OpenAI将支持成人内容，CoreWeave的AI牧场，苹果M5 Vision Pro | Zachary Perret、Panos Panay、Eric Wittman、Brian Baumgartner、Adam Ryan、Adit Abraham、Dan Shipper等

关于上周五生效的对华100%关税，我有一个更新，这让市场陷入了动荡。Polymarket显示，有16%的概率该关税将在11月1日仍然有效。所以大家某种程度上都在预期某种重新谈判，贸易战即将到来。希望——我的意思是，情况已经开始稳定。今天市场实际表现如何？

I have an update on the 100% tariff on China that went into effect last Friday, put the market in turmoil. Polymarket has a 16% chance that that 100% tariff is in effect on November 1. And so everyone's kind of pricing some sort of renegotiation in the trade war coming soon. And hopefully I mean, things are already starting to stabilize. What is the market actually doing today?

纳斯达克上涨了0.5%。比特币下跌了1.5%。但总体而言

We're up half a percent in the Nasdaq. Bitcoin's down one and a half percent. But overall

半个

Half a

点的适度增长。

point of modest modest increase.

是的。

Yes.

但我们会接受的。

But we'll take it.

泡沫破裂了。我们现在正在重新膨胀。这很棒。安妮说，我认识的这个人给他妻子买了一个这样的东西，还花了500美元做社交媒体广告，让她有事可做，从而资助了她的Etsy贴纸生意。现在他已经完全辞掉了工作，因为她赚的钱太多了。

The bubble popped. We're reinflating now. It's great. Annie says, this guy I know funded his wife's Etsy sticker business by buying her one of these and, like, $500 in social media ads to give her something to do. And now he's quit his job entirely because she's making them so much money.

哇。

Wow.

创始人。太神奇了。

Founder. That's amazing.

Etsy Etsy 现在感觉像是一个被低估的故事，尤其是在类似 TPT 的背景下。TPT，就像是长尾电商。是的。本·汤普森提到了一些

Etsy Etsy feels like an underrated story right now, especially with the sort of like TPT. TPT, like long tail commerce. Like Yeah. Ben Thompson told some

嗯，这个整合之所以很酷，是因为我可以想象一个未来，你可以生成一个物品的模型，嗯。然后找到一个 Etsy 卖家制作类似的东西，是的。直接请他们制作。完全正确。然后只需一键就能搞定，是的。

Well, the reason that integration is actually cool is I can imagine a future where you can generate, like, a mock up for an object Mhmm. And then find an Etsy seller that makes something like it Yep. And just ask them to make it. Totally. And it'll be like one click to go to Yeah.

就像，你生成一张图片，是的。或者视频什么的，然后就能找到合适的供应商，他们能实际为你实体制作出来。是的。只需一键，你知道，就能购买。

Like, you generate an image Yeah. Or something or video, and you can just find the right vendor that can actually physically make it for you. Yeah. And in one click, you know, purchase it.

这才是真正的，像是，3D 打印热潮。大约十年前就有这样的愿景，认为 3D 打印机现在能用了。每个人家里都会有一台 3D 打印机，他们能打印任何想要的东西。这并没有完全实现，但有很多 Etsy 卖家已经找到了特定的 3D 打印物品。我们家里刚买了一个，是为了更高效地拆卸婴儿车的轮子，它就是一个 3D 打印的小钥匙，用来卸下轮子。

This is the real, like, like, three d printing boom. Like, there was this whole vision about a decade ago of, like, three d printers now work. Everyone's gonna have a three d printer in their home, and they'll be able to print whatever they wanted. That didn't quite happen, but there are a lot of Etsy sellers that have figured out specific three d printed objects. We just got one at home for taking the wheel off a stroller more efficiently, and it's just a three d printed little key that takes that off.

本·汤普森谈到了一些 3D 打印设备。我忘了具体是什么，但好像是它能把 HDMI 线固定到特定设备上，以一个完美的角度，或者它以某种方式适配一个机架安装的东西。就是这样一种情况，没有真正的市场来完全量产，比如大规模生产，但一个 3D 打印机可以在 Etsy 上找到合适的客户。所以你看到这种超级长尾正在出现，这说实话非常迷人。令人兴奋。

Ben Thompson was talking about some some three d printed device. I I forget what it was, but it was like it holds an HDMI cable into a specific device, like, in a in a perfect angle, and there was like or or it adapts a rack mounted thing in some way. And it was just something where, like, there's no real market to fully productionize, like, the and mass produce it, but a three d printer can go on Etsy and and find the right customer. And so you're seeing this, like, super long tail emerge, which is honestly fascinating. Exciting.

好吧，在我们下一位嘉宾加入之前，我们的第一位嘉宾，让我告诉你关于 Julius。你想运行什么分析？与你的数据对话，在几秒钟内获得专家级的见解。它是为你工作的 AI 数据分析师。我们还有来自 Plaid 的 Zach Perre 在 restream 等候室。

Well, before we have our next guest join, our first guest, let me tell you about Julius. What analysis do you wanna run? Chat with your data and get expert level insights in seconds. It's the AI data analyst that works for you. And we have Zach Perre from Plaid in the restream waiting room.

让我们把他请到 TBP 和 Ultradome。怎么样

Let's bring him to the TBP And Ultradome. How are

扎克，你怎么样？

you doing, Zach?

各位，很高兴见到你们。谢谢你们的邀请。

Guys, nice to see you. Thank you for having me.

很高兴见到你。

Good to see you.

穿得很精神。你今天在华盛顿特区吗，还是只是为了TBPN的场合特意打扮的。TBPN。

Looking sharp. Are you in DC today, or is it just you're just dressed for the occasion of of TBPN. TBPN.

我是说，你们播新闻的时候都穿得这么好看。我觉得我也得打扮一下。不，我在华盛顿特区。就是特区。

I mean, you all dress so nicely when you're you're you're doing the news. I figured I had to dress up too. No. I'm I'm in DC. It's DC.

我懂了。原来你在这儿

I it. There you

等等。这太棒了。

Wait. This is awesome.

如果你没看出来，我们借了个地堡来拍摄这个。

If you couldn't tell, we we borrowed a bunker to to to film this from.

所以

So

看起来很棒。

It looks great.

不错，背景很好。

Nice, great background.

是的。隔音效果非常非常好。是的。给我们更新一下情况。观众都熟悉公司，但你那边有什么最新最棒的消息？

Yeah. Very, very, very sound padded. Yeah. Give us give us the update. Everyone in the audience is familiar with the company, but, what what's the latest and greatest in your world?

所以今天对我们来说是个大日子。我们每年发布两次重大产品。这是我们秋季的产品发布。有很多令人兴奋的新东西刚刚推出。其中最重要的可能是一款我们称之为LensScore的产品。

So today's a big day for us. We we do two big product releases each year. This is our fall product release. Lots of new exciting stuff that that just came out. Probably chief among it is a product that we call LensScore.

这是一个新的信用评分。嗯。我多年来一直在开发一系列信用分析产品，而这个LensScore能提供一个非常全面的潜在借款人画像。假设你找到了一份新工作，你的收入大幅增加了。

It is a new credit score. Mhmm. I've been working on a bunch of credit analytics products for many years, and this LensScore gives a really robust picture of a potential borrower. Let's say that you got a new job. Your income went up a lot.

我们实际上可以实时识别这类情况。比如说，你在过去24小时内申请了七笔贷款，这很可能相当可疑。是的，我们能实时识别这类行为。所以这是对信用评分系统的一次重大升级。

We could actually recognize that kind of thing in real time. Let's say that you sign up for seven loans all in the last twenty four hours, it's probably pretty sketchy. Yeah. We can recognize that kind of thing in real time. So a big upgrade to to to the credit scores.

我们还推出了大量关于风险与欺诈分析、支付等方面的功能。所以今天对我们来说是个大日子。

Also launched a bunch of stuff around risk and fraud analytics, payments, so on and so forth. So big big day for us today.

传统欺诈评分是如何运作的？就像我了解的，你知道，满分是800分，你会得到一个随机数字。即使你还清贷款，但总有些奇怪的事情，比如你会听到这样的建议：哦，不要提前还清这笔贷款，因为那会损害你的信用分数。这总是让人觉得非常难以理解。那么，你对现有系统有哪些不满意的地方？

How do legacy fraud scores work? Like, I'm I'm familiar with, you know, the it's out of 800, and you get a random number. And if you pay off but there's always, like, weird, like, there's these weird things and you hear recommendations like, oh, don't pay off this loan early because that'll actually hurt your score. It always feels very, very indecipherable. But, like, what what what did you not like about the current system?

目前存在什么问题？

What's currently broken?

嗯，你的感受正是过去大家对信用评分的普遍感受。它们通常是350到850分的评分体系，主要基于你的还款历史和过去申请的贷款数量。

Well, the the way that you feel is the way that everybody feels about credit scores in the past. They're they're kind of a three fifty to eight fifty score. It's largely based on your your history of payments and the amount of loans that you've taken out in the past.

嗯哼。

Mhmm.

而且它不能实时反映你生活中的变化。它还忽略了许多重要的数据输入。几十年来，贷款行业一直建立在这些信用评分之上，它们还算管用。但当你考虑普通贷款机构时，它们需要更多信息，因为如今贷款的基本面已经改变。贷款欺诈尝试大幅增加了。

And it's not responsive to the changes in your life in real time. It also skips a lot of important data inputs. Now the the lending industry has been built on these credit scores for for decades and they work okay. But when you think about an average lender, they want more information because today, the fundamentals of lending have changed. There's a lot more attempted fraud in lending.

