章节 1：Google的逆袭：Gemini 3与“创始人特权”

📝 本节摘要：

本节对话聚焦于 Google 最新发布的 Gemini 3 模型。演讲者重申了他长期以来的观点：凭借海量数据优势，Google 终将赶超并在 AI 领域占据主导地位。对话还深入探讨了 Google 联合创始人 Sergey Brin 回归对于打破公司内部官僚主义的关键作用，并提出了“创始人特权”（Founder's Privilege）或“隐形锤子”（Invisible Hammer）的概念——即只有创始人拥有一种特殊的权力，可以无视预算和常规流程，直接推动重大决策的执行。

[原文] [Speaker A]: dude have you seen Gemini 3 Yes Have you been using it Yes I think it's a godsend for marketers

[译文] [Speaker A]: 兄弟，你看到 Gemini 3 了吗？是的。你有在用吗？是的，我觉得这对营销人员来说简直是天赐之物。

[原文] [Speaker B]: Do you look you look excited about I can tell you've actually used it This is not BS So how do you feel excited about it What's so exciting to you

[译文] [Speaker B]: 你看——你看起来很兴奋啊。我能看出来你是真的用过了。这不是在吹牛。所以你觉得哪里兴奋？什么让你这么激动？

[原文] [Speaker A]: Haven't I been saying since the beginning that I believe Google catch up and dominate and not necessarily make chat GPT go away but they will become a dominant player in this space

[译文] [Speaker A]: 我不是从一开始就说，我相信 Google 会追赶上来并占据主导地位吗？不一定是要让 ChatGPT 消失，但他们会成为这个领域的主导玩家。

[原文] [Speaker B]: you have to to your credit since the beginning 3 4 years ago you have said this

[译文] [Speaker B]: 你确实——值得称赞的是——从一开始，三四年前你就这么说了。

[原文] [Speaker A]: because I'm a big believer that data helps you create the best LLM and Google has more data than anybody and anyone else and if you look at the person who created chat GPT they came from Google

[译文] [Speaker A]: 因为我坚信数据能帮你创造最好的 LLM（大型语言模型），而 Google 拥有的数据比任何人都多。而且如果你看看创造 ChatGPT 的人，他们也是来自 Google 的。

[原文] [Speaker A]: and then Sergey Brin decided to come back in and just handle crap himself when they lost their person and eventually what is this bureaucracy this is stupid like have you seen the all-in interview he this like this is so dumb like no we're just going to do

[译文] [Speaker A]: 然后当他们失去人才时，Sergey Brin（谢尔盖·布林）决定回来亲自处理这些烂摊子。最终，搞什么官僚主义，这太蠢了。就像你看了那个 All-In 的访谈吗？他就像是说，这太蠢了，不，我们就直接做。

[原文] [Speaker A]: He's like just and then someone gets in his way He's like s he he goes what's his name Sundar It's like Sundar just deal with this person like deal with it you know but only founders can do that in organization

[译文] [Speaker A]: 他就像是——然后有人挡了他的路。他就像——他去跟——那个人叫什么来着？Sundar（桑达尔）。就像是，“Sundar，把这个人处理掉，搞定这事”，你知道吧。但只有创始人在组织里能这么做。

[原文] [Speaker B]: Founders have uh uh Jason Cohen remember Jason Cohen from um uh WP Engine Yeah So when I did the interview with him he's like smarter bear Yeah And smart bear A smart bear

[译文] [Speaker B]: 创始人有——呃——Jason Cohen，记得 WP Engine 的 Jason Cohen 吗？对。当我采访他时，他说就像“Smarter Bear”。对，Smart Bear。Smart Bear。

[原文] [Speaker B]: So he was like founders have this invisible hammer where they have the only power to actually go around and make these types of moves And when when someone comes back in and does this only the founder can not even sundor can do this

[译文] [Speaker B]: 所以他说，创始人有一把“隐形锤子”（invisible hammer），他们拥有唯一的权力去绕过规则，做出这类举动。当有人回来做这种事时，只有创始人能做，甚至 Sundar 都做不到。

[原文] [Speaker A]: Yeah Because you know what When founders want to do stuff we and I I'm not Sergey Brennan I don't think like him I'm not the crazy mathematician or anywhere near as smart or successful Mh

[译文] [Speaker A]: 是啊。因为你知道吗？当创始人想做事时——我们，虽然我不是 Sergey Brin，我不像他那样思考，我不是疯狂的数学家，也没他那么聪明或成功，嗯。

[原文] [Speaker A]: But generally speaking with founders we don't look at things like budgets and oh this is what's realistic and you know it goes right over with planning and politician like well I just want to do this I don't really give a crap what anyone else thinks I want to do it

