Lenny's Podcast

安德鲁·安布罗西诺在OpenAI负责Codex桌面应用的开发。如今，OpenAI几乎100%的员工——不仅仅是工程师——每周都在使用Codex。作为一名终身创造者，他拥有工程、设计、产品管理以及创业公司的背景，现在他负责将Codex桌面体验打造成他所说的“有史以来最好的桌面应用，没有之一”。 *在这次深度对话中，我们讨论了：* 1. 为什么AI彻底颠覆了产品开发流程 2. 作为一项专业技能，“品味”的真正含义是什么，以及为什么它正成为AI优先工作场所中最有价值的能力 3. 为什么安德鲁认为如果他们在去年11月（而不是今年2月）发布Codex应用，它可能会失败 4. OpenAI产品经理在“人人皆可构建”环境下的“区域防守”工作模式 5. 安德鲁团队如何合并角色，以及为什么完全消除角色概念是一个大错误 6. 安德鲁如何使用Codex来运行自己的工作流程 7. 一个能够协调ChatGPT、Codex以及人们已使用工具的“大本营”愿景 *本集由以下赞助商支持：* WorkOS——通过SSO、SCIM、RBAC等功能，让您的应用达到企业级标准：https://workos.com/lenny Mercury——颠覆性的银行业务，现已推出Command功能：https://mercury.com/ *节目文字稿：* https://www.lennysnewsletter.com/p/openai-codex-lead-on-the-new-shape *Lenny播客所有文字稿存档：* https://www.dropbox.com/scl/fo/yxi4s2w998p1gvtpu4193/AMdNPR8AOw0lMklwtnC0TrQ?rlkey=j06x0nipoti519e0xgm23zsn9&st=ahz0fj11&dl=0 *如何找到安德鲁·安布罗西诺：* • X：https://x.com/ajambrosino • LinkedIn：https://www.linkedin.com/in/ajambrosino • 网站：https://ambrosino.io *如何找到Lenny：* • 新闻通讯：https://www.lennysnewsletter.com • X：https://twitter.com/lennysan • LinkedIn：https://www.linkedin.com/in/lennyrachitsky/ *本期节目涵盖内容：* (00:00) 安德鲁·安布罗西诺介绍 (02:30) AI如何改变产品工作的形态 (06:32) 何时使用文档 vs. 原型 (10:25) “品味”的真正含义 (12:06) 为什么AI在设计方面仍然表现不佳 (16:18) 设计流程真的消亡了吗？ (21:35) Codex团队的设计流程是怎样的 (23:41) 产品职能正在消失吗？ (27:22) 团队结构 (30:12) 个人贡献者 vs. 管理 (31:37) 规划路线图 (35:16) 构建尚未生效的功能 (38:13) 野心问题：当你过于沉迷AGI时 (39:17) 最新前沿：循环与自主开发 (52:05) 安德鲁如何使用Codex自动化整个工作 (46:52) 计算机使用与浏览器自动化的力量 (49:10) 我们会在Codex内运行所有SaaS应用吗？ (52:05) Codex的未来愿景 (57:20) 那位用Codex为Premiere Pro构建扩展的摄像师 (59:30) 失败角落 (1:01:50) 快问快答 (1:07:03) 幕后：我们的制作人如何使用Codex进行编辑 *参考资料：* • Codex：chatgpt.com/codex • 《原始印记：开端如何塑造创意发展的结局》：https://www.gsb.stanford.edu/faculty-research/publications/primal-mark-how-beginning-shapes-end-development-creative-ideas • Linear：https://linear.app • “品味”不仅仅是审美品味：https://x.com/thenanyu/status/2067327619897446721 • Linear打造备受喜爱的B2B产品的秘诀 | 南宇（产品负责人）：https://www.lennysnewsletter.com/p/linears-secret-to-building-beloved-b2b-products-nan-yu • 保罗·格雷厄姆的网站：https://paulgraham.com • 设计流程已死。这是取而代之的东西。| 珍妮·文（Claude设计负责人）：https://www.lennysnewsletter.com/p/the-design-process-is-dead • 案例研究工厂：https://essays.uxdesign.cc/case-study-factory • 为什么人类是AI最大的瓶颈（以及2026年将发生什么）| 亚历山大·恩比里科斯（OpenAI Codex产品负责人）：https://www.lennysnewsletter.com/p/why-humans-are-ais-biggest-bottleneck • OpenClaw：https://openclaw.ai • OpenClaw：构建、训练并与个人AI代理共存的完整指南：https://www.lennysnewsletter.com/p/openclaw-the-complete-guide-to-building • 从怀疑者到真正信徒：OpenClaw如何改变我的生活 | 克莱尔·沃：https://www.lennysnewsletter.com/p/how-openclaw-changed-my-life-claire-vo • 在你睡觉时工作的Codex功能：https://www.lennysnewsletter.com/p/the-codex-feature-that-works-while • AI悖论：更多自动化，更多人类，更多工作 | 丹·希珀：https://www.lennysnewsletter.com/p/the-ai-paradox-dan-shipper • Atlas：https://chatgpt.com/atlas • Anthropic：https://www.anthropic.com *推荐书籍：* • 《咕噜牛》：https://www.amazon.com/Gruffalo-Julia-Donaldson/dp/0803730470 • 《大橙色斑点》：https://www.amazon.com/Big-Orange-Splot-Manus-Pinkwater/dp/0590445103 _由 https://penname.co/ 制作和营销。_ _如需咨询赞助播客事宜，请发送邮件至 podcast@lennyrachitsky.com。_ Lenny可能是所讨论公司的投资者。