你本可以收集到更多新的见解，而且坦率地说，消费者的行为模式变化很大。因此，回顾七年前的数据可能没有太大意义。我自己曾去看验光师，结果那位验光师把账单寄到了我的旧地址。

There's a lot more new insights that you could be gathering and frankly consumers change their patterns a lot more. So it could the seven year look back doesn't make a ton of sense. I myself, I had gone to an optometrist and that optometrist self sent the bill to an old address for me.

确实。

Sure.

而我的信用评分因此受到影响，因为我从未收到那张账单。是的，持续了好几年。最后我才终于搞清楚是怎么回事。

And my credit score was hit because I never got that bill Yep. For years. And then I finally figured it out.

我也遇到过同样

I had the same

的情况。

thing happen.

所以我们的新版本

So our new version

是的。我的第一家公司在Verizon合同下有一大堆Wi-Fi热点设备。

Yeah. My first, my first company, we had a whole bunch of, like, Wi Fi hotspots on Verizon contracts.

这事也发生在我身上

This happened to me this

一模一样的情况

exact same thing

我也是。我把路由器寄回去了，结果信用分被扣了，他们却说没收到。我就说，所以是你们搞丢了，你们有我卡号存档，为什么不直接扣款呢？

to me too. My credit got dinged because I sent the router back, and they were like, we never got it. Yeah. I was like, so you lost it, and you have my card on file, and you didn't just charge me for it?

然后就只能自己处理。是啊，非常非常令人沮丧。肯定有很多不满。我还有很多问题，不过抱歉打断你了。

And then just deal with it. Yeah. It's very, very frustrating. I'm sure there's so much frustration. Have bunch more questions, but sorry I cut you off.

是的。我想我主要的问题是，显然行业和消费者理应获得更现代的信用评分体系，但现有的评分系统似乎根深蒂固。就像你可以做出更好的产品，但要如何真正让广大贷款机构广泛采用呢？

Yeah. I guess, my my main question is like, okay. Obviously, the industry and consumers deserve, like, a more modern credit score, but it feels like these the existing scores are heavily entrenched and it's it's like you can make a better product, but how do you actually get it to mass adoption among, you know, lenders broadly?

这是个很好的问题。我们推出的新产品专注于实时评估消费者。比如如何分析你的收入？如何分析你的支出？通过计算可以得出自由现金流，这应该是一个消费者每月或每年能偿还的金额。

It's it's a great question. The the new product that we launched is focused on looking at consumers in real time. So how do we look at your income? How do we look at your expenses? You can do the math to figure out free cash flow, and that should be the amount that a consumer can pay back in a given month or a year.

我们最初推出这种方式实际上是与许多伙伴合作的。所以我们会与现有征信局、众多现有贷款机构合作，将这项服务与FICO或VantageScore并列提供。这样你可以查看FICO评分，然后可以说，哦，但这个用户刚换了工作，让我获取新工作的数据。嗯。

And the way that we're launching it originally is actually in partnership with a lot of folks. So we'll partner with the existing bureaus, with with a lot of the existing lenders to add this alongside FICO or alongside VantageScore. So you can look at a FICO and then you can say, oh, but this user said that they just changed jobs. Let me get the data from that new job. Mhmm.

所以我们可以加入这一点。然后我们还有一个独特的洞察，这是只有我们才拥有的，因为我们与所有贷款机构合作，可以看到贷款申请的速度。我们称之为网络洞察。基于你在整个平台网络上的流量，我们还能看到许多其他数据点。是的。

So we can add that in. And then there's this interesting insight that we have which which only we have because we work with all of the lenders, and we can see the velocity of loans applied for. We call this network insights. There's a bunch of other data points that we can see just based on the traffic that you're doing across the entire platform network. Yep.

所以这算是一种独特的向量，人们现在可以据此放贷。最终效果是，我们看到，通常来说，这种新方法带来的结果质量提升了20%左右。而且实际上有更多人符合贷款资格，因为比如说他们收入很高，租金支付记录良好，这实际上应该表明他们是更好的贷款对象。

So it's kind of a a unique vector that people can now lend against. The net effect is we're seeing, you know, call it oftentimes 20% lift in the quality of outcomes from from from this new one. And we're seeing a lot more people actually be eligible for loans because let's say they have a really high income and good rent payment history, that actually should indicate that they're a better loan.

你是怎么考虑的？我猜想实际上有一个非常复杂的算法来决定这个分数。里面可能有一些机器学习或人工智能。但你也想提供一些人类可读的洞察，这样你可以告诉某人为什么他们的分数下降了，或者他们可以做些什么来提高分数。我有一个朋友，他从事的不是高频交易。更像是量化交易，所以大概是几天内的交易。

How do you think about I I imagine that there's a very complex algorithm to actually just to decide the score. There's probably some machine learning or AI in there. But you also wanna deliver something that's like a human readable insight so you can tell someone why their score dropped or what they can do to make their score better maybe. I I have a friend who works in it's not high frequency trading. It's like, out like, it's quantitative, trading, so it's like a couple days out.

然后他说，我们会处理所有这些数字，提出一些论点，比如一天内买入某家公司十亿美元的股票，然后在三天内卖出。我就问，有没有办法说，哦，是因为黄金上涨了之类的？你能把它变成业务逻辑或者人类能理解的if语句吗？他说，不可能。因为金融市场中寻找阿尔法收益涉及太多不同的参数了。

And, and he was like, we'll crunch all these numbers. We'll come up with some, like, thesis for buying a billion dollars of some company in a day, and we'll be selling in, like, three days. And I was like, is there any way that you could say, oh, well, it's because, like, if gold goes up like, could you put it in business logic or, like, if statements that a human could understand? He was like, it's impossible. Like, there's just so many different parameters that go into finding, like, alpha in the financial markets.

我觉得如果对消费者说，你的分数上升了一点，但我们完全不知道原因，也无法告诉你任何相关信息，只是这些张量、这些权重在这个庞大模型中激活了，然后你的分数就下降了，抱歉——这种情况下消费者会非常失望。

And I feel like there's a world where a consumer would be very disappointed if you're like, yeah. Your score went up one point, but we have no idea why. And we can't tell you the first thing about why it went up. Just these tensors, these weights activated in this massive model, and your score went down. Sorry.

那么你实际上是如何考虑教育消费者，让他们理解自己的分数是如何上升或下降的呢？

So how do you think about actually educating the consumer on like how how their score goes up or down?

所以我们确实做了很多机器学习和人工智能的工作，显然是为了清理数据并以结构化方式组织，以便我们能在其上构建模型。

So we we do a bunch of machine learning and AI to obviously like clean up the data and make it structured in a way that we can build our models on top

的。

of it.

但最终，我们的模型大致代表了消费者的财务健康状况。因此，消费者为更好地管理资金所做的几乎所有事情，最终都会带来更好的信用评分。而他们管理资金越糟糕，几乎都会导致信用评分变差。所以好处在于，当我们构建它时，我们实际上能够先展示给消费者看。他们能够进行审核。

But ultimately, our models are representative roughly of the financial health of the consumer. So almost everything that that a consumer would do to better manage their money will end up with a better credit score. Almost everything they would do to more poorly manage their money will end up with a worse credit score. And so the nice thing is when we build it, we're actually able to show it to the consumer first. They're able to review it.

如果他们愿意，点击提交，然后它就会发送给贷款机构。这样贷款机构随后就可以据此做出决定。

And if they want to, click submit, and then it goes to the lender. So the lender can then subsequently make the decision on the back of it.

这很酷。所以它对贷款机构非常友好，因为他们是选择加入的。如果他们只想使用他们的益博睿（Experian）评分或其他任何可选方案，他们可以那样做，或者他们可以把这个加进来，作为一种改进他们申请的方式。

That's cool. So so it's very lender friendly in that they're opting in. If they want to just run with their Experian score or whatever what other options they have, they can they can do that or they can add this on as a way to kind of improve their application.

正是如此。是的。所以这个想法是

Exactly. Yeah. So the idea is that

不过随着时间的推移，贷款机构会意识到，嘿，我们实际上在这里获得了更好的洞察力。更多的信号。我想，他们会愿意直接采用这个。

for time though over the over time though, the lender would realize, hey, we're getting actually much better insight here. More signal. We wanna just, you know, roll with this, I imagine.

这正是我们所希望的。但最初的目标是让消费者感觉他们可以获得更多更好的贷款。让贷款机构感觉他们可以提供更多更好的贷款。所以这最初是在打开市场。然后随着时间的推移，你知道，我们希望这能成为贷款机构做出决策的最重要方式之一，即使不是最重要的方式。

That's what we hope. But the goal at first is that consumers feel like they can get more and better loans. Lenders feel like they can make more and better loans. So it's kind of opening the market initially. And then over time, you know, we hope that this becomes one of the most important, if not the most important way that a lender makes their decisions.

现在，先买后付在这个生态系统中处于什么位置？我想它们都有自己的系统，但它们肯定希望获得更多数据，因为这可能有助于更好的风险评估。它们是如何融入这个生态系统的？

How does buy now, pay later fit into the ecosystem at this point? I imagine that they all have, like, their own systems, but then they would love to just have more data because that probably enables better underwriting. How do they fit into this ecosystem?

所以先买后付是成熟的贷款机构。贷款机构的成熟度是有层次的，你知道。有些贷款机构只会使用现成的FICO评分或LendScore评分来做贷款决策。而高度成熟的贷款机构，它们会想要大量的输入数据。是的。

So buy now, pay later are sophisticated lenders. There's a scale of sophistication of all lenders, you know. There are some lenders that will just use an out of the box FICO score or an out of the box LendScore score to make a lending decision. The highly sophisticated lenders, they'll want a zillion inputs. Yes.