[译文] [Speaker A]: 但总体而言，作为创始人，我们不看那些预算啊，或者“哦这是否现实”之类的东西。你知道这完全凌驾于计划和办公室政治之上。就像是“好吧，我就是想做这个，我真的不在乎别人怎么想，我就要做。”

章节 2：为速度让路：甚至可以无视预算

📝 本节摘要：

本节通过一个发生在“黑色星期五”的真实案例，展示了创始人如何在既定预算与业务速度之间做取舍。尽管 2026 年的预算已经“板上钉钉”，但联合创始人 Mike 提议为了加快产品迭代，打破原定的招聘计划提前录用工程师。对话详细拆解了提前招聘带来的额外成本（约 9.6 万美元），但两位创始人最终达成共识：为了不错失产品发展的关键窗口期，必须利用“创始人特权”跳过财务官僚主义，优先保证执行速度。

[原文] [Speaker A]: And the same goes with our company because we're recording this podcast literally on Black Friday Yeah And we have most of our 2026 budgets set in stone

[译文] [Speaker A]: 我们公司也是一样的情况，因为我们录这个播客的时候正好是黑色星期五。是的。而且我们要把 2026 年的大部分预算都定死（set in stone）了。

[原文] [Speaker A]: So I get a text message from Mike uh co-ounder Mike Okay My co-founder Mike right And Mike's like "Yo you know uh I want to hire this developer and get them to start sooner because we already have our hiring planned out for 2026."

[译文] [Speaker A]: 然后我收到了 Mike 的短信，呃，联合创始人 Mike。好的。我的联合创始人 Mike，对吧。Mike 说：“哟，你知道吗，呃，我想雇这个开发人员，让他们早点入职，因为我们已经规划好了 2026 年的招聘计划。”

[原文] [Speaker A]: Okay keep in mind we haven't started the year yet So you know if this is how we start the year it's going to go off Yeah It snowballs and makes it even more worse and it's probably going to irritate people over time

[译文] [Speaker A]: 好吧，你要记住我们（2026）这一年还没开始呢。所以你知道，如果是以这种方式开始新的一年，事情就会走偏。是的。这会像滚雪球一样，让情况变得更糟，而且随着时间的推移可能会惹恼大家。

[原文] [Speaker A]: So the quick question is how quickly will it irritate Not if if more so how So like this is our budget for 2026 And you didn't say that your finance said that Our finance says so they send us the budget per division

[译文] [Speaker A]: 所以快速的问题是，这会多快惹恼大家？不是“会不会”，而是“怎么个恼法”。比如这就是我们将原本 2026 年的预算。而且这话不是你说的，是你们财务说的。我们的财务是这么说的，他们按部门把预算发给我们。

[原文] [Speaker A]: When you hire engineer as you know it's not necessarily that you have a budget for 2026 per engineer It's also when do they start Because if an engineer costs to keep the math simple 120 grand a year that's 10 grand a month

[译文] [Speaker A]: 当你招聘工程师时，如你所知，预算不一定是按每个工程师在 2026 年的总额算的。还要看他们什么时候入职。因为如果一个工程师的成本——为了算数简单点——是一年 12 万美元，那就是一个月 1 万美元。

[原文] [Speaker B]: I know there's other miscellaneous costs payroll costs HR etc Yeah Yeah Yeah

[译文] [Speaker B]: 我知道还有其他杂项成本，工资成本，人力资源等等。是的是的是的。

[原文] [Speaker A]: But let's just say it's 10 grand a month Well if this person was planned to be hired middle of the year you would only have 60 grand worth of expenses not 120 grand

[译文] [Speaker A]: 但我们就假设是一个月 1 万。好吧，如果这个人计划是在年中受雇的，那你只有 6 万美元的支出，而不是 12 万。

[原文] [Speaker A]: So Mike's messaging me the other day and he's like "Yo dude I want to hire this engineer sooner and do this and oh by the way it'll uh cost us 96 grand that's not budgeted."

[译文] [Speaker A]: 所以 Mike 那天给我发信息说：“哟兄弟，我想早点雇这个工程师，把这事办了，哦顺便说一句，这会花费我们 9 万 6千美元，这不在预算里。”

[原文] [Speaker A]: And him and I are looking at him like "Well we have to make all these product changes to really get the product to where we want right We need maybe more engineers." Yeah

[译文] [Speaker A]: 然后我和他看着这事儿就像：“好吧，我们必须做所有这些产品变更，才能真正把产品做到我们想要的样子，对吧？我们可能需要更多工程师。” 是的。

[原文] [Speaker A]: So I like so why are we waiting to do this Well this is the budgeting plan that was created like "Yeah we were both texting like who cares It's our company Yeah let's just hire them How quickly can we get them?"