托尼·法德尔创造了iPod，联合创造了iPhone，并创立了Nest（以32亿美元出售给谷歌）。他联合拥有超过300项专利，曾是通用魔术传奇团队的一员，并撰写了面向建造者的最重要且鼓舞人心的著作之一《创造》。 *在本次深度对话中，我们探讨了：* 1. 关于iPhone是否应有物理键盘的内部激烈争论 2. 为何基于观点的决策对初代产品至关重要 3. 为何营销与产品本身同等重要，以及iPod险些失败的教训 4. 为何语音最终将成为与AI交互的主要界面 5. 为何对AI的认知投降是当今产品建造者面临的最大风险 *本集由以下赞助商支持：* WorkOS——通过SSO、SCIM、RBAC等功能让您的应用企业就绪：https://workos.com/lenny Vanta——利用AI自动化合规、管理风险并加速信任：https://vanta.com/lenny *节目文字稿：* https://www.lennysnewsletter.com/p/father-of-the-ipod-and-iphone-on *Lenny播客全部文字稿存档：* https://www.dropbox.com/scl/fo/yxi4s2w998p1gvtpu4193/AMdNPR8AOw0lMklwtnC0TrQ?rlkey=j06x0nipoti519e0xgm23zsn9&st=ahz0fj11&dl=0 *联系托尼·法德尔：* • X：https://x.com/tfadell • LinkedIn：https://www.linkedin.com/in/tonyfadell • 网站：https://www.buildc.com *联系Lenny：* • 新闻通讯：https://www.lennysnewsletter.com • X：https://twitter.com/lennysan • LinkedIn：https://www.linkedin.com/in/lennyrachitsky/ *本期内容时间线：* (00:00) 托尼·法德尔介绍 (02:23) 黑莓与iPhone键盘之争 (07:50) 微观管理vs善意谎言：优秀产品真正需要什么 (15:57) Nest恒温器与烟雾报警器故事 (21:22) 如何判断值得建造什么：痛点加新技术 (27:36) 三代法则：为何首次尝试总不成功 (34:20) 完整客户旅程：为何营销定义你的产品 (40:53) 故事的力量与先写新闻稿的方法 (48:37) 产品管理的演变与建造者角色 (50:27) 为何AI生成代码会制造脆弱、不可维护的产品 (58:00) 叙事技巧 (1:05:45) 下一个iPhone (1:13:15) 硬件回归 (1:17:01) 托尼最兴奋的事 (1:21:38) 与托尼共事 (1:25:36) 伦理、道德与产品建造者的责任 (1:32:40) 如何联系托尼与Build Collective *参考资料：* • 黑莓：https://www.netflix.com/title/81725542 • 功能系统：https://x.com/bhalligan/status/2051873396896518558/photo/1 • Nest：https://en.wikipedia.org/wiki/Google_Nest • 人人都是工程师：v0创造一亿建造者的使命 | Guillermo Rauch（Vercel创始人兼CEO，v0与Next.js创造者）：https://www.lennysnewsletter.com/p/everyones-an-engineer-now-guillermo-rauch • Hermann Hauser的LinkedIn：https://www.linkedin.com/in/hermannhauser • Acorn Computers：https://en.wikipedia.org/wiki/Acorn_Computers • 臭鼬工厂项目：https://en.wikipedia.org/wiki/Skunkworks_project • Netscape Navigator：https://en.wikipedia.org/wiki/Netscape_Navigator • 解析亚马逊独特工作方式 | Bill Carr（《逆向工作》作者）：https://www.lennysnewsletter.com/p/unpacking-amazons-unique-ways-of • 通用魔术：https://www.imdb.com/title/tt6849786 • Dario Amodei网站：https://www.darioamodei.com • Flighty：https://flighty.com • Dave Chappelle：https://en.wikipedia.org/wiki/Dave_Chappelle • Humane Inc.：https://en.wikipedia.org/wiki/Humane_Inc • 《她》：https://www.imdb.com/title/tt1798709 • Spike Jonze：https://en.wikipedia.org/wiki/Spike_Jonze • Waymo：https://waymo.com • Snapchat CEO：为何分销已成为最重要护城河 | Evan Spiegel：https://www.lennysnewsletter.com/p/snapchat-ceo-why-distribution-is • Simbe Robotics：https://www.simberobotics.com • Greyparrot：https://www.greyparrot.ai • Grok：https://grok.com • Cerebras：https://www.cerebras.ai • 以斯帖记4:14：https://www.biblegateway.com/verse/en/Esther%204%3A14 • iPod发明者与Nest创始人托尼·法德尔被任命为MAD首位驻场设计师：https://mad.mit.edu/news/ipod-inventor-and-nest-founder-tony-fadell-named-mit-morningside-academy-for-design-s-inaugural-designer-in-residence *推荐书籍：* • 《创造：打造值得打造之物的非正统指南》：https://www.amazon.com/dp/0063046067 • 《逆向工作：来自亚马逊内部的洞察、故事与秘密》：https://www.amazon.com/Working-Backwards-Insights-Stories-Secrets/dp/1250275717 _由 https://penname.co/ 制作与推广。_ _如需赞助播客，请发送邮件至 podcast@lennyrachitsky.com。_ Lenny可能投资了讨论中涉及的公司。