它们可能仍然会获取FICO评分。它们会获取PlaidLendScore，但它们希望查看底层输入，即原始数据，然后在此基础上构建自己的模型。我认为先买后付贷款机构在贷款方式上相当成熟。因此，它们会引入大量数据，并在此基础上构建自己的定制模型。我们的输入仍然是它们模型的输入，但它们不会将我们的评分作为唯一的决定因素。

They'll still get maybe a FICO. They'll they'll get a PlaidLendScore, but they wanna look at the underlying inputs, the underlying raw data, and then they'll build their own models on top of it. I'd say the BNPL lenders are quite sophisticated in the way that they lend. And so they're they're pulling a ton of data into the picture and then building kind of their own custom models on top of it. Now our inputs are, yes, still inputs to their model, but they're not looking at our score as the sole determining factor.

是的。LENS评分的商业模式与之前的信用评分相同还是有所不同？你怎么看？挺相似的。

Yeah. Is the business model for the LENS score the same as the previous credit scores or different in some way? How do you think about Pretty similar.

是的，挺相似的。我们按每笔贷款申请来销售。我们的客户是贷款机构，它们为此向我们付费。

Yeah. Pretty similar. We sell it on a kind of per loan application basis. Our customers are the lenders. They pay us for it.

明白了。

Got it.

我们有一些方法可以帮助它们持续跟踪信用状况。而FICO或Vantage通常是一次性的快照。有时贷款机构可能希望向消费者提供优惠，如果它们意识到想增加信用额度，或者如果认为消费者能更快还清贷款，可能会提供更优惠的付款条件。因此，我们的评分是持续性的信用评分。这可能是唯一的主要区别。

There's a handful of things that we can do to actually help them track credit worthiness on an ongoing basis. So whereas FICO or Vantage oftentimes is a one time snapshot. Sometimes a lender will want to maybe make offers to a consumer if they realize they want to increase their credit line or maybe they want to kind of give better preferential payment terms to a consumer if they think they can pay a loan down faster. And so ours is an ongoing credit score. That's maybe the only big difference.

您如何看待消费者获取信用评分的界面？我记得，几年前，你实际上必须去三家信用机构。是的，你得去信用机构，然后支付大约30美元才能查看。然后你会拿到报告，再带着它去申请贷款。

How do you think about what the consumer's interface is to the credit scores? I remember, like, years ago, you used to have to actually go to the three credit Go to the credit store. Yeah. You had to go to the credit store and, like, check out and pay, like, $30. And then you'd get that, and then you'd take that to a loan application.

有时候，有时候他们会在后台处理，他们会支付费用。但随着时间的推移，我觉得信用评分几乎变得商品化或免费了。比如，它经常出现在我的银行应用里。还有像Credit Karma这样的公司，他们找到了方法，基本上将其作为客户获取成本免费提供，然后在上面提供一系列可以货币化的产品。您认为LENS评分的长期消费者采用会是什么样子？

Sometimes sometimes they just do it on the back end, they'd pay. But then over time, I feel like the credit scores, just became almost commoditized or free. Like, it just shows up in my, you know, banking app a lot. And there were also, like, Credit Karma and different companies that figured out ways to basically pay that as a, like, a customer acquisition cost for free, and then they give you a bunch of products on top that they would monetize. How do you think the the the what do you think the long term consumer adoption of the LENS score looks like?

嗯，我们希望它非常相似。实际上，我们希望消费者能够免费清晰地查看他们的信用档案。我们正在努力构建方式，让消费者可以进来查看。并且，最终客户将是贷款机构。对我们来说，让消费者更了解这一点是有益的。

Well, we hope it's pretty similar. Actually, we want consumers to be able to get a clear view of their credit profile for free. We're working on building ways that a consumer can come in and and look at that. And, know, ultimately the customer then would be the lender. Like, it's it's to our benefit to have consumers that understand this more.

让消费者信任这个评分以及我们构建的东西，对我们是有益的。因此，我们希望尽可能直接和开放地与消费者沟通。所以很多这方面的内容即将推出，您会在未来几个季度看到我们在这方面的一些发布。

It's to our benefit to have consumers that trust the score and that that that trust the things that we build. And so we want to be as direct and open with the consumer as we can. So a lot of that stuff is to come, you'll see some launches from us kind of over the coming quarters on that front.

除了Lens，今天还有哪些重点是？

What else besides Lens was the focus today?

所以我们对反欺诈模型进行了一次重大升级，并推出了一个更广泛的产品套件，称为Protect。Protect对我们来说是一套产品，它查看消费者在几乎所有金融产品上的底层行为。我们可以将其转化为消费者的总体风险评分和一系列属性。有些属性非常容易解释，比如这个人在过去三十分钟内注册了三个加密应用。有些属性则不那么容易解释，更多地涉及机器学习方面。

So a big big upgrade to our anti fraud model and we launched a kind of broader suite of products called Protect. So again, Protect for us, it's a set of products that looks at consumers' underlying actions that they take across almost any financial product. We can back it into an overall score of riskiness for the consumer and then a bunch of attributes. So some of the attributes are very easily explainable such as this person just signed up for three crypto apps in the last thirty minutes. Some of the attributes are less easily explainable more on the the the machine learning side.

但这些信息将输入到我们许多客户的模型中，然后他们可以根据用户的相对风险水平，构建这种非常吸引人的升级或降级要求。对于许多试图更快注册更多用户的金融科技公司来说，这是一个巨大的解锁。我们可以有效地创建一条快速通道。所以我们以前见过您。我们识别您的设备。

But this stuff will feed into the the models for a lot of our customers and then they can build kind of this really fascinating step up, step down requirement for users based on their relative levels of riskiness. There's been a huge unlock for a lot of the fintech companies that are trying to sign up more users faster. We can create a fast lane effectively. So we've seen you before. We recognize your device.

我们已经看到了你的电话号码。你只需完成双重验证码，就可以顺利进行了。哇。

We've seen your phone number. You just do a two factor code and you're good to go. Wow.

然后呢

And then

我们还可以创建一个慢速通道。所以，嘿，我们以前从未见过你。你的设备似乎与你拨号使用的IP地址不匹配。这是怎么回事？让我们增加一些验证步骤，确保你真的是你声称的那个人。

we can also create a slow lane. So hey, we've never seen you before. Your device doesn't seem to match the IP address that you're you're dialing in from. What's going on with all this stuff? Let's let's add a bunch of step ups and, you know, be really, really certain that you are who you say you are.

这就是金融服务通道，也是我们正在努力构建的。

So that's lane for financial services, and that's what we're trying to build.

这听起来像是进行大规模网络金融诈骗的人最可怕的噩梦。

This sounds like the worst nightmare of somebody doing large scale financial fraud on the Internet.

Plaid目前AI应用的形态是怎样的？我觉得，显然你们可能在软件开发方面做了大量AI工作。但同时，如果我在使用Plaid时它问我：嘿，现在想和我聊天吗？我会说：不用了。

What is, what what's the shape of AI adoption at Plaid right now? I feel like, obviously, there's probably a ton of AI that you're doing on just, like, software development stuff. But at the same time, I feel like if I was offing with Plaid and it was like, hey. Do you wanna chat with me right now? I'd be like, no.

谢谢。我其实点几个按钮就够了。那么，你们看到很多LLM令牌消耗吗？还是更专注于机器学习、定制化的承保模型，即针对特定场景训练专门模型，而不是依赖大型令牌工厂处理大量工作？

Thanks. I'm actually good to just click a couple buttons. So, like, are you seeing a lot of, like, LLM token consumption or more just focused on the machine learning, the custom models for underwriting where you train a specific model for that and you're not just leaning on a big, like, token factory for a lot of work?

我会说我们对代币的使用，我称之为中等程度。当然，我们会构建一种自定义训练数据集，然后通过那个模型运行大量数据，并且对训练数据集进行标注。其中很多工作只是清理交易记录，识别那些毫无意义的内容。正在发生什么样的嗅探行为。

I I'd say our our use of tokens, I would call it, like, medium. Sure. We do build kind of a custom training dataset and then run a ton of the data through that that that model and the dating the training dataset. A lot of it is around just cleaning up transactions, identifying That makes no sense. What sniffing is happening.

是的。比如，文本分析在这方面非常非常有帮助。当然。所以这就像是我们许多模型的输入。是的。

Yeah. Like, the textual analysis is very, very helpful for that. Sure. So that's that's like the input into a lot of our models. Yeah.

但你说得对。我们并没有为任何消费者构建活跃的聊天应用。我们做的很多人工智能是用来对抗人工智能的，意思是存在大量试图进行的人工智能欺诈，然后我们有很多人工智能在另一边试图对抗人工智能欺诈。然后这就像是构建基础数据集，我们在此基础上建立许多模型。

But you're right. We don't build like an active chat app for any consumer. A lot of the AI that we're doing is used to fight AI, meaning there's a lot of attempted AI fraud and then we have a lot of AI on the other side to try to fight AI fraud. And then it's it's kind of building the foundational kind of dataset upon which we build a lot of the models.

是的。所以每当你在后台有大量文本时，你可以运行大型语言模型来处理它，获得更细粒度的数据，进行分类，理解这些大型文本转储的实际内容。但你甚至不一定需要将这些直接呈现给客户，即最终用户，这很棒。

Yeah. So whenever you have a big pile of text in in kind of behind the scenes in the back end, you can run LLMs over that and get a lot finer grained data, categorize, understand what's actually going on with those large text dumps. But you don't necessarily even need to surface that to the to the customer, the end user, which is great.

通常，不会直接呈现。通常，我们会间接地将其呈现给消费者，比如，嘿，你的银行费用升高了。我们甚至不会直接发送给消费者，而是发送给我们的客户。

Generally, not directly. Generally, we'll we'll surface it to the consumer indirectly saying, hey. You had elevated bank fees. And we won't even send that to the consumer. We'll send that to our customer.