[译文] [Speaker A]: 所以我就想，那我们为什么要等呢？好吧，虽然这是已经制定好的预算计划，但就像——“是啊，我们俩发短信说，谁在乎呢？这是我们的公司。是的，我们就雇他们吧。我们能多快搞定他们？”

[原文] [Speaker A]: Well it's going to take 30 days to recruit them We already started recruiting Yeah Right And this is for engineers that we're supposed to hire No joke That's like if Yeah Six months out

[译文] [Speaker A]: 嗯，招聘他们需要 30 天。我们已经开始招聘了。是的，对。而且这还是针对我们要雇的那些工程师——不开玩笑——那就像是，是的，要在六个月后才入职的。

[原文] [Speaker A]: He's like "Uh well the ones that we were supposed to hire in beginning of the year we already started but I already started for the four that we're supposed to recruit middle of the year."

[译文] [Speaker A]: 他说：“呃，好吧，那些我们本该在年初雇的人，我们已经开始招了；但我已经开始招那四个本该在年中才招的人了。”

[原文] [Speaker A]: And and I don't know what countries we're recruiting from but he's just like "Yeah might as well just start getting them now." so we can just do all the stuff we want I was like "Okay sounds good."

[译文] [Speaker A]: 而且我不——我不知道我们是从哪些国家招聘的，但他就是觉得：“是啊，不如现在就开始把人弄进来吧。”这样我们就可以做所有我们想做的事了。我就说：“行，听起来不错。”

章节 3：系统减负：砍掉“附加组件”提升40%效率

📝 本节摘要：

本节深入探讨了产品开发中鲜为人知的“隐形官僚主义”。演讲者发现，SaaS 产品中看似灵活的“附加组件”（Add-ons）功能，由于逻辑复杂，竟导致开发速度降低了 40%。为了解决这一问题，他提出了一项反直觉的改革方案：废除所有附加组件，转而在后端创建数十个包含不同功能组合的固定“套餐”（Plans）。这种“以空间换时间”的策略，在不影响前端用户体验的前提下，成功消除了工程端的逻辑负担，大幅释放了开发产能。

[原文] [Speaker A]: Same with other bureaucracy in a business Like the reason our product moves slower check this out So when you have products you have monthly pricing plans right Uh for SAS as you know and then we have these things called add-ons

[译文] [Speaker A]: 这跟公司里的其他官僚主义是一样的。比如我们产品开发变慢的原因，听听这个。当你有产品时，你有月度定价计划（Plans），对吧？呃，你知道对于 SaaS 来说，然后我们还有这些叫做“附加组件”（Add-ons）的东西。

[原文] [Speaker A]: So that way people can just like upgrade for different parts of it Did you know that add-ons on our system because it has so much more logic it slows down development an estimated 40%

[译文] [Speaker A]: 这样人们就可以只针对他们想要的部分进行升级。你知道吗？在我们系统里的附加组件，因为它包含太多的逻辑，导致开发速度估计降低了 40%。

[原文] [Speaker B]: Wow 40%

[译文] [Speaker B]: 哇，40%。

[原文] [Speaker A]: 40% 40% So I'm not an engineer you know this So I just got on the phone with engineer I'm like "So plans are less efficient right?" He's like "Yeah." I was like "All right I got a solution for you." He's like "What Remove all add-ons."

[译文] [Speaker A]: 40%。40%啊。我不是工程师，你知道的。所以我刚跟工程师通了电话，我说：“所以（现在的）计划效率很低对吧？”他说：“是的。”我说：“行，我有个解决方案给你。”他说：“什么？”（我说：）“把所有附加组件都删了。”

[原文] [Speaker A]: He's like "Neil this would be too complex What would we do in instead?" create a plan for every single add-on

[译文] [Speaker A]: 他说：“Neil，这太复杂了，那我们用什么来代替呢？”（我说：）为每一个附加组件创建一个单独的计划（Plan）。

[原文] [Speaker B]: Wait add-ons for Uber Suggest What did what are your talk about like feature add-ons

[译文] [Speaker B]: 等等，你是说 Ubersuggest 的附加组件吗？你在说什么，比如功能附加组件？

[原文] [Speaker A]: Uh feature add-ons or I'll give you a very specific example Let's say you are tracking your rankings and your visibility on chat GPT and you want to do it for Gemini it's add-on

[译文] [Speaker A]: 呃，功能附加组件，或者我给你一个非常具体的例子。假设你正在追踪你在 ChatGPT 上的排名和可见度，然后你也想追踪 Gemini 上的，这就是一个附加组件。

[原文] [Speaker A]: So I say forget the add-ons just keep plans that we have in our back end that aren't visible on our pricing page and you can do a new plan for someone who has Chad GBD plus Gemini and then if they don't want that just drop them to the older plan that just has Chad GV