本尼迪克特·埃文斯是一位独立分析师，曾任职于安德森·霍洛维茨基金，担任多年内部“思想家”，追踪最重要的技术趋势。过去六年间，他持续发布深度研究的演讲报告，探讨科技发展方向，近期重点关注人工智能对经济的变革。他的研究成果被创始人、投资者和运营者广泛阅读，帮助他们在喧嚣的领域中理清思路。他最具争议的观点是：人工智能的重要性堪比互联网或移动互联网——但也仅此而已。 *在我们的深度对话中，我们讨论了：* 1. 为什么我们正处于人工智能的“1997年”——早期、令人兴奋，且对下一步发展充满不确定性 2. 价值将如何在人工智能技术栈中实际积累 3. 反人工智能的反弹情绪及其可能走向 4. 人工智能公司中咨询和专业服务领域的意外繁荣 5. 为什么随着软件构建变得更容易，分销正成为终极护城河 6. 关于你的工作，正确的问题不是“人工智能能完成百分之多少？”而是“这是任务还是工作？” 7. 为什么情况可能会好起来——以及你需要如何准备 *本期节目由以下赞助商支持：* WorkOS——通过SSO、SCIM、RBAC等功能，让你的应用达到企业级标准：https://workos.com/lenny Vanta——利用人工智能自动化合规、管理风险并加速信任：https://vanta.com/lenny *节目文字稿：* https://www.lennysnewsletter.com/p/a-rational-conversation-on-where *Lenny播客所有文字稿存档：* https://www.dropbox.com/scl/fo/yxi4s2w998p1gvtpu4193/AMdNPR8AOw0lMklwtnC0TrQ?rlkey=j06x0nipoti519e0xgm23zsn9&st=ahz0fj11&dl=0 *如何找到本尼迪克特·埃文斯：* • LinkedIn：https://www.linkedin.com/in/benedictevans • 新闻通讯：https://www.ben-evans.com/newsletter • 网站：https://www.ben-evans.com *如何找到Lenny：* • 新闻通讯：https://www.lennysnewsletter.com • X平台：https://twitter.com/lennysan • LinkedIn：https://www.linkedin.com/in/lennyrachitsky/ *本期节目涵盖内容：* (00:00) 本尼迪克特·埃文斯介绍 (02:19) 人们对人工智能影响尚未定价的部分 (06:24) 为什么我们处于人工智能的1997年时刻 (09:44) 专业服务和咨询领域的意外繁荣 (17:44) 为什么分销正成为终极护城河 (23:17) 即将到来的工作变革：真实与恐慌 (27:33) 为什么通用人工智能的定义不断变化 (38:11) 价值将如何积累：模型与应用程序 (42:55) 分销之战：谷歌、Meta、苹果与OpenAI (48:12) 反人工智能情绪与反弹 (53:11) 如何在人工智能未来中养育孩子 (58:27) 应该选择或避开哪些工作 (59:20) 关于人工智能没人问的问题 (1:06:25) 如何在未来取得成功 (1:08:43) 人工智能角落 (1:11:43) 快速问答环节 *参考资料：* • 安德森·霍洛维茨基金：https://a16z.com • 人工智能吞噬世界：https://youtu.be/niJpDnNtNp4 • VisiCalc：https://en.wikipedia.org/wiki/VisiCalc • 麦肯锡公司：https://www.mckinsey.com • 贝恩公司：https://www.bain.com • 埃森哲：https://www.accenture.com • 杰文斯悖论：https://en.wikipedia.org/wiki/Jevons_paradox • 本尼迪克特在LinkedIn上关于Excel的帖子：https://www.linkedin.com/posts/benedictevans_younger-people-may-not-believe-this-but-activity-7303217994459938816-PNqu • 人工智能原生初创公司：5款产品、七位数收入、100%人工智能编写代码 | Dan Shipper（Every联合创始人/CEO）：https://www.lennysnewsletter.com/p/inside-every-dan-shipper • Dario Amodei在X平台：https://x.com/DarioAmodei • 马克·安德森：真正的人工智能热潮尚未开始：https://www.lennysnewsletter.com/p/marc-andreessen-the-real-ai-boom • Frame.io：https://frame.io • 食品营销协会：https://en.wikipedia.org/wiki/Food_Marketing_Institute • Llama：https://www.llama.com • Steven Sinofsky在X平台：https://x.com/stevesi • Drake表情包：https://imgflip.com/memegenerator/343699919/Drake-Hotline-Bling-Transparent-Background • 前谷歌CEO在毕业典礼演讲中谈论人工智能时被嘘 | WSJ新闻：https://www.youtube.com/watch?v=tNH43a1EI7s • 乔纳森·斯威夫特名言：https://www.goodreads.com/quotes/9838985-you-cannot-reason-a-person-out-of-a-position-he • 乔治·卡林名言：https://www.brainyquote.com/quotes/george_carlin_391403 • 富士通：https://global.fujitsu • O*NET在线：https://www.onetonline.org • Pete Holmes网站：https://peteholmes.com • 《第七封印》：https://www.imdb.com/title/tt0050976 • 爱立信R310s手机：https://en.wikipedia.org/wiki/Ericsson_R310s • i-mate手机：https://en.wikipedia.org/wiki/I-mate *推荐书籍：* • 《三人同舟》：https://www.amazon.com/Three-Men-Boat-Jerome-K/dp/1512099899 • 《自然的大都市：芝加哥与大西部》：https://www.amazon.com/Natures-Metropolis-Chicago-Great-West/dp/0393308731 _制作与营销由 https://penname.co/ 提供。_ _如需咨询播客赞助事宜，请发送邮件至 podcast@lennyrachitsky.com。_ Lenny可能是所讨论公司的投资者。