我们的客户随后会说，好吧。在你的预算应用中，你的银行费用升高了，看看所有这些不同的银行费用。

Our customer will then say, alright. In your budgeting app, you had elevated bank fees and look at all these different bank fees.

这非常有道理。好吧，恭喜发布。非常感谢你的到来。总是很愉快

That makes a ton of sense. Well, congrats on the launch. Thanks so much for stopping by. Always great to

再见。玩得开心。

see you. Have fun.

谢谢

Thank you

邀请我。

for having me.

祝你一切顺利。

Have a good one.

干杯，扎克。

Cheers, Zach.

在我们请出节目的第二位嘉宾之前，让我介绍一下Fall，这是一款面向开发者的生成式媒体应用。全球顶尖的生成式图像、视频和音频模型齐聚一堂，使用无服务器GPU和按需集群开发微调模型。而我们的下一位嘉宾即将进入直播等候室。

Before we hop on with our second guest of the show, let me tell you about Fall, the generative media app for developers. The world's best generative image, video, and model audio models all in one place, developing fine tune models with serverless GPUs and on demand clusters. And our next guest is about to be in the restream waiting room.

你可以回到时间线。贝莱德和英伟达正参与一项400亿美元的数据中心收购。这是个好消息。包括XAI和微软也参与其中。这真是好消息。

You can go back to the timeline. BlackRock and NVIDIA are participating in a $40,000,000,000 data center takeover. That's great news. Including XAI and Microsoft. That's great news.

今天《金融时报》报道，一个包括贝莱德全球基础设施合伙公司在内的投资财团——这个名字对一家从事全球基础设施投资的基金来说真是恰如其分。没错。当然，还有阿布扎比基金MGX，已达成以400亿美元收购全球最大数据中心运营商之一的交易，同时启动一项旨在支持人工智能基础设施的计划。我相信我们的下一位嘉宾已在Restream等候室。让我们请他进来。

This is in the Financial Times this morning invest an investment consortium that includes BlackRock Global Infrastructure Partners, and that's a great name for a fund that does global infrastructure investing. Yes. And, of course, Abu Dhabi Fund MGX, has struck a $40,000,000,000 takeover of one of the world's largest data center operators as it launches an initiative to underwrite the infrastructure for an artificial intelligence. And I believe we have our next guest in the Restream waiting room. So let's bring him in.

来自亚马逊的Ponak。欢迎来到节目。你最近怎么样？Ponak。

Ponak from Amazon. Welcome to the show. How are you doing? Ponak.

很高兴你能来。

Great to have you.

非常感谢你抽空参加。恭喜发布。抱歉有点延迟，但非常兴奋能与你交流。想先请你做个自我介绍，比如你在组织中的职位，然后我们可以聊聊一些新闻和更新。

Thanks so much for hopping on. Congratulations on launch. Sorry for, the delay, but very excited to catch up with you. Would love to start with just the introduction on yourself, kind of where you sit within the organization, and then we can go into some of the news and some of the updates.

当然。我是设备与服务的高级副总裁。简单来说，如果用品牌来框架可能更容易理解。对。但想想亚马逊所有的设备。

Sure. I'm the senior vice president, devices and services. So what that means is, just if I frame it around brands, it probably helps a little bit. Yeah. But think about devices across Amazon.

是的。从Project Kuiper开始，这是我们正在建设的低地球轨道卫星星座。另一端是Zoox，你可能在拉斯维加斯见过的自动驾驶汽车。对。但在这之间是消费品牌，如Alexa、Ring、Fire TV、Fire平板、Kindle、Echo设备，你知道，就是我们在亚马逊打造的各种面向消费者的设备阵列。

Yep. You'd start with Project Kuiper, which are the satellites for LEO constellation that we're building right now. And on the other end, you would think about Zoox, which are the self driving cars that you might have seen in Las Vegas. Yeah. But in between kind of the consumer brands, Alexa, Ring, Fire TV, Fire tablets, Kindle, Echo devices, you know, just an array of all the kind of consumer front devices that we build here at Amazon.

嗯。那么硬件生态系统的最新更新是什么？最近宣布了哪些消息？

Yeah. And and what, what what's the latest update across the hardware ecosystem? What was announced most recently?

我们刚刚发布了一系列新的Echo设备，非常酷，可以说是专为Alexa Plus打造的。这是我们新的Alexa Plus，是我们使用Alexa的新AI界面，更加对话化。这是一个非常有趣的产品，能帮你完成各种事情。有两款带屏幕的设备，分别是8英寸和11英寸，还有两款新音箱，音效极佳。

We just launched a set of new Echo devices, super cool, kind of built for Alexa Plus. This is our new Alexa Plus is our new AI interface using Alexa, just more conversational. It's a really fun product. It helps you get stuff done. Two two screen based devices, an eight inch and 11 inch and two new speakers, killer sound.

顺便说一句，它们绝对漂亮，简直美极了。如果你想的话我们可以聊聊。是的，还有三款新的Kindle，我们称之为Kindle Scribe。

They're absolutely beautiful, by the way. They're just gorgeous. We can talk about it if you want. Yeah. Three new Kindles, we call the Kindle Scribe.

这些设备中有一款是彩色的，拥有绚丽的彩色书写功能。还有一系列4K摄像头与Ring配合使用，是的，包括门铃也是4K的。这带来了非常清晰的画质，相当酷。这些都是很酷的产品。

These and one of them is in color, just gorgeous color writing. A whole array of four k cameras with Ring Yeah. Including doorbells Four k. Which is bringing that clarity with quite cool. Just these are cool your cool products.

还有一些新的Fire TV，其实就是电视机，从40英寸起。

And some new Fire TVs, just quite literally TVs from, you know, 40 inches

哦，哇。

Oh, wow.

最大到75英寸，包括你所说的Fire TV Stick。

Up to 75 inches, including, you know, what you call Fire TV Stick.

是的。你对Alexa有什么看法？我的意思是，我们一直在关注，比如Siri很久以前就推出了，感觉就像是激活了Siri。感觉他们现在正处于这样一个阶段：之前的系统更像是业务逻辑，有一堆不同的路径，Siri应用可以沿着这些路径运行，而现在他们进入了LLM时代，完全进入了AI生成式AI时代。也许这需要重新思考、重建或建立合作伙伴关系。

Yeah. How are you thinking about Alexa? I mean, we've been following like, Siri launched a long time ago, and it feels just activated Siri. And it feels like they are at a point where the current the the previous system was very much like business logic with a bunch of different routes that the that the Siri app could go down, and now they're kind of in the LLM era, in the completely, AI generative AI era. And maybe that necessitates a rethink or a rebuild or a partnership.

亚马逊是如何应对AI助手架构可能正在改变这一观点的？

How has Amazon grappled with this idea that the architecture of what an AI assistant is might be changing?

是的。可以从几个角度来思考这个问题。首先，我认为你提出的观点是，你知道，以前这类系统大多是确定性的。是的。如果你说对了话，基本上就能得到正确答案。

Yeah. If you there's a couple ways to think about it. First, I think the way you framed it is, you know, previously, a lot of this was deterministic. Yeah. If you said the right thing, you kinda got the right answer.

或者，可以说是问答式的点对点模式，问题、答案、问题、答案。但现在这种情况正在发生根本性转变，因为当大型语言模型介入时，你必须重新架构整个问题。你从这种确定性模式转变了——以前只要说对话就能得到正确答案。现在感觉有点受限，你知道。这就像是对之前某些代理系统说话时需要采用不同的方式。

Or, you know, call it point and shoot where question, answer, question, answer. It's it's flipping on its head now because when LLMs come into play, you have to rearchitect the whole problem. You move from this deterministic model where if you say the right thing, you get the right answer. It kinda feels limited, you know. It's like a different type of speak that you have towards one of some of these what in the previous agents, if you will.

但现在你进入了这个对话式世界，你可以问任何问题，拥有无限知识，特别是对Alexa来说，你会发现有个性因素在起作用。有同理心在起作用。所以你不仅获得了这种个性化体验，还获得了记忆功能。你和她交谈得越多，她就越了解你。所以你从这种可以说是即问即答的模式，转变为了深入理解的对话，并最终真正完成一些事情。

But now you're in this conversational world, anything you ask, unlimited knowledge, specifically to Alexa, you you you found you there's personality that comes into play. There's empathy that comes into play. There's so you get this personalization, but you also get memory. And so the more you talk to her, the more it knows about you. So you've gone from this, what you would say, just give and take moment to conversation and depth understanding and then actually getting some stuff done.

不过它的强大之处在于，你不仅拥有所有这些功能，还有这种环绕你的环境感知能力。如果你连接了几个Alexa设备，它们都会更新到新的Alexa Plus AI基准。嗯。现在你可以随时随地对话，询问任何你想问的事情，把手机放在口袋里，在那些关键时刻，你不需要再去那样做研究。只需要进行对话。

The power of it though, not only do you have all that, you have this kind of ambient world around you. If you have a few Alexa devices plugged in, they'll all update with the new Alexa Plus AI baseline. Mhmm. And you and you and now you can just have a conversation from anywhere, ask anything you want, keep your phone in your pocket, and those, like, critical moments where you don't have to go do the research that way. Just have the conversation.

这相当强大。

It's pretty powerful.