[译文] [Speaker A]: 所以我说，忘掉附加组件吧，只保留我们在后端有的那些计划——那些在定价页面上不可见的——你可以为那些拥有 ChatGPT 加上 Gemini 的人做一个新计划；如果他们不想要那个，就把他们降级到只有 ChatGPT 的旧计划。

[原文] [Speaker B]: So all cart and different bundles Is that what you're saying

[译文] [Speaker B]: 所以是单点式（a la carte）和不同的捆绑包，你是这个意思吗？

[原文] [Speaker A]: Uh kind of I'm just grouping all the add-ons So let's say if we only had four plans just have 30 plans And then and I broke this down with my engineer He's just like "Yeah if you end up doing that then we wouldn't have and that speeds up capacity your your development rate." Yeah Yeah

[译文] [Speaker A]: 呃，算是吧。我只是把所有的附加组件组合起来。所以假设我们本来只有 4 个计划，现在就搞 30 个计划。然后我跟我的工程师拆解了这个想法，他就像是：“是啊，如果你最后这么做，我们就不用（处理那些逻辑），这会提升产能，也就是你的开发速度。” 是的，是的。

[原文] [Speaker A]: And I was like "All right sounds good to me." And he's like "Yeah that works." He's like "You okay with 30 plans?" I was like "Yeah 30 plans How's that going to look?" Not visible on our back end so wouldn't slow things down Got it Got it Okay that makes sense

[译文] [Speaker A]: 然后我就说：“行，听起来不错。”他说：“是啊，这行得通。”他问：“你能接受 30 个计划吗？”我说：“行啊，30 个计划。那看起来会怎样？”（只要）在后端不可见（或者不影响前端展示），就不会拖慢速度。明白了，明白了。好的，这有道理。

章节 4：实战演练：用Gemini构建自动化人脉网络

📝 本节摘要：

谈话转向 Gemini 3 的实际应用。Speaker B 分享了他使用 Gemini AI Studio 进行“氛围编码”（Vibe Coding）的经历：他只需上传一张他觉得很酷的产品截图，Gemini 就瞬间生成了与其一模一样的前端代码。更进一步，他正在构建一个名为“Nexus”的自动化项目——一个类似 Juicebox 的智能体系统。该系统能自动寻找潜在的合作伙伴和销售线索，获取真实邮箱，并生成可视化的人脉网络图谱。他认为，虽然 Replit 和 Lovable 很不错，但 Gemini 展现出的能力已处于另一个层级。

[原文] [Speaker B]: Um but all that to say so going back to Gemini 3 is a godsend for marketers In in what ways are are we using it right now Because I've seen so many examples I'll give you an example

[译文] [Speaker B]: 呃，总而言之，回到“Gemini 3 是营销人员的天赐之物”这个话题。我们现在到底在怎么用它？因为我看过太多例子了，我给你举个例子。

[原文] [Speaker B]: So we've talked about Replet We've talked about uh lovable We've talked about these vibe coding tools right And uh Google Gemini the the Gemini AI studio The things you can build now is absolutely insane

[译文] [Speaker B]: 我们聊过 Replit，我们聊过呃 Lovable，我们聊过这些“氛围编码”（vibe coding）工具，对吧？而 Google Gemini，那个 Gemini AI Studio，你现在能用它构建的东西绝对是疯狂的。

[原文] [Speaker B]: Like I don't know if you've tried this but literally I designed so I I built the front end for a product right And then I designed it I literally went to a website I dragged I took a picture of a product that I thought was cool I said "Make it look like this." Boom It makes it look like that Right

[译文] [Speaker B]: 我不知道你试过没有，但我真的设计了——所以我构建了一个产品的前端，对吧。然后我设计它时，我真的只是去了一个网站，我拖拽——我拍了一张我觉得很酷的产品照片，我说：“把它做成这样。” Boom。它就做成那样了。对吧。

[原文] [Speaker B]: And then it's actually finding uh because I keep talking about juice box recruiting tool I was like I wonder if I can make a juice box for like sales and for like partnerships and like conferences and all that right

[译文] [Speaker B]: 然后它实际上在寻找——呃，因为我一直在这个 Juicebox 招聘工具。我就想，我想知道我能不能做一个用于销售、用于合作伙伴关系、用于会议之类的 Juicebox，诸如此类，对吧。

[原文] [Speaker A]: And it what do you mean by that I'm a little confused Juice box does outreach for recruiting I already got that

[译文] [Speaker A]: 那是什么——你这是什么意思？我有点困惑。Juicebox 做招聘外联，我已经知道了。

[原文] [Speaker B]: So what uh what this product is called is Nexus Like Gemini actually named this product Nexus right