Elena Verna是Lovable的增长负责人，Lovable是领先的AI驱动应用构建平台，在不到一年内仅凭100名员工就实现了2亿美元的年经常性收入。在本期播客中，Elena第四次做客，分享了传统增长策略如何被AI公司彻底改写。她解释了为什么Lovable更注重创新而非优化，如何从激活转向构建新功能，以及为什么免费提供产品已成为他们最强大的增长策略。 *我们讨论的内容：* 1. 为什么60%到70%的传统增长策略在AI领域不再适用 2. 为什么每3个月必须重新寻找产品市场契合点 3. 推动Lovable空前增长的具体增长策略 4. 为什么免费提供产品是一种超越付费广告的增长策略 5. “最小可爱产品”作为新标准（而非最小可行产品） 6. 为什么激活现在属于产品团队，而非增长团队 7. 你是否应该加入一家AI初创公司（诚实的权衡） *本期节目由以下赞助商提供：* WorkOS——面向B2B SaaS的现代身份平台，最高100万月活跃用户免费：https://workos.com/lenny Vercel——你的协作AI助手，用于设计、迭代和扩展全栈网络应用：https://vercel.com/lennyspodcast Persona——数字身份验证的全球领导者：https://withpersona.com/lenny *文字记录：* https://www.lennysnewsletter.com/p/the-new-ai-growth-playbook-for-2026-elena-verna *我的最大收获（面向付费订阅者）：* https://www.lennysnewsletter.com/i/181207556/my-biggest-takeaways-from-this-conversation *如何找到Elena Verna：* • X：https://x.com/elenaverna • LinkedIn：https://www.linkedin.com/in/elenaverna • 新闻通讯：https://www.elenaverna.com *如何找到Lenny：* • 新闻通讯：https://www.lennysnewsletter.com • X：https://twitter.com/lennysan • LinkedIn：https://www.linkedin.com/in/lennyrachitsky/ *本期节目涵盖内容：* (00:00) Elena Verna介绍 (05:19) Lovable的规模与增长 (08:55) 对Lovable作为业务的信心 (12:17) Lovable的留存率 (15:02) Lovable独特的增长杠杆 (28:13) 营销在Lovable成功中的作用 (38:09) 推出新功能 (40:59) 招聘与团队动态 (43:17) 氛围编码的价值 (49:46) 社区的重要性 (51:47) 免费提供产品 (56:26) 公司规模扩大三倍 (01:00:23) 产品市场契合挑战 (01:08:50) 加入AI公司的建议 (01:12:00) 工作与生活平衡 (01:15:20) 在Lovable工作的体验 (01:19:45) 科技领域的女性 (01:25:29) 最终想法与快速问答 *参考资料：* • Elena Verna谈B2B增长如何变化、产品驱动增长、产品驱动销售、为什么选择免费增值而非试用、哪些功能应免费等：https://www.lennysnewsletter.com/p/elena-verna-on-why-every-company • 产品驱动销售的终极指南 | Elena Verna：https://www.lennysnewsletter.com/p/the-ultimate-guide-to-product-led • 10种永远无效的增长策略 | Elena Verna (Amplitude, Miro, Dropbox, SurveyMonkey)：https://www.lennysnewsletter.com/p/10-growth-tactics-that-never-work-elena-verna • Lovable：https://lovable.dev • 构建Lovable：60天内15人实现1000万美元ARR | Anton Osika（联合创始人兼CEO）：https://www.lennysnewsletter.com/p/building-lovable-anton-osika • Stripe：https://stripe.com • 最高绩效产品团队的差异化因素 | John Cutler (Amplitude, The Beautiful Mess)：https://www.lennysnewsletter.com/p/what-differentiates-the-highest-performing • 如何在AI时代获胜：每周发布一个功能、拥抱技术债务、无情削减范围、创造竞争对手无法复制的魔法 | Gaurav Misra（Captions CEO兼联合创始人）：https://www.lennysnewsletter.com/p/how-to-win-in-the-ai-era-gaurav-misra • “我听过最蠢的主意”到1亿美元ARR：Gamma的崛起内幕 | Grant Lee（CEO）：https://www.lennysnewsletter.com/p/how-50-people-built-a-profitable-ai-unicorn • Eric Ries在LinkedIn上：https://www.linkedin.com/in/eries • Elena在LinkedIn上关于Lovable Missions的帖子：https://www.linkedin.com/posts/elenaverna_everythingispossible-lovableway-activity-7401627519646474242-hn6e • SheBuilds：https://shebuilds.lovable.app • Shopify + Lovable：https://lovable.dev/shopify • 产品市场契合跑步机：为什么每家AI公司都在原地冲刺：https://www.elenaverna.com/p/the-product-market-fit-treadmill ...更多参考资料请访问：https://www.lennysnewsletter.com/p/the-new-ai-growth-playbook-for-2026-elena-verna 制作与营销由 https://penname.co/ 提供。如需咨询播客赞助事宜，请发送邮件至 podcast@lennyrachitsky.com。 Lenny可能是所讨论公司的投资者。