你认为是否存在风险或某种紧张关系，比如在Alexa开发者、应用商店之间？我知道有些人开发了专门做特定事情的应用程序。但如果你在上面部署一个更通用的LLM系统，这实际上可能会形成竞争。你某种程度上是在与你的客户竞争。你们是如何向开发者社区传达这一点的？

Do you think there's a risk of or or some sort of tension between, like, the Alexa developers, the App Store, the I I I know some folks who have built specific apps that do specific things. But if you drop a more generalized LLM system on top of it, that might actually be competitive. You're kind of competing with your customers. Like, how have you messaged to the developer community around that?

开发者们现在对此非常兴奋，对吧

Developers are pretty pumped about it right

嗯哼。

now. Mhmm.

因为，你知道，这些技能我们仍然称之为，基本上你是在调用这些路由到某个技能的API。所以虽然它不是确定性的即点即用，但仍然可以将其视为系统中的不同代理，你可以调用它们。所以你看到这些开发者编写应用程序，嗯哼。其中LLM做出决策，比如产品Alexa进行协调并说，你正在尝试叫车，让我来调用Uber。

Because, you know, these skills we're still calling, you know, these basically, you're calling these APIs that are routing to a skill. So while it's not deterministic point and shoot, it still is kind of think of it as different agents on the system that you can that you can call on. So you you're you're seeing these developers write apps Mhmm. Where the LLM makes, you know, the product Alexa makes the decision, kind of orchestrates and says that you're you're trying to order a car. Let me we'll call Uber.

是的。然后Uber负责安排车辆。是的。我认为当你看到这个例子，或者你想从Grubhub之类的地方订餐，答案就是，你只需提出要食物，然后，你知道，那个服务就会行动并送达。我认为这有一个根本的好处，对开发者来说实际上更好，因为如果你在服务客户，他们坐在这里，试图告诉你他们想要什么。

Yeah. And then Uber makes the car happen. Yeah. And I think as you as you see that as an example or you wanted to order food from Grubhub or whatever the answer was, you just ask for the food and and then, you know, that service makes its move and, you know, delivers it. I think there's this fundamental of it's actually better for developers because as you if you're serving your customer, they're sitting here, they're trying to tell you what they want.

他们不再需要那么具体了。他们的请求几乎可以很泛泛。

They don't have to be as explicit anymore. They can almost be general in their asks.

是的。

Yeah.

在这一点上，如果它调用一个应用程序、技能或我们称之为附加组件的东西，开发者们现在非常兴奋。我们现在有一个很长的开发者名单。我们可能是目前连接到任何LLM的最大代理群体，嗯哼。所以这是一个相当罕见的情况，因为其他地方还没有这样做。

And at that point, if it's calling on an app or a skill or an add on as we call it, developers are pretty pumped about it right now. We have a pretty long list of developers right now. We're probably the largest group of, if you will, agents attached to any LLM out there. Mhmm. And so you're it is a pretty rare situation as it's just not being done anywhere else.

你对智能眼镜和可穿戴设备的未来怎么看？亚马逊已经推出了很多智能设备，从你提到的Kindle电视到Alexa。这感觉是件显而易见的事。你们已经宣布或已经推出了多款产品，但你认为这个方向会如何发展？你最喜欢哪些产品形态？

How do you think about the future of smart glasses and wearables? Amazon's done so many smart devices from the TV you mentioned in the Kindle to the Alexa. It feels like an obvious thing. You you have a number of products that have been announced or already live, but where do you think that goes? What formats do you like most?

那么你在哪些领域看到最多的实际应用呢？

And where are you seeing the most adoption really?

我们看到使用量大幅增长，Alexa Plus在我们现有的设备阵容中表现突出。我们有眼镜，是的，就是人们佩戴的眼镜。我们看到使用量在上升，耳机也在增长。因为当你随身携带一个智能助手时，它的力量相当强大。

So we see a ton we see our usage kind of just spiking with Alexa Plus on the current device lineup that we have today. We have frames Yeah. Frames that people wear. We see the usage go up, the buds are going up. Because when you take an agent that's smart with you, it's quite powerful.

所以，我们已经看到它发生在不同的形态因素中。我的意思是，这取决于用户的偏好。我要说的是，你提到了一个我认为值得注意的观点。我认为有很多工作正在转向所谓的AI设备。

So you we already see it happening in different form factors. I mean, it's down to user preference. I will say this. I mean, you're hitting a point that I think is just worth noting. You're I think there are plenty of jobs moving to what you would call AI devices.

这只是有点...如果你想想科技不同部分随时间的演变，让我带你回顾一下。当我大约14年前开发时，我正在做笔记本电脑，有人来找我说，你在做什么？笔记本电脑已经死了。我现在看不到你们，但假设你们面前有笔记本电脑。

It's just kinda And if you think about kind of the the evolution of different portions of tech over time let me take you back. When I was developing, I think, fourteen years ago, I would I'd I was making a laptop and somebody had come to me and said, what what are you doing? The laptop is dead. I can't see you guys right now. But assuming you have laptops in front of me.

我们有，我们确实有。我们已经死了。谢谢。

We we do. We are dead. Thank you.

是的，不是...是的。

Yeah. Not Yeah.

就像，大家花一分钟想想这个。那可是十四年前的事了，伙计们。真的，我当时我们正在创造这个产品。我们要做这台笔记本电脑。它会创造，我有一个愿景，而你们都在追随它。

Like, just think about that for a minute. Like, that was fourteen years ago, guys. Like, literally, I was we were creating this product. We're gonna make this laptop. It's gonna create and I have this vision and you're getting all after it.

你们会说，这太美了。看看它在每个细节上能做什么。然后你们去做了，但第一个反应是，为什么要造笔记本电脑？那东西已经过时了。是啊。

You're like, it's this beautiful. Look what it can do in every detail. And you're going and the first thing is like, why would you build a laptop? That thing is dead. Yeah.

手机已经取代了它。十四年前笔记本电脑就完蛋了。

The phone has replaced it. It's over for the laptop fourteen years ago.

是啊。这些东西确实持久。真的吗？很久了。

Yeah. These things stick Really? Long time.

真正发生的是，你知道，并不是笔记本电脑在消亡，而是它在自己擅长的领域变得更加强大。

What really was happening, you know, wasn't that the laptop was dying, it was just getting stronger at what it did great.

是啊。

Yeah.

乔布斯某种程度上转向了手机，手机上对你重要的东西变得活跃起来，通讯，然后，你知道，社交媒体的出现，可能还有购物，那些在手机上更方便的事情。但那些重要的东西，比如现在你的笔记或其他对笔记本电脑重要的东西，它们只是变得更强大。我认为我们现在正处于同样的时代。我认为AI设备正在到来。是啊。

Jobs had kind of moved to the phone and the things that mattered to you on a phone came to life, communications, then, you know, the advent of social media, shopping probably, like things that were easier there. But the things that mattered, like right now, your notes or your other things that mattered for the laptop, they just got stronger. I think we're kind of in that same era. I think AI devices are coming down the pipe. Yeah.

它们在某些方面会非常出色。正如你所说，很多设备将是便携式的，无论是眼镜、手腕上的设备，还是口袋里的东西。你知道，我不谈论实验室里的东西，但可以肯定的是，你会想象随着任务的变化，会出现更重要的形态因素。我认为这只会强化现有的形态因素，就像笔记本电脑变得更强大一样，在我看来，因为当手机出现时，你在笔记本电脑上的任务处理能力变得更好，就像笔记本电脑出现时，台式机在游戏和开发方面变得更强大。

They're gonna be great for certain things. And to your point, a lot of them will be on the go, whether it's glasses or something on your wrist or, you know, something in your pocket. You know, I don't talk about, you know, what's in the lab, but for sure, you would you would imagine that as the jobs move, there's gonna be form factors that evolve that are gonna be more important. I think it just strengthens the form factors that exist today, just like the laptop got stronger, in my opinion because when the phone came, you know, you got better at the jobs on your laptop just like when the laptop came. The desktop got stronger for gaming and developers.

你只是看到了这种演变，这些形态因素并不会消失，它们会在本职工作上变得更好，然后新的形态因素出现。我认为这正在发生。

And you just see that evolution where the it's not that these form factors go away, they get better at what they are meant to be and then new form factors show up. Think that's happening right now.

你如何看待单个模型的个性和品牌？我觉得有一个世界，我会喜欢一个Alexa，我可以直接说它是Claude。我在和一个Alexa硬件设备交谈，但我喜欢Claude的个性和功能。所以我希望默认运行那个软件，就像我可以选择默认浏览器一样。我有一个MacBook，但我用Chrome作为默认浏览器。

How do you think about the personality and brand of individual models? I feel like there's a world where I would love an Alexa that I can just say it's Claude. And I'm talking to an Alexa hardware device, but I like the personality and functionality of Claude. And so I want to just kind of run that piece of software by default in the same way that I can select, like, a default browser. I have a MacBook, but I run Chrome as my default browser.

有没有可能实现这一点？

Is there a world where I could be able to do that?

这是个有趣的想法。目前Alexa不是这样工作的。是的。但这很有趣。我不认为，比如，如果我想想客户，你知道，那种我不知道该怎么称呼，但只是基准客户。

It's an interesting thought. It's not how it works for Alexa right now. Yeah. But it's pretty it's it's interesting. Don't think for, like, if I if I think about the customer that, you know, kind of a I don't know what you would call it, but just the bay the the baseline customer.

我想说的是

What I'm trying to

我是非常在线的。我有我的看法关于

do is I'm extremely online. I have I have opinions about

是的。就像，个人你是真实的，我甚至不知道该告诉你什么，就像

Yeah. Like, individual you're real I don't even know what to tell you, like

但很多人不在乎。没错。

But many people don't care. That's right.