[译文] [Speaker B]: 所以——呃，这个产品叫 Nexus。就像 Gemini 实际上把这个产品命名为 Nexus，对吧。

[原文] [Speaker B]: So it will basically look for people that we should reach out to partner with that we should do content collaborations with that we should do sales with and it'll make like a a web and like it shows me who I connect with over time and the web will just continue to expand and it will do it automatically like Juicebox so I can make agents doing it

[译文] [Speaker B]: 所以它基本上会寻找我们应该去联系合作的人，我们应该做内容协作的人，我们应该做销售的人。它会制作一个像网络一样的东西，展示我随着时间推移与谁建立了联系，并且这个网络会继续扩展。它会自动完成这些，就像 Juicebox 一样，所以我可以让智能体（Agents）来做这件事。

[原文] [Speaker B]: And so I'm just working on that right now and like it's finding real emails It's finding the right people and I'm like dude if I just keep playing with this for like a couple hours a week or whatever and I do it over a couple months like I can actually build something real right

[译文] [Speaker B]: 所以我现在正在做这个，而且它正在找到真实的电子邮件。它正在找到合适的人。我就想，兄弟，如果我每周只花几个小时摆弄这个，或者不管多少时间，坚持做上几个月，我就真的能构建出一些真实的东西，对吧。

[原文] [Speaker B]: Imagine how this is going to be in a year But like this is the first time like sure replet's kind of cool you know um Lovable is cool but like it hasn't it's not at this level dude

[译文] [Speaker B]: 想象一下一年后这会变成什么样。但这就像是第一次——当然 Replit 挺酷的，你知道，呃，Lovable 也挺酷的，但就像它还没——还没达到这个水平，兄弟。

章节 5：市场洞察：算力护城河与AI泡沫预警

📝 本节摘要：

本节对话从产品工具转向宏观市场竞争。演讲者预测，尽管像 Lovable 这样的公司增长迅速，但在长跑中，拥有底层模型和庞大生态系统的巨头（Google、OpenAI、Anthropic）将“碾压”这些应用层公司，因为巨头们并不依赖单一工具的 SaaS 收入。对话还提到了 Sam Altman 对 Google 追赶速度的担忧，以及 Google 拥有自家芯片（TPU）从而减少对 Nvidia 依赖的硬件优势。最后，他们发出了严厉的预警：目前 90% 到 95% 的 AI 初创公司都是泡沫，许多创始人融资后便不再与投资人沟通，缺乏实质性的商业构建。

[原文] [Speaker B]: I think I I know everyone's like lovable is a big company 200 million They just broke 200 million Yeah In ARR and they they crossed the extra 100 million in like 4 months or something like that Really fast Yeah Really fast

[译文] [Speaker B]: 我想——我知道大家都觉得 Lovable 是家大公司，2 亿美元。他们刚刚突破了 2 亿美元。是的，ARR（年度经常性收入）。而且他们大概只用了 4 个月左右就完成了额外的 1 亿美元增长。真的很快。是的，真的很快。

[原文] [Speaker A]: I believe Gemini and Chad GPT and Anthropic are going to end up in the long run crushing the lovables Not necessarily meaning Lovable is going to be worth zero dollars They're going to take over the market share for that dude

[译文] [Speaker A]: 我相信从长远来看，Gemini、ChatGPT 和 Anthropic 最终会碾压 Lovable 这类公司。不一定意味着 Lovable 会变得一文不值，但他们（巨头）会夺走那部分市场份额，兄弟。

[原文] [Speaker A]: Because at least for right now AI Studio Gemini is not charging me for it Or maybe I'm paying for the pro version right But they're not going to like they they don't care about that so much because they just want to have people using their products and they want them in their ecosystem That's how they make their money Yes It's not from the SAS revenue

[译文] [Speaker A]: 因为至少现在，AI Studio Gemini 没有收我费。或者可能我付的是专业版的钱，对吧。但他们不会——他们不太在乎那个，因为他们只想让人们使用他们的产品，想把人们留在他们的生态系统里。那是他们赚钱的方式。是的。不是靠这些 SaaS 收入。

[原文] [Speaker B]: I think so too And I think this is where um I read something on the information last week cuz I still have a subscription to the information and it's like Sam Alman is like basically the article is like how Sam Alman's like oh crap this like oh crap moment and like internally he's like trying to push the team now cuz they need to be pushed because this Gemini situation like they are starting to get lapped

[译文] [Speaker B]: 我也这么认为。我觉得这就是——呃，上周我在《The Information》上读到了一些东西，因为我还订阅着它。文章基本上是说 Sam Altman（萨姆·奥特曼）就像是“噢糟糕”，这就像是一个“噢糟糕”的时刻。他在内部试图推动团队，因为他们需要被推动，因为 Gemini 的情况就像——他们（OpenAI）开始被套圈（被赶超）了。

[原文] [Speaker B]: Uh but yeah no it's and their TPUs too So they they're not as reliant on Nvidia chips as the other companies are

