] would uh encourage uh her or him to to [50:12] see whether any change is warranted I [50:14] mean they don't have to change any, you [50:15] know, anything if they like it the way [50:16] it is, but they equally they should feel [50:18] free to to change things if they thought [50:20] that would improve things. [50:22] >> Okay. Just from talking to you, you're a [50:25] very first principles person, it feels [50:28] like kind of a big skepticism of [50:30] conventional wisdom. I was sort of the [50:31] same way back in the day. What is [50:34] conventional wisdom right about that you [50:36] do that maybe you were like, "This is [50:38] wrong. We're going to rethink it, but [50:40] ah, it's on our wheels. We should just [50:41] do what everybody else does." you know, [50:44] we we've tried to build this company [50:45] from the ground up. Uh yeah, questioning [50:47] everything, not because we thought we [50:49] knew better, but because we believe [50:50] that. So, we had this ambition of [50:52] building one of the most successful [50:53] companies of all time, knowing full well [50:55] that we would very likely not succeed. [50:58] But we thought that if that were if that [51:00] was the the how how high the bar was [51:03] going to be, the only chance we had to [51:05] to to achieve that level of ambition was [51:07] to to to do something differently. if [51:10] you if you emulate what everybody else [51:11] is doing, you're guaranteed to be [51:12] approximately mediocre. I mean, a little [51:14] bit better, a little bit worse. Um, and [51:17] so we we never even bothered [51:19] understanding what conventional wisdom [51:21] was. We just try to to for each [51:23] important problem, we just try to find [51:25] the the optimal solution given our [51:27] particular context and and and vision. [51:29] Um, so yeah, I don't I don't know. I'm [51:31] sure there are plenty of things in [51:32] conventional conventional wisdom that [51:34] that makes sense. I don't know like [51:35] being nice to people. I I guess that I [51:37] agree. One thing that's kind of [51:40] interesting that people ask me about all [51:42] the time is AI native companies seem to [51:45] be running a little bit different than [51:46] companies founded before Chat GBT. [51:50] You're kind of famous for simplifying [51:51] things. So I want to hear about that. A [51:54] lot of people are just doing a lot of [51:55] stuff at once. Like in year two they're [51:57] going international. In year three [51:59] they're starting. They're acquiring a [52:01] business. Like things are just moving so [52:03] much faster. Stuff that took HubSpot [52:04] eight years people are doing in the [52:06] first couple years. And in HubSpot, I [52:08] was a big simplification guy. I even got [52:10] myself a no hat. [52:12] >> Oh, that's awesome. [laughter] [52:14] >> I go to a big meeting. And good thing [52:16] about HubSpot, everyone always got great [52:18] ideas. We should do this. We And we [52:20] would I would always we're definitely a [52:22] company that was much more likely to die [52:24] of overeating and starvation. And so I [52:26] bought this for everyone, the no hat. [52:28] Things seem to be changing. I've noticed [52:31] these companies like they're wearing yes [52:33] hats and they're getting a lot done and [52:35] they're pulling it off. Uh what's your [52:37] reaction to that? [52:38] >> It's really a winning strategy to to be [52:41] a radical simplifier. [52:43] >> Yeah. [52:43] >> Uh because complexity means [52:47] uh costs means being slower, being worse [52:50] generally speaking. But there's no [52:54] um but the perfect balance depends on on [52:57] on on your particular context. [52:59] Obviously, if you were to build a [53:02] business uh let's say in in the 1800s [53:05] with no technology, no internet, [53:10] you know, no no probably labor market [53:12] that was way less efficient. Clearly you [53:14] you should uh at the time you should [53:16] have chosen to do a lot less than uh you [53:18] know uh would have been reasonable in [53:20] the 1990s. Uh and so if with modern [53:24] technology and AI and what we've learned [53:25] we I'm not surprised we can do more [53:28] given the same let's say resources than [53:30] we could in the past. But I think as a [53:32] general rule uh leaning less rather than [53:37] more is a very very useful thing even in [53:39] this day and age particularly because I [53:41] think the fact that some of these [53:42] companies are pulling it off is it [53:44] doesn't mean that you know that that [53:46] they've chosen the right path. We don't [53:48] know how well they would have done had [53:49] they done three but rather than than [53:52] four things and dropping the smallest [53:54] least important one. Maybe the the the [53:57] biggest most important one would have [53:58] been 35, you know, 35% better and and [54:02] bigger and maybe it's a net positive. [54:04] So, um you