Maor Shlomo是Base44的创始人，这是一款AI驱动的应用构建工具，他在短短六个月内以白手起家的方式将其做到超过8000万美元，并被Wix收购。作为一名患有严重ADHD的独立创始人，他在产品上线仅三周后就实现了100万美元的ARR，并将用户数增长到超过40万，期间还经历了以色列的两场战争，且从未获得过任何外部融资。 *你将学到：* 1. 让Base44从三个朋友发展到40万用户、且未在营销上花一分钱的增长策略 2. 他如何在三个月内没有写过一行前端代码——以及如何构建代码仓库，让AI更容易为你编写代码 3. 他的AI生产力工具组合，使他能够与资金雄厚的竞争对手抗衡 4. 为什么在AI领域做独立创始人可能是终极优势（以及差点毁掉业务的婚礼故事） 5. 在伊朗战争爆发时签署8000万美元收购协议的故事 6. 如何判断何时出售、何时保持独立（以及为什么Maor在高度盈利的情况下仍选择被收购） 7. 那个反直觉的产品决策：移除一个“有用”的功能后，激活率提升了三倍 8. 在LinkedIn上公开构建如何带来比任何付费渠道更快的增长 *文字记录：* https://www.lennysnewsletter.com/p/the-base44-bootstrapped-startup-success-story-maor-shlomo *我的最大收获（面向付费订阅用户）：* https://www.lennysnewsletter.com/i/167384119/my-biggest-takeaways-from-this-conversation *本节目由以下赞助商支持：* Sauce——将客户痛点转化为产品收入：http://sauce.app/lenny Dscout——从构思到生产，在每个阶段捕捉洞察的UX平台：https://www.dscout.com/ Contentsquare——创造更好的数字体验：https://contentsquare.com/lenny/ *在哪里找到Maor Shlomo：* • X：https://x.com/ms_base44 • LinkedIn：https://www.linkedin.com/in/maor-shlomo-1088b4144/ *在哪里找到Lenny：* • 新闻通讯：https://www.lennysnewsletter.com • X：https://twitter.com/lennysan • LinkedIn：https://www.linkedin.com/in/lennyrachitsky/ *本期节目涵盖：* (00:00) Maor和Base44介绍 (08:16) 起源故事：Base44是如何诞生的 (14:55) 白手起家与独立创业：挑战与洞见 (22:52) 独立创始人的生产力技巧和技术栈 (27:23) 如何开始使用Base44 (28:47) 关于融资的思考 (34:05) AI时代的推广策略 (36:09) 雄心与目标 (40:05) 增长策略：从第一批用户到成千上万 (51:32) 公开构建 (57:42) 独立创始人的旅程 (01:00:23) 社区支持 (01:03:23) 黑客马拉松与合作伙伴关系 (01:06:42) 产品开发中速度的重要性 (01:08:20) 技术栈与基础设施洞见 (01:15:24) 激活经验教训 (01:18:19) 与Wix的收购历程 (01:25:14) 给创始人的最终建议与思考 *参考资料：* • Base44：https://base44.com/ • Retool：https://retool.com/ • Tzofim：https://www.israelscouts.org/ • Y Combinator：https://www.ycombinator.com/ • RescueTime：https://www.rescuetime.com/ • Cursor：https://www.cursor.com/ • Wix：https://www.wix.com/ • Cursor的崛起：工程师们爱不释手的3亿美元ARR AI工具 | Michael Truell（联合创始人兼CEO）：https://www.lennysnewsletter.com/p/the-rise-of-cursor-michael-truell • 打造Lovable：60天内15人实现1000万美元ARR | Anton Osika（CEO兼联合创始人）：https://www.lennysnewsletter.com/p/building-lovable-anton-osika • 走进Bolt：从濒临死亡到5个月内约4000万美元ARR——历史上增长最快的产品之一 | Eric Simons（StackBlitz创始人兼CEO）：https://www.lennysnewsletter.com/p/inside-bolt-eric-simons • 产品背后：Replit | Amjad Masad（联合创始人兼CEO）：https://www.lennysnewsletter.com/p/behind-the-product-replit-amjad-masad • 现在人人都是工程师：v0创造一亿构建者的使命 | Guillermo Rauch（Vercel创始人兼CEO，v0和Next.js的创造者）：https://www.lennysnewsletter.com/p/everyones-an-engineer-now-guillermo-rauch • Snowflake：https://www.snowflake.com • Yoav Orlev在LinkedIn上：https://www.linkedin.com/in/yoav-orlev-4a044b72 • WhatsApp：https://www.whatsapp.com/ • Amazon：https://www.amazon.com/ • Google：https://about.google/ • MongoDB：https://www.mongodb.com/ • Deloitte：https://www.deloitte.com/ • Render：Render.com • Claude 4：https://www.anthropic.com/news/claude-4 • Gemini：https://gemini.google.com/app • Cloudflare：https://www.cloudflare.com/ _制作与营销由 https://penname.co/ 负责。_ _如需咨询播客赞助事宜，请发送邮件至 podcast@lennyrachitsky.com。_ Lenny可能是所讨论公司的投资者。

No template online will get you a reply, except "leave me alone." Rob Snyder, Author, The Power of Pull, on "founder magic" outreach: skip the pitch and just get them to talk with you.

They give you feedback, buy into your vision, and never buy from you. Rob Snyder, Author, The Power of Pull , on why design partners are a negative signal to people who actually need your product.

Linear beat Jira with a better experience. So why can't every startup win that way? Rob Snyder, Author, The Power of Pull, on why better only wins when the buyer is already pulling for it.

AI progress looks like sudden jumps from the outside. From the inside, it's a smooth line that crosses a threshold. Emmanuel Ameisen, Interpretability Researcher at Anthropic, on why 80% accuracy is useless, 99% is useful, and the model that got Claude writing most of his code.