你在问题中提到了极端情况，但我是这么想的。是的。我认为，Alexa基本上是模型无关的，就像，什么才是适合这项工作的正确模型？当然。而且，我们当然有专门为Alexa微调的模型，专门用来完成你需要的任务。

You hit the extreme in the question, but here's how I think about it. Yeah. I think, like, Alexa is basically model agnostic, like, what's the right model for the job? Sure. And, of course, we have fine tuned models specifically that are built specifically for Alexa to do the jobs that you're looking for.

我认为，对我们的客户来说最重要的是它能为你个性化。嗯。所以我越多地和它谈论我喜欢的事情，它就越多地理解我。我想你今天在其他产品中也看到了这一点。但当你进行自然对话或要求下一件事时，你不必从头再来。

The thing that matters most, I think, to our customers is that it personalizes to you. Mhmm. And so the more I talk to it about the things that I love, the more it understands me. I think you see that today in other products. But then as you're in natural conversations or you're asking for the next thing, you don't have to go all the way.

Alexa可以主动为

Alexa can be proactive for

你服务。嗯。

you. Mhmm.

我我给你举个简单的例子。也许有点傻，但很简单，就是很实用。如果我问Alexa，我有一个摄像头。我现在有一个Ring摄像头。你你把它放在我的狗吃饭的地方，在狗碗那里。

I I give you a simple example. Maybe it's a silly one, it is simple, just kind of being practical. If I ask Alexa I have a camera. I have a Ring camera right now. You you have it on the in the where my dogs eat in the dog bowls.

如果我现在问Alexa，我可以直接问它。这是个相当酷的功能。虽然简单来说就是人工智能，但还挺有趣的。今天有人喂狗了吗？这在我家是个经典问题。

And if I axe I can ask Alexa now. It's a pretty cool feature. It's just AI in its, you know, simple sense, although kind of of fun. Did anyone feed the dogs today? That's a classic question at my house.

我每天晚上回家都会遇到麻烦。'你喂狗了吗？'而我刚进门。你知道，就是那种情况。然后我就可以转身问Alexa，Alexa会说：'没有，今天没人喂狗。'

I get in trouble every night when I get home. Have you fed the dogs? And I'm like, just walked in. You know, that sort of thing. Like, and and then there's that moment where I can just turn and ask Alexa, Alexa says, no, nobody fed the dogs today.

所以它们一直在禁食。这真的很酷，因为它知道没人喂狗。我们使用Ring视频描述功能。我们知道有狗。我们知道有人喂了食物或者给狗喂了食。

So it's They've been fasting. It's just very cool because it knows nobody fed the dogs. You know, we use Ring video descriptions. We understand there's a dog. We understand that somebody fed the food and or fed the dog the food.

所以如果狗没被喂食，我可以询问。但三四天后，它可以主动提醒说：'嘿，今天没人喂狗。'然后你就知道了。对客户来说，是哪个大语言模型或智能体创建了这个场景并不重要。重要的是它具有个性化。

So if the dog's not fed, I can ask. But in three days or four days later, it can be proactive and just say, hey, nobody's fed the dogs today. And, like, you just take that. It didn't matter which LLMs or which agents kinda created that scenario to the customer. What mattered was it had the personalization.

它知道我在乎这件事，然后主动给我反馈这个信息。我认为这就是保持技术中立性的本质。嗯。但我们确实使用不同的模型。是的。

It knows I care about it, and then kinda proactively gave me that information back. I think it's that's the essence of just staying agnostic to it. Mhmm. But we do use different models. Yeah.

而且，你知道

And, you know

您如何看待广告应该如何整合到人工智能助手当中？因为现在有很多争论。人们会向助手寻求产品推荐和购买建议。我认为有一个非常现实且重要的讨论：这个像值得信赖的顾问一样的助手，应该在什么时候、以什么方式展示基于广告的推荐？什么时候是合适的？

How are you thinking about the way that advertising should integrate into AI assistance? Because there's a lot of debate happening right now. People are going to assistants for product recommendations and advice on things to buy. And I think there's a very real and important debate around when should this assistant, that's like a trusted adviser, how should ad, you know, ad based recommendations, you know, show up? When is it appropriate?

它应该在什么时候推荐呢？就像，你知道，通常一个类别里没有客观上的最佳产品，因为每个消费者都不同。但你对此有什么框架吗？因为显然亚马逊是非常支持广告的。

When should it just recommend? Like, you know, there's no there's usually no objectively best product in a category because every consumer is different. But what's your framework for that? Because obviously You Amazon's very pro ad.

你认为应该在什么时候推荐，或者应该推荐什么？嗯，你的提法是对的。我认为就像任何助手一样，它越了解你，推荐就越好。所以我认为广告确实有效的时候，是当你真正帮助客户找到他们想要的东西。这确实归结为个性化方面和一些历史记录。

When do you think you should recommend or what should it recommend? You well, the way you framed it is right. I think the like any assistant, the better it knows you, the better the recommendations can be. And so where I get where I think ads do work is when you're actually helping the customer, you know, find what they're looking for. And then it does comes down to the that personalization aspect and some of that history.

如果你经常使用某样东西，并且有一个广告实际上能帮助你，如果它被推荐给你，那我会感到兴奋，因为我认为这是一种整体的方法，你知道，那种循环。当它只是随机的，你会想，这是什么？我真的觉得那很痛苦。就是在那时候变得痛苦。而且我认为

If you if there's something you use quite often and you're having you know, and there's an ad that actually helps you if it's raised to you, that's when I get excited because I think it's like a holistic approach to, you know, that kind of circle. When it's just random and you're like, what is this? I I really that's painful. It's just where it gets painful. And I I think

人们真的不介意去你的Alexa，然后亚马逊从发生的商业交易中抽成，因为你看到亚马逊的卡车，看到配送网络，看到东西送达的速度有多快。所以那里有明显的价值渗透。所以我认为大多数客户都乐意让亚马逊基本上收取一种类似联盟费或抽成。是的。问题在于如何保持中立。是的。

People really have no problem with the idea of an going to your Alexa and then Amazon taking a cut of the commerce transaction that happens because you see the Amazon truck and you see the delivery network and you see how fast things arrive. And so there's clear value pervade there. And so I think most customers are are happy to have Amazon take a a essentially, like an affiliate fee almost on that or cut. The yeah. The issue is is how how agnostic Yeah.

有没有一种情况，Alexa会像平台上的联盟一样，有效地成为一个联盟，就像它是一个外部流量来源？因为我想要的是，基本上说，嘿，Alexa，抱歉家里的任何人，如果你们在听这个，我会触发你们的Alexa设备。但是，嘿，Alexa，我想买——我用纸巾架的例子，因为我有一个，而且我希望它来自一个超过50年历史的品牌。就像那样，那是我想要从信誉良好的品牌购买的产品，该品牌一直在运营，而不是仅仅从海外进口最便宜的产品。

Is there a world where where Alexa would be getting an an like, effectively be an affiliate on the platform like it was an external, like, you know, source that was sending traffic? Because what I would want is to arrive on is basically say, hey, Alexa, and sorry to anybody at home. I'm setting off your your Alexa device if you're listening to this. But, hey, Alexa, I wanna buy I I use the example of a paper towel holder because I I've had a and I want it to be from a brand that is over 50 years old. And like that is the kind of per that's the kind of product I wanna buy from a reputable brand that has been operating and that isn't just, you know, importing the cheapest possible product from from overseas.

所以那对我来说是一个非常宝贵的推荐。作为消费者，我不在乎Alexa是否抽成，但那不同于我只是在搜索栏里搜索。

And so that is, like, a very valuable recommendation to me. And as a consumer, I don't care if Alexa takes a cut, but that's different than me just searching in in, you know, in a search bar.

我同意。我的意思是，我不太清楚你所说的Alexa抽成具体指什么，但在那种情况下，我希望看到的是，而且我认为现在就是这样，因为我们如此对话化，我们可以为你做那个搜索。我们可以把它找出来。我们也可以只是查看你的购物历史，如果你选择启用的话，然后说，看这里。这是你一直在买的纸巾尺寸。

I agree. I mean, I don't know about I'm not exactly sure what you mean by Alexa taking a cut there, but what I would like to see happen in that scenario, and I think is what happens today, is because we're so conversational, we can do that search for you. We can bring it up. We can also just look at your shopping history if you choose to enable that and then say, look here. This is the size of paper towels you've been buying.

从根本上说，你要明白，如果不仅仅停留在'我听到了你的问题并给出答案'这个层面，而是为你做一些逻辑梳理，提供几个选项供选择。我认为这才是建立情感连接点的时刻。我知道这听起来可能有点傻——我们讨论的是纸巾这种小事。但当你间接地提出需求，自己也不完全清楚具体想要什么，而系统能给出接近的解决方案并展开对话，想想任何助手类模式都是这样的。

And fundamentally, know, if you just go to that next level of beyond just I'm gonna I heard what you said and here's the answer, maybe just do a little bit of logic flow for you and bring back a few options. I think that's when it does get, like, these are those emotional connection points. I know it sounds silly. We're talking about paper towels. But if you're asking for something indirectly and you don't know exactly what you're pointing and shooting for and you get something that's close and you can have a conversation, just think about any assistant kind of mode.

就像是在和某人交谈。我只是要进行这样的对话，我们一起 brainstorming。你已经开始在Alexa上看到这种趋势了。重点在于不断精炼你真正想要的东西。

I'm just talking to somebody. I'm just gonna have this conversation, and we're gonna brainstorm it together. You're you're starting to see that happen on Alexa. What? It's more so about refining what it is you're getting after.

在大多数方面你都能看到这种趋势。

You can see that in kind of most aspects.