know, I'm I'm we're big on we [54:08] have a value actually or quality uh that [54:10] we call relentless simplification. We we [54:12] we measure ourselves against and we [54:14] demand [54:15] >> you measure yourself. That's [54:16] fascinating. [54:17] >> Yeah. measure I meant like we we we yeah [54:19] we we assess ourselves against it's not [54:21] and frankly it's not very very [54:23] quantifiable but we want people to every [54:26] time so the rule goes like this every [54:28] time someone recommends that we should [54:31] add complexity whether it's a step in a [54:33] process feature uh a team headcount to a [54:36] team you know you name it any element of [54:38] the system the burden of proof is on [54:41] those making that recommendation those [54:43] who uh believe it's a bad idea don't [54:45] need to bother bring in any proof, they [54:48] they're fine. Uh and so if we choose to [54:50] go forward, there needs to be very clear [54:52] proof that it makes sense or it has to [54:54] be revertible and the people who drive [54:57] the initiative need to then determine [54:59] whether it was a clear success and [55:00] otherwise revert it pretty quickly. And [55:03] the other part of the rule is that we [55:04] need to stay alert and and question [55:07] complexity that's been there for a while [55:09] because we we you know we become blind [55:11] to it as human beings. Uh and and anyone [55:14] who who does that should always be [55:16] applauded even if it's a nuisance that [55:18] oh my goodness again we're questioning [55:19] this process or or this or this policy. [55:22] Um but yes if you have better technology [55:25] AI is amazing and uh you know as someone [55:27] who failed with AI startup in 2010. I [55:29] mean I didn't fail 2010. We started in [55:31] 2010 failed in 2013. Clearly I believe [55:34] in in the power of technology um I'm not [55:38] surprised we can do a lot more today [55:40] than we could 20 years ago. you decided [55:42] to base a company in Milan. We were [55:44] based in Boston. [55:46] I felt like there were some big pluses [55:48] and big minuses to being in Boston. A [55:51] lot of people think just you can't [55:52] create a great company out of Silicon [55:54] Valley. That's obviously not me. What is [55:56] your why did you do Milan? Uh what are [55:59] the pluses and minuses of building [56:00] outside the valley? Do you do you [56:02] perennally feel like you're like not a [56:04] cool kid because you're in Milan? [56:06] >> You know, I actually couldn't care less [56:08] about coolness. someone who buys a [56:11] company like AOL cannot possibly care [56:13] about cool, you know, [laughter] [56:17] >> if anything, or maybe I do care, but I [56:20] think that the uncool is cool. You [56:22] choose, but it's one or the other. I'm [56:23] not certain. Um um [56:26] >> what are the pluses and what are the [56:27] minuses? Like my my my thesis is San [56:31] Francisco is the best place in the world [56:32] to start a company, but holy crap is it [56:34] hard to scale there. It's so expensive [56:36] that there's not a lot of loyalty. It's [56:38] just darn hard to scale. What's your [56:40] thesis on it? [56:41] >> Yeah, I mean, look, I I haven't built a [56:43] company in San Francisco, so I can't [56:44] really tell for sure. Uh, it does strike [56:46] me as a place if you have the coolest, [56:50] highest potential, be best resourced [56:53] company on the planet. I think that's [56:55] where we want to build it. You can [56:56] probably absorb the most talent the [56:58] fastest there than anywhere else. If you [57:01] are more of a midling project, uh, and [57:04] by the way, midling by Silicon Valley [57:07] standards is a pretty high bar, then it [57:09] may be harder there than in many other [57:11] locations. In any case, we we chose [57:13] Italy, Milan. Today, we're an [57:15] international company. We're hiring way [57:16] more people outside of Italy than in [57:18] Italy at this point, but you know, [57:19] including now more and more in the [57:20] States. I think we'll probably open [57:22] office there in the next year. But yes, [57:24] we started it in Milan, Italy, uh, on [57:27] purpose. We actually were in in Denmark [57:29] with my co-founders where we had [57:31] completed our engineering studies and [57:33] had launched and then shut down our our [57:36] AI startup. Um we had two main reasons. [57:40] One was um extremely you know commercial [57:44] and selfish. Basically what you [57:46] described, we thought Italy, 60 million [57:48] people, excellent educ education, [57:50] relatively few amazing companies, [57:52] certainly very few if any upandcoming [57:55] tech companies could be a win-win for us [57:57] in the talent. Like we we brought a lot [57:59] of ambition. We thought an idea that [58:00] made a lot of sense, a project that we [58:02] thought was exciting and and so a lot of [58:04] people would would otherwise be uh [58:07] relegated to joining a inferior business [58:12] or or moving abroad, but they had their [58:14] family there. they loved it there could [58:16] actually stay and and have, you know, [58:18] everything both an exciting job and and [58:20] stay with with their families. Um, and [58:22] so we we thought we'd be able to have [58:24] build a better team basically for the [58:26] same dollars or, you know, or, you know, [58:28] a a an equally good team for for [58:31] cheaper, [58:33] which I think has proven to be true. Um, [58:35] and and a significant advantage. The [58:37] other reason was more missionary. We I [58:39] believed and I think my co-founders [58:41] did did too and we certainly believe so [58:44] today that the world in general is a [58:46] better place if [58:48] knowledge, opportunity, wealth are a [58:51] little bit better distributed. Um I [58:53] think it's totally fine and probably [58:54] even a good thing that you have one or [58:56] two hubs like Silicon Valley that are [58:59] one notch above. I think the world [59:01] probably needs that to to achieve [59:02] maximum you know uh prosperity. But I [59:05] think if the gap becomes immense and [59:08] basically everything that matters [59:09] happens in one or two places, it's a [59:11] really big problem including for the [59:13] people who live in those places in the [59:15] long run. And so we we wanted to do our [59:18] part, you know, 1% make a 1% change by [59:20] showing you can build a world beating [59:22] company, massive ambitions in in a in a [59:25] in unlikely place, Italy being one of [59:28] many clearly, but you know, we are [59:30] Italian. It was easy. We understood the [59:32] culture better. Um, so that was it. and [59:35] we don't regret it. I think it was a [59:37] good move. You can never AB test life, [59:38] so I cannot definitively say that we [59:40] would have done less well had we started [59:43] in in Silicon Valley, but I I do suspect [59:45] that this has been a better choice for [59:46] us. [59:47] >> My last question, so I interviewed three [59:51] of the founders, didn't interview, I [59:53] called three of the founders of [59:54] companies you acquired, [59:56] and all three of them said the same [59:59] thing. They said, "You are exceptionally [60:02] high integrity." [60:04] I don't hear that a lot about CEOs [60:06] necessarily and I don't hear that a lot [60:08] in preparation for these calls. Are [60:11] there some principles around that you [60:13] live by and what are they and what can [60:16] we borrow from you? [60:18] >> Oh, I don't know that you know it's it's [60:20] up to you to determine whether it's [60:21] worth borrowing, but um um it's not a a [60:24] point of expedience. It's a it's a moral [60:27] point. So that morality is very [60:29] personal. I um at some point early in my [60:32] adult life, I decided that I never [60:37] so I mean I I can be transparent about [60:39] it. It's not so in my early early 20s I [60:42] I cheated on my girlfriend at the time [60:45] and uh and I was so ashamed. Sometimes I [60:48] think making a big mistake [60:51] is a better way of learning than uh [60:54] rationalizing how you should conduct [60:56] yourself. I was so ashamed of myself [60:59] that I made a promise to myself to never [61:02] lie uh but also never omit something [61:05] important that people should know um [61:07] even if it's to my detriment. And so [61:09] I've tried to extend that to all aspects [61:12] of my life. Uh I I suspect that although [61:15] it's sometimes costly in the short run, [61:18] it tends to pay off in the long run. Uh [61:20] I'm very glad you know I you just made [61:22] my day sharing that feedback. I I didn't [61:24] know you talked to them and uh [61:28] but I'm glad you did and I'm glad they [61:29] said that over uh you know praising my [61:31] looks or something because [laughter] [61:35] >> No, because it really matters. No, jokes [61:37] aside, it really matters to me. It's a [61:38] it's a it's a it's a life mission I [61:40] settled for myself to to be that kind of [61:42] person you can trust is telling you [61:44] things as they are. At least as as as I [61:46] think they are. Um [61:48] and yeah, I think it does pay off in the [61:50] long run when when you're you become [61:52] known as someone who can be trusted. Uh [61:55] it's easier to to have a good life, you [61:57] know, make good friends and uh [61:59] relationships that that last that [62:00] matter. Do you know do business, you [62:02] know, like people will come to you and [62:04] say, "Okay, we're looking to sell. We [62:06] believe you guys. Your word means [62:09] something." And um so but but it's not a [62:11] it's it doesn't come from a place of of [62:13] seeking expediency. It's it's really [62:15] just a moral point for me. And it's very [62:17] subjective in that regard. [62:18] >> I love that. I'm going to leave it [62:19] there. Congratulations on building [62:21] something really special. Good luck to [62:23] you on your journey to a trillion [62:25] dollars, my friend. [laughter] [62:27] >> We'll do our best. Thank you. It's been [62:28] a pleasure. Thank you for having me, [62:29] Brian. [62:30] >> Appreciate it. [62:31] >> Okay. I thought Luca was really [62:33] interesting. His company's doing really [62:35] well and the guy beats to his own [62:36] drummer. Just to stick on the ending [62:39] there. I mean, for CEOs, integrity is [62:42] really an underrated thing. Uh, you are [62:45] what your word is. You've got a great [62:47] reputation probably, but you do one [62:49] stupid thing and you can kind of have [62:50] that crash down. So, I like that. I like [62:52] that he's like, I never lie and I never [62:56] omit important info. And he could have [62:58] just said that and I'd be like, that's a [63:00] bunch of BS, but I had three of the [63:02] companies he acquired kind of say the [63:04] same thing. It reminds me of my last [63:06] interview with Ali Goi from Data Bricks [63:09] where he's like the worst thing you can [63:11] be as a CEO is conflict averse and like [63:14] if you're never going to admit important [63:16] information when talking to someone, [63:18] you're going to have conflict. And so [63:20] that just sort of reinforces that in my [63:22] mind. Other things I really thought were [63:24] unique was like the way he does hiring [63:26] is really unique along a lot of [63:28] different levels. I mean, he's very [63:30] conscious about the top of his funnel [63:33] and how he brings people in. is very [63:34] conscious about his employee brand. This [63:38] does rhyme with HubSpot. We we had a [63:41] woman named Katie Burke who who she was [63:44] a marketer actually that we moved into [63:46] run HR and she was excellent at it and [63:49] she treated it like marketing like how [63:50] do we how do we how do we create a [63:52] flywheel? How do we pull people in and [63:55] create a real employee brand and now [63:56] she's the COO actually at Harvey. Uh, [63:59] but I think that's really important for [64:00] scaling companies. Create a a hiring [64:03] brand. Getting very good at hiring, I [64:05] think, is important. He's got a whole [64:07] darn data science team looking at [64:09] hundreds or thousands of signals, trying [64:11] to find the right candidates, and sounds [64:13] like he does lots of interesting things [64:15] in that process. One of the least [64:17] interesting things it sounds like he [64:19] does is the interview itself. I kind of [64:22] agree with that. Like when I interview, [64:24] everyone's good at being interviewed [64:25] now. And so I discount that a little bit [64:28] relative to other signals. So I thought [64:31] that was I thought that was pretty [64:32] interesting. I mean he's pretty [64:34] obsessive on talent. He you know he said [64:36] it he said it on this pod. I've heard [64:38] him say it other times like you just [64:40] can't drop the bar. And like as [64:41] companies scale they they inevitably do [64:43] drop the bar. And it sounds like he [64:45] really hasn't and he's stuck on that and [64:47] it causes a lot of suffering and a lot [64:49] of people working very long hours. Uh [64:52] but he he sticks with that. Um, and you [64:56] know, he weighed in on the the age-old [64:58] battle that I hear from CEOs all the [64:59] time. Should I hire for experience or [65:01] slope? And he's at a pretty big scale [65:03] and he's still like full-on slope over [65:06] experience. I think his management teams [65:08] probably other than his founders are all [65:10] in their 20s uh, and have high slope and [65:13] it's working for him. I'm not saying [65:15] that's the answer. I think for some [65:17] people experience matters. also buys a [65:20] lot of companies in the in like the PLG [65:22] type sales motion where you know it's [65:24] not necessarily enterprise sales or [65:26] maybe some experience might be useful [65:28] but that's where he comes in on that he [65:30] talks so much about team team but then [65:33] at one point he's like actually you can [65:35] have a mediocre team if your strategy [65:36] [music] [65:37] is really good which I thought hurt my [65:40] head a little bit and then he also [65:42] talked about when he buys a company is [65:44] it more his picking or his operating so [65:46] operating you know he's a really good [65:47] operator [65:48] I think the reality is you have to be [65:50] really good at all of that stuff. You [65:51] need to have a really compelling [65:53] strategy and that's unique relative to [65:55] [music] your competitors and is [65:57] compelling to your potential customers. [65:59] You need to be an amazing operator and [66:02] you need to have an amazing team. [66:04] There's not really any shortcut on any [66:06] of that stuff. Anyway, that's Luca. I [66:08] thought he was really interesting and [66:10] smart and I hope you liked it.