Rob Snyder is the author of The Power of Pull and writes the newsletter The Physics of Startups. He has advised more than 50 startups on growing from zero to $1 million in ARR, drawing on his own experience as a founder who spent two years pushing a product the market didn’t want before accidentally stumbling onto what actually works. In this episode, we walk step by step through his framework for finding product-market fit, from coming up with an idea to closing your first handful of deals. *We discuss:* 1. The PULL framework: Project, Unavoidable, List of options, Limitations 2. Where the best startup ideas come from 3. Tactics for cold outreach and running your first sales call 4. Why Rob advises against design partners 5. Goblin mode: what to do when nothing is working 6. How to apply the PULL framework inside larger companies *Brought to you by:* https://www.lennysnewsletter.com/p/an-86-minute-guide-to-finding-pmf WorkOS—Make your app enterprise-ready, with SSO, SCIM, RBAC, and more: https://workos.com/lenny DX—Engineering intelligence for the AI era: https://getdx.com/?utm_source=lenny&utm_medium=sponsorship&utm_campaign=q42026 *Episode transcript:* https://www.lennysnewsletter.com/p/an-86-minute-guide-to-finding-pmf *Archive of all Lenny's Podcast transcripts:* https://www.dropbox.com/scl/fo/yxi4s2w998p1gvtpu4193/AMdNPR8AOw0lMklwtnC0TrQ?rlkey=j06x0nipoti519e0xgm23zsn9&st=ahz0fj11&dl=0 *Highlights:* https://lennyspodcast.com/mostreplayedmoments *Where to find Rob Snyder:* • LinkedIn: https://www.linkedin.com/in/rsnyder1 • Newsletter: https://thephysicsofstartups.substack.com • Podcast: https://open.spotify.com/show/4Cic56E1UEbU25P3v14Vkq • Website: https://robsnyder.org *Where to find Lenny:* • Newsletter: https://www.lennysnewsletter.com • X: https://twitter.com/lennysan • LinkedIn: https://www.linkedin.com/in/lennyrachitsky/ *In this episode, we cover:* (00:00) Introduction to Rob Snyder (01:13) The PULL framework: what makes customers buy (07:30) The “PULL” acronym (09:44) How Jump found pull with financial advisors (13:53) Why pull is discovered, not created (16:10) Selling to a person, not a company (17:13) Finding startup ideas through firsthand experience (22:34) Testing your pull hypothesis with two slides (26:41) Recognizing real buying intent (27:24) When solving your own problem works (29:02) The Hail Mary question that can unlock a pivot (31:01) Common traps to avoid (32:50) Quick recap (34:46) Landing your first customer conversations (39:18) Outreach benchmarks and diagnosing what isn’t working (41:27) How to run your first sales call (46:56) Successful pitch tactics (49:59) Does your product need to be 10 times better? (54:08) Closing the deal (01:00:41) Using the PULL framework to set pricing (01:02:22) Revising your hypothesis and going into goblin mode (01:06:56) Designing onboarding to make customers stay (01:09:28) What churn can teach you about your ideal customer (01:11:37) Building a repeatable case study factory (01:16:15) Applying the framework inside established companies (01:20:03) Lightning round and final thoughts *Referenced:* • The PULL framework: https://thephysicsofstartups.substack.com/p/the-pull-framework • The ultimate guide to JTBD | Bob Moesta (co-creator of the framework): https://www.lennysnewsletter.com/p/the-ultimate-guide-to-jtbd-bob-moesta • How to find work you love | Bob Moesta (Jobs-to-be-Done co-creator, author of “Job Moves”): https://www.lennysnewsletter.com/p/how-to-find-work-you-love-bob-moesta • Jump: https://jump.ai • Linear’s secret to building beloved B2B products | Nan Yu (Head of Product): https://www.lennysnewsletter.com/p/linears-secret-to-building-beloved-b2b-products-nan-yu • Josh Woodward on LinkedIn: https://www.linkedin.com/in/joshwoodward/ • Clearly AI: https://clearly-ai.com • Emily Choi-Greene on LinkedIn: https://www.linkedin.com/in/emilycg • Spanx: https://spanx.com • Floodgate: https://www.floodgate.com • Snoop Dogg on X: https://x.com/SnoopDogg • Jira: https://www.atlassian.com/software/jira • Linear: https://linear.app • “Sell the alpha, not the feature”: The enterprise sales playbook for $1M to $10M ARR | Jen Abel: https://www.lennysnewsletter.com/p/the-enterprise-sales-playbook-1m-to-10m-arr • How to close $100K+ enterprise deals, step by step | Jen Abel: https://www.lennysnewsletter.com/p/how-to-close-100k-1m-deals-step-by • Teero: https://www.teero.com ...References continued at: https://www.lennysnewsletter.com/p/an-86-minute-guide-to-finding-pmf _Production and marketing by https://penname.co/._ _For inquiries about sponsoring the podcast, email podcast@lennyrachitsky.com._ Lenny may be an investor in the companies discussed.

Anthropic's Emmanuel Ameisen explains what happens when he realized Claude was better at SQL than him.

Is Anthropic's "go slowly" message just marketing? Emmanuel Ameisen, Interpretability Researcher at Anthropic, on why the alignment team's work comes down to one belief: this technology is going to keep improving seriously.

You can ask an AI what it was just thinking. Then you can check its answer against what's actually going on inside. Emmanuel Ameisen, Interpretability Researcher at Anthropic, on the Spanish-to-French experiment that suggests models have two kinds of processing: a higher-order one and a lower-order one.

The model knew the answer was 12. It kept saying 36. Then it started writing in all caps that it was possessed by a demon. Emmanuel Ameisen, Interpretability Researcher at Anthropic, on what they found when they looked inside, and why it's so hard to catch a model thinking badly.

Anthropic's Emmanuel Ameisen explains how experts understand AI models.

The best ideas at Google Labs never came from a design sprint. Josh Woodward, VP, Google Labs, Gemini & AI Studio, on where NotebookLM, Flow, AI Studio and Google Beam actually started: tiny groups of obsessed people who wouldn’t stop.