在Kindle方面，你们是否收到用户希望在产品中集成更多AI功能的需求？是的。我能想象——当我拿起一本很久没读的书时，如果能说'你能帮我总结前三章吗'就太好了。

On the on the Kindle side, are you getting, like, feature requests from users that want more AI integrated into products? Yeah. I can imagine I That's I pick up a book I haven't read in a while, and I and it'd be nice to be able to say, can you summarize the first three chapters?

用一句话总结整本书。

Summarize the entire book in one sentence.

我说完了。好吧。我说完了。但我们正在

I'm done. Okay. I'm done. But We're getting

大量接收这类需求。你知道吗，这很好——我来回答你：我们有个叫'Books So Far'的功能，让我给你介绍一下。这正是你刚才要求的功能。请注意，这是基于AI实现的。

bunch of it. You know, it's this is a fine I'm gonna answer your we have a thing called Books So Far, so I'm gonna tell you about that. That was the exact feature you just asked for. Please. And and it's AI based.

这基本上就是书籍版的‘继续观看’功能。嗯哼，你明白我的意思吗？就像你看了某个剧的前三集，它就会给你推送更新，或者不管是什么剧，你都能看到剧情回顾然后接着看下一集。我经常遇到这种情况，因为我电视看得不够多。

And it's basically continue watching for books. Mhmm. Do you know what mean? Like, you watch the first three episodes of whatever it is, you know, brings a power and you're like, just keep me up to date or, you know, whatever it is, what show it is, and you just see the recap and then you go into your the next episode. This happens to me all the time because I don't watch enough TV.

是的。

Yeah.

我们现在有书籍功能了，就是说，嘿，给我总结一下这本书我看到哪儿了，提醒我一下。你还可以向书本提问。是的，任何问题都可以。你可以问Kendall关于这本书的任何事情。这是个很大的功能需求，比如我就想讨论角色，想知道这里到底发生了什么。

We have books so far, is that, you know, hey, summarize how far I am to this book, remind me. You also can ask the book Yeah. Anything you You can ask Kendall anything you want about the book. And that was a big feature request, like, I just wanna talk about the character. I wanna know more about what really happened here.

现在这个功能实现了。所以这就是AI增强的地方。但你必须非常小心，因为Kindle有一种纯粹的品质，是的，顾客很喜欢。这一点对我来说特别重要。

And now that's happening. So that's where AI enhances it. But you have to be really careful because there's something pure about Kindle Yeah. That customers love. And this is, like, near and dear to my heart.

你必须把握一个非常微妙的界限。我们确实有，比如刚发布的Kindle Scribe，就有针对所有笔记的AI搜索功能。所以如果你有几百页的笔记，现在你只需要搜索，然后可以说，总结所有我写的关于约翰的笔记之类的。然后你会得到所有总结，它会为你做一个智能摘要，然后你可以把它们整理进去。所有这些都能做到。

You you have to be a super fine line. We do have like, on the Kindle scribes that we just launched, we do have AI search for any of your notes. So if you have hundreds of pages of notes now, you just search for or and then then you can say, summarize all the notes that I wrote about John or whatever. And, you know, you'd get them all summarized and it would do a smart summary for you and then you can put them in. You can do all that.

但这里有一个我们非常谨慎的风险。Kindle就像一个圣殿。你沉浸其中，就想完全投入进去。我们的用户，他们热爱他们的书。你知道他们看书时讨厌什么吗？

But here's here's the risk that we're very cautious of. Like, Kindle's like a sanctuary. It's like this you you get into it and then you just wanna disappear into it. Our users, like, they love their books. You know what they hate when they're reading books?

是分心的事物。

It's distractions.

是的。

Yeah.

我甚至不会轻易用这个词，但情况就像是，如果给你弹个广告、放个链接，或者用 Jordy 的名义发条短信渗透进来，你会觉得，天啊，我只是想读会儿书，哥们。这是我唯一能消失的地方。这很疯狂，因为读者无处不在冒出来，新读者。你知道吗，现在我们有 60% 的新购买来自完全新的读者。

I don't even I don't use that word lightly, but it is like, if you popped them an ad or put a link up or I infiltrated you with a text coming in, if you will, from Jordy, you'd be like, goddamn. I'm just trying to read, man. This is my one place to disappear. And it's crazy because readers are popping up everywhere, new readers. I think, like, 60% of our new purchases are from totally new readers right now, just so you know.

所以现在有点激增，因为这个类别又在增长。关键在于保持那种圣殿的感觉，当你拿起它阅读时，那就是它的意义。除了书之外，你在里面加入的技术越多，你就得越小心。我想我知道那条线在哪里。团队也知道那条线在哪里。

So there's a bit of a spike happening because it's a growing category again. The the trick is to hold that as, like, that sanctuary where when you do pick up to read, that's what it becomes. The more tech you put in there beyond a book, you know, you you have to be careful. We I think I know where that line is. The team knows where that line is.

我们有一个很棒的 Kindle 团队。他们已经做了很多很多年了。他们非常了解客户。但就像你说的，客户已经问过，我能不能得到我们称之为“书签”或类似“继续观看”的功能？所以我们创建了那个，或者说是摘要功能。

We have an amazing Kindle team. They've been doing it for, you know, years and years. This is an and they are they they understand the customer so well. But to your point, the customer has already asked, can I can I get to what we call book so far or continue watching is the best way to say it in in analogous? And so we created that or summarization.

我们有那个功能。而且在你的写作中，可以搜索你的笔记。但我们非常谨慎，不过度技术化，如果这说得通的话。你得小心，因为归根结底，它是一本书或一个你想在上面书写的笔记本。你可以有很多其他设备来让你分心，但这个是我们有点珍视并相当保护的。

We have that. And also in your writing, can search your notes. But we're being very selective not to over techify it, if that makes sense. You gotta be careful because at the end of the day, it is a book or it is a notepad that you wanna write on. And and you you can have plenty of other devices to let your distractions kick in, but this is the one that we kinda hold coveted and and protect quite a bit.

还有最后一个问题。但这可能是个技术问题，关于 Kindle 和 AI。我认为 Kindle，我不想说它性能不足，但它设计是为了长电池寿命，你知道，不是最疯狂的屏幕。它应该是你可以扔进背包，旅行时拿出来，知道它还有电的东西，对吧？

Have one last question. But but maybe this is a technical one, but on on Kindle and AI Yeah. I think of Kindle as, I don't wanna say underpowered, but, like, it's it's designed for long battery life, you know, not the craziest screen. It's supposed to be something you can, you know, throw in your backpack and pull out on a trip and know that it's still charged. Right?

所以我猜想

So I imagine that there's

你应该去当Kindle的产品经理。

You should be a product manager for Kindle.

好的。是的。所以

Okay. Yeah. And so

保持这个思路。如果你坚持这个框架，老兄，是的。你太完美了。就像，你只需要虔诚地坚持那些观点。

hold that line. If you hold if you hold that frame, man Yeah. You're perfect. Like, you just just religiously hold on to those statements.

太棒了。那么，所以我想AI工作负载需要在云端完成，你们不考虑设备端AI，这通常是正确的权衡，因为摘要可以异步处理，或者可以排队，通过WiFi网络或4G来回传输。我只是有点好奇，比如，对于Kindle未来来说，中位数连接性与设备端计算之间的权衡大概是怎样的。

Fantastic. And and and so I imagine that the the AI workloads need to be done in the cloud, and you're not thinking about on device AI, and that's generally the right trade off because the summaries can be processed asynchronously or or they can be queued up and and kind of gone back and forth over the over the WiFi network or or for four g. I I'm just kind of interested in, like, what the what the median connectivity versus, on device compute trade off looks like for Kindle going forward.

是的。这很棘手。我们现在正如你所说，在边缘端做了很多工作，也就是设备端。当然。我认为必须这样做，因为它可以提升速度，有时候确实如此。

Yeah. It's tricky. We're doing we're doing a lot of work on the edge right now as you you put it on device. Sure. I think you have to because it just it it can envelope speed and sometimes Yeah.

不过，基本上，我们透明地为客户服务，是的，无论哪种方式合适。但我们确实在努力，当时机成熟时，当时机适合使用设备端时，比如，如果你想减少书写延迟并进行一些美化，那就是一个例子。是的。是的。

Though, you basically, we serve the customer transparently, yeah, wherever it's right. But we are doing work to when it's time and when it's time to use on device, so let's call it, like, if you're trying to reduce latency on writing and you're doing some beautification. That's an example. Yeah. Yeah.

你会想在设备端完成。完全正确。如果你想，你知道，处理书籍到目前为止，像是云端的事情。我的意思是，如果你开始在iPhone上阅读，这是一个很好的例子。你在iPhone上开始阅读，然后转到你的Kindle，拿起来继续。

You're gonna wanna do it on device. Totally. If you want to, you know, do books so far, like, it's a cloud based thing. Mean, what if you start reading on your is a good example. You start reading on your iPhone, you came over to your Kindle, and you picked it up.

你想要的是云端体验。这不是你希望在设备上实现的。所以，这中间有取舍，但我们两者都做。这是个好问题。顺便说一句，我认为今天我们讨论的每一款设备都是如此。

You you want that to be a cloud based experience. That's not something you wanted on device. So, like, there's just there's trade offs, but we do both. It's a good question. I think that's true, by the way, across every single device we talked about today.

是的，完全同意。

Yeah. Totally.

比如，如果我确实想要，你知道，取决于我们是出于安全、速度、成本还是其他任何原因，边缘工作负载在何处更重要，你总是在尝试为客户找到正确的平衡点。顺便说一句，当你说到它，我能展示

Like, where is the edge workload more important if I did wanna you know, depending on if we were doing whether whether it was for security or speed, your or cost, whatever it might have been, you're always trying to find the right balance for the customer. By the way, when you say it, I'd can I show

这个吗？

you this?