[译文] [Speaker B]: 呃，是啊。而且还有他们的 TPU（张量处理单元）。所以他们不像其他公司那样依赖 Nvidia 的芯片。

[原文] [Speaker A]: Yeah You saw the Facebook deal right It caused Nvidia stock to go down No What happened Uh Facebook bought Google TPU Uh I don't know what the exact name is but yes it could be TPUs or what They also have uh I know it wasn't their quantum computing tensor processing units Yes I know it wasn't their quantum computing chips because I'm assuming they're not for sale but they bought their quote unquote Nvidia uh copycat clone

[译文] [Speaker A]: 是的。你看到 Facebook 的交易了吧？它导致 Nvidia 股价下跌了。不，发生了什么？呃，Facebook 买了 Google 的 TPU。呃，我不知道确切的名字叫什么，但是是的，可能是 TPU 或别的什么。他们还有——我知道不是他们的量子计算张量处理单元。是的。我知道不是量子计算芯片，因为我假设那是非卖品，但他们买了那种所谓的“Nvidia 山寨克隆版”。

[原文] [Speaker A]: I'm not trying to say it in a bad way I just don't know what which version they bought but it caused Google stock to go up It got close to a $4 trillion market cap and Nvidia went down

[译文] [Speaker A]: 我不是想说得很难听，我只是不知道他们买了哪个版本，但这导致 Google 股价上涨。它的市值接近 4 万亿美元，而 Nvidia 下跌了。

[原文] [Speaker A]: But Google and all of them will just start dumping a ton of money in and catch up really Yeah When you have a cash machine like they have like it's you know it's it's I was talking to my dad yesterday about this like do do you think AI is a bubble And I'm like because you have these companies that generate so much free cash flow that are kind of reinvesting um into this space I don't think that is

[译文] [Speaker A]: 但 Google 和所有这些公司会开始砸入大量资金并真正追赶上来。是的。当你像他们那样拥有印钞机时——你知道——我昨天还在跟我爸聊这个，比如“你觉得 AI 是泡沫吗？”我说，因为你有这些能产生如此多自由现金流的公司正在向这个领域再投资，我不觉得那是泡沫。

[原文] [Speaker A]: I think maybe 90 to 95% of these AI startups are a bubble Um and what I mean by that is there's a lot of like I don't know if you're reading this um but I I'm seeing on Twitter now like one of my friends retweeted this He's a VC You actually know him

[译文] [Speaker A]: 我觉得大概 90% 到 95% 的这些 AI 初创公司是泡沫。呃，我的意思是，有很多像——我不知道你有没有读到这个，但我现在在 Twitter 上看到，就像我的一个朋友转发了这个。他是个风投（VC）。你其实认识他。

[原文] [Speaker A]: Um but what's happening is these these AI companies will raise a bunch of money and then what will happen is um they'll keep the money They won't get they won't talk to the VCs anymore And that's how easy it is to raise money in the AI space right now So I think 95% of these AI startups are a bubble but I think the scaleups are not

[译文] [Speaker A]: 呃，现在发生的情况是，这些 AI 公司会融一大笔钱，然后发生的是——呃，他们会把钱留着。他们不会再——他们不再跟风投说话了。现在在 AI 领域融资就是这么容易。所以我认为 95% 的 AI 初创公司都是泡沫，但我认为那些规模化扩张的公司（scaleups）不是。

章节 6：落地难题：组织架构、人效与招聘新标准

📝 本节摘要：

最后一节从工具层面回归到管理层面。演讲者坦言，虽然公司内部 AI 的采用率尚可，但在效率和产出质量（Satisfaction）上仍有巨大落差，主要问题在于员工制造了大量低质量的“数字垃圾”（Slop）且缺乏人工校验。对话还涉及了管理 Z 世代（Gen Z）员工的挑战——他们往往追求“酷炫”的营销活动而忽视底层的 KPI。最后，Speaker B 分享了从 Rippling CEO Parker Conrad 那里学到的招聘新招：在面试前直接发送公司的战略文档或投资者讲演稿，以此筛选出那些能真正理解业务指标、进行深度战略对话的高阶人才。

[原文] [Speaker B]: Which leads to another question How what percent of your company right now do you think you're happy with from a AI fluency standpoint like 10% 20% 50%

[译文] [Speaker B]: 这引出了另一个问题。你觉得你公司现在在 AI 熟练度（AI fluency）方面，有多少比例是你满意的？比如 10%、20% 还是 50%？

[原文] [Speaker A]: Per employee or you talking about our progress in leveraging because it's two different things

[译文] [Speaker A]: 是按员工个人算，还是说我们在利用（AI）方面的整体进度？因为这是两码事。

[原文] [Speaker B]: Second one Uh the second one was our progress overall implementing I would say I'm probably 30ish 40% of people 30 to 40% satisfied Call 35% 35% adoption rate where it's like hey these people know beyond just chat GPJ they can do other things that are a little more advanced for our specific industry and field being marketing