A PM told him not to launch a Gemini feature. He replied all to say thank you. Josh Woodward, VP, Google Labs, Gemini & AI Studio, on the signs it’s time to kill an idea, and building a team that can say, “I don’t think this is working.”

Forget learning rate. How fast can someone learn something and then walk away from it? Josh Woodward, VP, Google Labs, Gemini & AI Studio, on the two signals he’s hiring for right now: unlearning rate and explosive endurance.

Emmanuel Ameisen is an interpretability researcher at Anthropic, where he looks inside LLMs like Claude to understand how they think. Before joining Anthropic, he spent years in applied ML at Stripe and Insight, and is the author of Building Machine Learning Powered Applications. His research has produced some of the most striking findings in the field: that models form general representations of concepts, plan many tokens ahead without stating those plans explicitly, and may exhibit something structurally similar to the distinction between conscious and unconscious processing in humans. *We discuss:* 1. Why the “next-token predictor” framing undersells how models actually plan and think 2. Why anthropomorphizing AI can actually be the right approach 3. Why models are “grown,” not built 4. What the “J space” reveals about model self-awareness 5. The smooth scaling curve behind what look like sudden capability jumps 6. How interpretability connects to AI alignment and safety *Brought to you by:* WorkOS—Make your app enterprise-ready, with SSO, SCIM, RBAC, and more: https://workos.com/lenny DX—Engineering intelligence for the AI era: https://getdx.com/?utm_source=lenny&utm_medium=sponsorship&utm_campaign=q42026 *Episode transcript:* https://www.lennysnewsletter.com/p/a-mind-expanding-conversation-on *Archive of all Lenny's Podcast transcripts:* https://www.dropbox.com/scl/fo/yxi4s2w998p1gvtpu4193/AMdNPR8AOw0lMklwtnC0TrQ?rlkey=j06x0nipoti519e0xgm23zsn9&st=ahz0fj11&dl=0 *Highlights:* https://lennyspodcast.com/mostreplayedmoments *Where to find Emmanuel Ameisen:* • X: https://x.com/mlpowered • LinkedIn: https://www.linkedin.com/in/ameisen • Website: https://www.mlpowered.com *Where to find Lenny:* • Newsletter: https://www.lennysnewsletter.com • X: https://twitter.com/lennysan • LinkedIn: https://www.linkedin.com/in/lennyrachitsky/ *In this episode, we cover:* (00:00) Introduction to Emmanuel Ameisen (01:30) Why AI models are grown, not built (05:57) Why next-token prediction is harder than it sounds (11:49) How models represent concepts and understand the world (17:06) How LLMs plan ahead (23:10) Could AI be conscious? (24:44) Scaling laws and the power of text (28:29) Why smooth progress creates sudden breakthroughs (32:59) Why watching AI isn’t enough to solve alignment (38:21) The Golden Gate Claude experiment (39:01) Why Anthropic takes AI safety seriously (42:06) When anthropomorphizing AI is useful (48:06) What AI can teach us about intelligence (51:44) Reasons for optimism about AI alignment (55:43) Are we living in a simulation? (59:00) Jev, decision models, and AI’s gut instincts (01:01:37) Two more surprising findings about how models think (01:03:33) How to get into interpretability research (01:04:55) Lightning round and final thoughts *Referenced:* • Chris Olah on X: https://x.com/ch402 • Distill: https://distill.pub • Ilya Sutskever on X: https://x.com/ilyasut • Tracing the thoughts of a large language model: https://www.anthropic.com/research/tracing-thoughts-language-model • A global workspace in language models: https://www.anthropic.com/research/global-workspace • Scaling Laws for Neural Language Models: https://arxiv.org/abs/2001.08361 • Head of Claude Code: What happens after coding is solved | Boris Cherny: https://www.lennysnewsletter.com/p/head-of-claude-code-what-happens • Claude’s Constitution: https://www.anthropic.com/constitution • Golden Gate Claude: https://www.anthropic.com/news/golden-gate-claude • The Rise and Fall of Agent Civilizations: The whole OpenAI/Hugging Face story in plain English: https://www.dwarkesh.com/p/openai-huggingface • The Utility of Interpretability — Emmanuel Ameisen, Anthropic: https://www.latent.space/p/circuit-tracing • Swyx on X: https://x.com/swyx • Jev for beginners: how to use it and what to build: https://www.lennysnewsletter.com/p/jev-for-beginners-how-to-use-it-and • Jev: https://typesafe.ai/blog/introducing-system-one-models-and-jev • What is System 1 and System 2 Thinking?: https://thedecisionlab.com/reference-guide/philosophy/system-1-and-system-2-thinking • Neuronpedia: https://www.neuronpedia.org • Obsession: https://www.imdb.com/title/tt37287335 • Bref 2 on Hulu: https://www.hulu.com/series/bref2-251c6388-18e9-4545-9e5d-d5f3f6bf20d7 • Apple Watch Ultra: https://www.apple.com/apple-watch-ultra-4 *Recommended books:* • A Brief History of Intelligence: Evolution, AI, and the Five Breakthroughs That Made Our Brains: https://www.amazon.com/dp/0063286343 • Stories of Your Life and Others: https://www.amazon.com/Stories-Your-Life-Others-Chiang/dp/1101972122 • Four Thousand Weeks: https://www.amazon.com/Four-Thousand-Weeks-Management-Mortals/dp/1250849357 _Production and marketing by https://penname.co/._ _For inquiries about sponsoring the podcast, email podcast@lennyrachitsky.com._ Lenny may be an investor in the companies discussed.

Early product-market fit isn’t DAU. It’s whether people’s eyes light up. Josh Woodward, VP, Google Labs, Gemini & AI Studio, on falling in love with the problem instead of the product and the one signal he watches for when showing early prototypes.