[译文] [Speaker B]: 第二个。呃，第二个是我们整体实施的进度。我会说我大概对 30% 到 40% 的人——30% 到 40% 感到满意。我们就说是 35% 吧。35% 的采用率，就是指这些人不仅知道 ChatGPT，他们还能做其他更高级的事情，特别是针对我们这个特定的营销行业和领域。

[原文] [Speaker A]: Yes Yes Okay So are you saying 35% satisfied or 35% adoption rate

[译文] [Speaker A]: 是的，是的。好的。所以你是说 35% 的满意度还是 35% 的采用率？

[原文] [Speaker A]: 35% satisfied 30 to 40% So we'll take what do you think the gap is right now Our biggest problem isn't the adoption Our biggest problem is the efficiency

[译文] [Speaker A]: 35% 的满意度，30% 到 40% 吧。所以我们来看看——你觉得现在的差距在哪里？我们最大的问题不是采用率。我们最大的问题是效率。

[原文] [Speaker A]: So let me give you an example I I'll use one that everyone who does marketing can understand Let's say you use AI to do keyword research You then take the keyword research You find the gaps between you and your competition and then you focus on writing content

[译文] [Speaker A]: 让我给你举个例子。我用一个做营销的人都能听懂的例子。假设你用 AI 做关键词研究。你拿到关键词研究结果，找到你和竞争对手之间的差距，然后你专注于撰写内容。

[原文] [Speaker A]: And let's say you have humans in the loop and then you're modifying the content Okay I would say 35% of the output 3 out of all the people in the organization I'm happy 35% of the time

[译文] [Speaker A]: 假设你让人类参与其中（humans in the loop），然后你在修改内容。好吧，我会说 35% 的产出——组织里所有的人中——我有 35% 的时间是满意的。

[原文] [Speaker A]: The biggest issue isn't the usage of the product The biggest issue is the slop that's being created with AI Part of that slop is technology which I'm assuming you agree with Part of the slop that a lot of companies don't want to talk about is the actual people using the AI and not working better at prompting or training something

[译文] [Speaker A]: 最大的问题不是产品的使用。最大的问题是用 AI 制造出来的“垃圾”（slop）。这种垃圾一部分源于技术本身，我想你会同意这点；但很多公司不愿意谈论的另一部分垃圾，是源于实际使用 AI 的人，他们没有在提示词（prompting）或训练方面做得更好。

[原文] [Speaker A]: Uhhuh And checking Checking I I put as standard If they don't check there's a problem with them and they shouldn't be in the organization if you're not willing to check your work But a lot of it is I wouldn't call it human error It's how humans use AI

[译文] [Speaker A]: 嗯哼。还有检查。检查，我把它设为标准。如果他们不检查，那就是他们的问题，他们就不该待在组织里——如果你不愿意检查你的工作的话。但这其中很多——我不称之为“人为错误”，而是人类如何使用 AI 的问题。

[原文] [Speaker A]: And if you don't get what you want you can modify it Uh the easiest way is is if it's giving you wrong information you can then end up going and being like okay go pull from this source You know you can leverage MCPs and you can say hey I'm going to use MCPs so then that way I can get a better output when I'm using AI

[译文] [Speaker A]: 如果你没得到你想要的结果，你可以修改它。呃，最简单的方法是，如果它给你错误的信息，你可以去说：“好吧，去从这个来源提取信息。”你知道你可以利用 MCPs，你可以说：“嘿，我要用 MCPs，这样我用 AI 时就能得到更好的输出。”

[原文] [Speaker A]: What we're finding is it takes a little bit longer for us what we've experienced for getting older people to adopt Not always but I'm seeing that And not just in my company but across the board But on the flip side with the younger people they're quick to adapt but a lot of times when you give them feedback they believe they know how to do it better and they don't keep in mind business metrics

[译文] [Speaker A]: 我们发现——根据我们的经验——让年纪大一点的人采用 AI 需要更长的时间。不总是这样，但我确实看到了这种现象。不仅是在我的公司，而是普遍如此。但在另一方面，年轻人适应得很快，但很多时候当你给他们反馈时，他们相信自己知道怎么做得更好，却不把业务指标（business metrics）放在心上。

[原文] [Speaker A]: Like they'll show like look at all this cool stuff Like there's this guy in San Diego He's really on top of with AI crypto uh social media marketing specifically out of all marketing channels And he has this Gen Z lingo like he'll use words like bet not know and stuff like that So sometimes I have a hard time keeping up with him

[译文] [Speaker A]: 比如他们会展示：“看这些东西多酷。”就像在圣地亚哥有个家伙，他在 AI、加密货币、呃，特别是所有营销渠道中的社交媒体营销方面真的很在行。他用这种 Z 世代（Gen Z）的行话，比如他会用“Bet”、“Not no”之类的词。所以我有时候很难跟上他的节奏。

[原文] [Speaker A]: But he updates me on the lingo and what it means Yeah And I'm like instead of not know why don't you just say yes but he's like no dude that's not what people say I'm like all right I still say yes or no instead of the not no Not no

[译文] [Speaker A]: 但他会教我这些行话和它们的意思。是的。我就说，与其说“Not no”，你为什么不直接说“Yes”呢？但他就像：“不，兄弟，人们不那么说话。”我就想：“行吧，我还是说 Yes 或 No，而不是那个 Not no。”Not no。

[原文] [Speaker A]: But he'll tell me like no dude this is what's wrong to do marketing with this crypto company This is what we need to do And I'm like he's like look how cool this is And he'll show me something like dude I get this but this doesn't actually move the bottom line

[译文] [Speaker A]: 但他会告诉我：“不，兄弟，这家加密货币公司的营销这么做是错的，我们需要这样做。”然后我就像——他就像：“看这多酷。”他给我看一些东西，我就说：“兄弟，我懂这个，但这实际上并不能推动底线（bottom line，意指实际利润/业绩）。”

[原文] [Speaker A]: I'm like "Dude this is not tied to any KPI We're getting paid to help them move the KPIs dude." Like it's it's just going around in circles dude

[译文] [Speaker A]: 我说：“兄弟，这没有跟任何 KPI 挂钩。我们拿钱是为了帮他们推动 KPI 的，兄弟。”这就像是在原地打转，兄弟。

[原文] [Speaker B]: I um you know the Ripling CEO right Uh Parker Conrad So um you know Brian Hagan the CEO of HubSpot I started a new podcast and I was like "Oh he's interviewing all these CEOs I want to listen." So I I listened to Parker and this is a really interesting tactic that I started trying actually as soon as this week

[译文] [Speaker B]: 我呃——你知道 Rippling 的 CEO 对吧？呃，Parker Conrad。还有你知道 HubSpot 的 CEO Brian Halligan。我开始听一个新的播客，我就想：“噢，他在采访所有这些 CEO，我想听听。”所以我听了 Parker 的那期，这是一个非常有趣的策略，我这周其实也开始尝试了。

[原文] [Speaker B]: And so what Parker does is he actually sends um recruits that he's going to talk to Um he sends them the investor decks all these things the strategy around the company His expectation is when you come into the the meeting with him you're going to talk you're going to have questions and you're going to have a deep strategic conversation

[译文] [Speaker B]: Parker 做的是，他实际上会给那些他准备谈话的应聘者发送——呃，他发给他们投资者讲演稿（investor decks），所有这些东西，关于公司的战略。他的期望是，当你进会议室跟他面谈时，你要能谈论这些，你要有问题，并且你要能进行一场深度的战略对话。

[原文] [Speaker B]: And you can tell by that conversation over 30 minutes or so if that person is they they know what they're talking about when it comes to metrics and how they think how their motor is right then after that he just focuses on backdoor reference checks

[译文] [Speaker B]: 通过那大约 30 分钟的对话，你就能判断那个人在指标方面是否懂行，他们的思维方式如何，他们的驱动力（motor）如何。对，然后在那之后，他就专注于做背景调查（backdoor reference checks）。

[原文] [Speaker B]: I thought that was interesting That is interesting Yeah Worth trying I tried I tried it this week I tried it on one person Okay And she could not hold it together She was just like she kept talking about herself She kept going back to herself and she may have she might have just asked like two or three questions around the document that I sent which is the plan for next year So what will you do with this person we will not hire this person

[译文] [Speaker B]: 我觉得那很有趣。确实有趣。是的，值得一试。这周我试了，我在一个人身上试了。好的。结果她完全撑不住场面。她就像——她一直谈论她自己。她一直把话题绕回她自己身上，她可能——她大概只问了两三个关于我发给她的那份明年计划文档的问题。所以你会怎么处理这个人？我们不会雇佣这个人。

[原文] [Speaker B]: I think every time you double as a company like a lot of things are going to break Processes are going to break People are going to break as well right And so some people will scale like Noah has seen this right Some people will scale some people will not be able to And it's it's all about making sure that you're setting the right standards right And this is one of the standards

[译文] [Speaker B]: 我觉得每当公司规模翻倍时，很多东西都会崩溃。流程会崩溃，人也会崩溃，对吧？所以有些人能跟上规模扩张——就像 Noah 见过的——有些人能跟上，有些人则不能。这一切都是为了确保你设定了正确的标准，对吧。这就是其中一个标准。