Google Labs VP Josh Woodward explains why every company now needs to think and operate like a labs team

The majority of actions on the internet will be taken by agents. Tibo Sottiaux, Head of ChatGPT & Codex, on what builders aren't pricing in yet, and what happened to Notion's traffic after they shipped an MCP.

Josh Woodward is the head of Google Labs, the Gemini app, and AI Studio. He has spent over 17 years at Google and oversees its longest-running and largest labs team, whose products include NotebookLM (now Gemini Notebook), Project Genie, and Flow. His job is to find, experiment with, and scale new AI products, and he has built one of the most distinctive product cultures in the industry for generating ideas, running experiments, and, even more importantly, knowing when to quit. *In our in-depth conversation, we discuss:* 1. Why every company now needs to think and operate like a labs team 2. Where the best ideas really come from 3. The “almost possible” framework 4. How to recognize genuine product-market fit early 5. When to kill an idea 6. What skills are rising in value in the AI era 7. Why roles are not actually blurring into one universal builder role 8. How to set up your internal labs team *Brought to you by:* WorkOS—Make your app enterprise-ready, with SSO, SCIM, RBAC, and more: https://workos.com/lenny DX—Engineering intelligence for the AI era: https://getdx.com/?utm_source=lenny&utm_medium=sponsorship&utm_campaign=q42026 *Episode transcript:* https://www.lennysnewsletter.com/p/why-every-company-now-needs-to-think *Archive of all Lenny's Podcast transcripts:* https://www.dropbox.com/scl/fo/yxi4s2w998p1gvtpu4193/AMdNPR8AOw0lMklwtnC0TrQ?rlkey=j06x0nipoti519e0xgm23zsn9&st=ahz0fj11&dl=0 *Highlights:* https://lennyspodcast.com/mostreplayedmoments *Where to find Josh Woodward:* • X: https://x.com/joshwoodward • LinkedIn: https://www.linkedin.com/in/joshwoodward *Where to find Lenny:* • Newsletter: https://www.lennysnewsletter.com • X: https://twitter.com/lennysan • LinkedIn: https://www.linkedin.com/in/lennyrachitsky/ *In this episode, we cover:* (00:00) Introduction (02:53) Why every company now needs to think like a labs team (05:43) Where great ideas come from (07:43) The “almost possible” framework (11:52) Finding product-market fit (16:09) Falling in love with the problem, not the solution (17:57) When to kill an idea (19:35) Overseeing Google Labs and the Gemini app (22:21) Next consumer AI breakthroughs (24:32) Google’s vision for a personal assistant (27:18) What is overhyped and underhyped in AI right now (31:36) Skills rising in value: unlearning rate, explosive endurance, trust (35:45) Why roles are not blurring into one universal builder role (38:59) Explosive endurance rhythms (41:31) How planning has changed: rolling windows and 100-day milestones (43:37) Labs lifecycle stages (47:37) Tips for setting up an internal lab (54:00) The future of products in an AI world (56:26) Google Labs culture, rituals, and closing thoughts *Referenced:* • Google Labs: https://labs.google • Gemini Notebook: https://notebook.google • Google Flow: https://flow.google.com • Google Beam: https://beam.google • Whisk: https://whisk-ai.io • Nano Banana: https://aistudio.google.com/models/nano-banana • Google Stitch: https://stitch.withgoogle.com • Gemini Spark: https://gemini.google/overview/agent/spark • Gemini Personal Intelligence: https://gemini.google/overview/personal-intelligence • Meta Muse: https://muse.ai • AI’s third era: the rise of persistent AI coworkers | Tara Seshan (Product Lead ChatGPT Work): https://www.lennysnewsletter.com/p/ais-third-era-the-rise-of-persistent • Lenny & Friends Summit: https://www.lennyssummit.com • Dan Shipper’s website: https://danshipper.com • Inside OpenAI | Logan Kilpatrick (head of developer relations): https://www.lennysnewsletter.com/p/inside-openai-logan-kilpatrick-head *Recommended book:* • The Master: The Long Run and Beautiful Game of Roger Federer: https://www.amazon.com/Master-Long-Beautiful-Roger-Federer/dp/1538767368 _Production and marketing by https://penname.co/._ _For inquiries about sponsoring the podcast, email podcast@lennyrachitsky.com._ Lenny may be an investor in the companies discussed.

Everyone assumed 17 people reviewed it. He wrote it in 30 minutes and shipped it. Tibo Sottiaux, Head of ChatGPT & Codex, on OpenAI's culture of autonomy and taking production down on his third day.

Want ChatGPT to recommend your plugin? Build a good one. Tibo Sottiaux, Head of ChatGPT & Codex, on how retention and quality decide which plugins get surfaced to users and why it's not an AEO game.

OpenAI's Tibo Sottiaux on what everyone's getting wrong about AI loops.

Typing fast is trending down. Great taste is trending up. Tibo Sottiaux, Head of ChatGPT & Codex, on the skills that matter in hiring now and why OpenAI has more than 120 ex-YC founders.

OpenAI's Tibo Sottiaux explains why he thinks that AI will keep radically transforming how we work.

Five minutes before his DevDay live demo, his AI agent pinged him: production is down. Tibo Sottiaux, Head of ChatGPT & Codex, on his dot, an agent that works 24/7, connecting the demo and the outage on its own and offering to fix it.

Setting up and fiddling with agent loops isn't the future. Tibo Sottiaux, Head of ChatGPT & Codex, on a permanent, active intelligence that's available on any screen and gets out of the way when you don't need it.