是的，请展示。

Yeah, please.

你可能无法在屏幕上看到它。

You may not be able to see it on the screen.

这个会

Will that this

新的颜色，肯德尔。你能看到吗？这个

the new color, Kendall. Can you see it? The

屏幕哇。

screen's Wow.

是不是很酷？那个颜色是

Isn't that rad? Is that is that color

看它有多薄。这东西只有5.4毫米薄。它很有趣，而且电池续航时间和我之前说的一样，同样的延迟，同样的速度，我现在一边和你们说话一边用它记笔记，但它现在就像个性能猛兽。处理器性能将翻倍。

look how thin it is. Like, the thing is 5.4 millimeters thin. Like, it's It's so fun and it has the same battery life I've talked about, same latency, same speed and I'm I have it here taking notes as I'm talking to you guys but it it's like a it's a powerhouse now. The processor there is gonna double the power.

那是，那是，像是电子墨水屏吗？就像在户外也能用，是的。

Is that, is that, like, e ink screen? Like, it works outside and Yeah.

是电子墨水屏。

It's e ink.

所以当你看着它时，感觉非常柔和，对吧？它有前光，而且前光均匀分布在整个屏幕上，所以不会刺眼。

So it almost feels like when you look at it, it's so soft on the eyes. Right? It has a front light and the front light Sure. Basically across the screen so it's not blasting you.

是的。

Yeah.

是的。所以它依然保持着那种细腻感。你在看它时不会感到疲劳。无论明暗环境下都能使用。这块屏幕没有任何损失。

Yeah. So it still has that subtlety. You don't feel as tired as you're looking at it. You can do it in the dark and the light. There's nothing that you lose this screen.

这是块磨砂屏。非常适合书写。

It's a matte screen. It's perfect for writing.

哇。

Wow.

这样你就能获得所有这些特性。精妙之处在于色彩。所以总会有这种较量，但它不是LCD那种美化效果，你知道的，像OLED等等，我是说，这有一种美丽而微妙的感觉，几乎像是屏幕上的彩色铅笔。嗯。所以即使你用彩色书写，感觉也是如此。

And you're and so you get you get all those pieces. The refinement was color. So there's always this battle, like but it's not an LCD with the beautification of, you know, OLED and so forth, I mean, there's something beautiful and subtle about this, almost like colored pencils on the screen. Mhmm. And so even when you're writing in color, feels that way.

这相当...这很漂亮

It's quite it's it's pretty

好吧，我们得搞一台。得想办法越狱，侧载Sora的垃圾内容推送。我要让这东西充满脑残内容。

Well, we gotta get one. We gotta figure out how to jailbreak it, sideload the Sora slob feed. I want brain rot on that thing.

别跟我说那些，老兄。不想听

Don't tell me any of that stuff, man. Don't wanna

最后问一句，你觉得亚马逊的AI机遇是否被普遍低估了？我的意思是，感觉有太多服务领域被严重低估了。零售机遇、Anthropic合作。你们已经有一套人们使用并喜爱的设备进入了千家万户。感觉在很多方面，其他公司，你知道的

Last thing, do you think Amazon's AI opportunity is broadly underappreciated? I mean, it feels like there's so much service area So underappreciated. Retail opportunity, anthropic. You guys already have a suite of devices that people use and love that are in people's homes. It feels like in many ways other, you know

你们还有数据中心呢。

You guys also have data centers too.

是有那么几个。

A few of those.

确实有几个。

Few of them.

你们有几排机柜呢。

You got a few racks.

数据中心可是他们发明的。

They invented the data center.

当然。是的。当然。我的意思是，我不知道还能怎么说，

For sure. Yeah. For sure. I mean, I don't know how else to say it,

就像是的。

like Yeah.

毫无疑问，你知道，亚马逊的每一个部分每一个部分

Hands down, you know, you're every part every part of Amazon

是的。

Yeah.

都深深扎根于推动人工智能的发展。无论是Bedrock、AWS、CloudWise正在发生的事情，我们运行的所有模型，我们提供给客户的所有服务，一直到商店以及它们从根本上改善您购物体验的努力，Prime Video以及它们的举措。我们的Fire TV有一个场景跳转功能。你们知道吗？非常酷。

Is, like, entrenched in driving through AI. Whether it's Bedrock, AWS, what's happening, like, CloudWise, all the models we run, we make available to our customers, all the way through stores and what they're doing fundamentally to help your shopping experience better, Prime Video and what they're doing. We have a jump to scene with Fire TV. Did you guys know that? It's very cool.

嗯，那是

Well, that's

非常棒

very Great

电影场景。

movie scene.

我的意思是，我认为世界上最大的机器人部署也是弹药。

I mean, also the largest deployment of of robots, I believe, in the world is ammo.

没错。哦，不过我不确定这是否准确。

Correct. Oh, well, I don't know if that's I don't know if that's accurate.

也许在美国。我看到一些图表显示你们已经超过一百万台机器人了。

Maybe in The US. I I saw some chart where you guys were were passing a million robots.

是的。规模相当庞大。然后，你知道，由...控制，还有另一面。现在所有的Alexa设备都将由Alexa Plus驱动。这就像是核心基础

Yeah. It's pretty massive. And then, like, you know, controlled by and then you have the other side of it. All the Alexa devices now are about to be powered by Alexa Plus. This is, like, core foundation

当然。

Sure.

的，就像是最实用意义上的消费级人工智能。

Of, like, consumer AI in its most practical sense.

是的。是的。

Yeah. Yeah.

你现在所处的这个位置是，你知道，我和家人一起在餐桌旁，不是有人掏出手机查答案或研究什么，而是直接和Alexa对话，你能获得基于LLM产品所期望的全部深度，但它非常实用且随时可用。所以，我认为整个范围你是对的。它目前确实被低估了。我不知道在另一面如何恰当地描述它而不显得太，你知道，但我确实认为这个

You're you're in this place now where, you know, I'm at the dinner table with the family, and instead of somebody pulling out their phone to get the answer or research something, we just have a conversation with Alexa, you got the full depth of what you would have expected from an LLM based product, but it's just very practical and ready for you. And so, like, I think the whole gamut is you're right. It's it's quite underappreciated right now. I don't I don't know what the right way to frame it is on the other side without sounding too like, you know, but I I do think this I

我同意。所以我认为，而且我认为市场最终会意识到的。

I agree. So I think and I think I think the market will will will realize eventually.

我也这么认为。我认为，你知道，首要的是服务客户，服务客户，服务客户，并以正确的方式去做。我认为，你知道，这是我们的重点，我们会继续这样做。但你是对的。我认为它会得到注意、被看到、被理解。

So I do too. I think, you know, first thing, serve the customer, serve the customer, serve the customer and do it in the right way. I think, you know, that's our focus and we'll we'll we'll kinda continue to do that. But you're right. I think that it gets it'll get noticed, seen, understood.

我现在对此感到非常自豪。当你看到Alexa Plus与Ring的结合，Ring视频搜索或Ring故事描述，以及总结你的一天等功能时，它真的开始将AI提升到一个新的水平，因为它让人们觉得非常实用。就像，它就在那里。别误会，你也可以使用alexa.com来做人们通常与传统LLM关联的所有事情。

I'm pretty proud of it right now. And when you see what's coming with Alexa Plus with Ring, with Ring video search or Ring story descriptions and, like, summarizing your day, it really starts making it something it takes AI to the next level in my mind because it just makes it practical for people. Like, it just puts it right there. Don't get me wrong. You can actually use alexa.com too to do all the things that people are associating with traditional LLM Sure.

应用也很酷。但归根结底，有一整套全新的实用方式，我认为会让我们的客户感到非常愉快。

Is quite cool as well or the app. But at the end of the day, there's just a whole new set of practical ways to use it that I think are gonna be pretty delightful for our customers.

新的Scribe到底什么时候发布？我们聊天的时候我刚试着去买，是的。它显示即将推出。

When's the new Scribe actually releasing? I just tried to buy it while we were talking, and Yeah. It says coming soon.

我不能在这里给你具体日期，但确实是。所以你得加入等候名单。目前我们的需求比预期要高一些。我们正在生产中。所以你先登记上。

I I'm not allowed to give you the date on here, but it's yeah. You so you have to get on the you have to get on the wait list. Our demand is a little bit higher than we expected it to be right now. We're in we're in production. So you put it on.

你

You

你们没能做好需求规划吗？

guys couldn't get the demand planning right?

我想要一个 那是需求规划。

I want a It was a demand planning.

那是几天的事。面板。九个。假期前你就能拿到。你会有的

That's days. Panels. Nine. You'll have it before the holiday. You'll have in

两天。

two days.

只要是我们

As long as something we're

非常关心这件事。我会确保你们得到，我会确保你们得到

very about it. I'll make sure you get I'll I'll make sure you guys get

这太棒了。

it. It'd be awesome.

写下你的名字，我会确保的。

Put your name down, I'll make sure.

我们会的。我们会的。太棒了。非常感谢你加入我们。这真是太棒了。

We will. We will. Awesome. Well, thank you so much for joining us. This is great.

在所有进展上。

On all the progress.

是的。真的，真的是对这个生态系统的一次迷人游览。感谢你为我们详细解释。非常感谢。

Yeah. Really, really fascinating tour of the ecosystem. I appreciate you breaking it down for us. Thanks so much.

很愉快，各位。你们保重，等等。

Been fun, guys. You take Wait.

另外，我觉得很明显，现在说已经太晚了，但Panos是第一个——我想你是我们节目有史以来第一位来自亚马逊的嘉宾。

Also, I think obviously, it's too late now, but is Panos the first are you I think you're the first person from Amazon ever on our show.

我想我们已经有过

I think We've had over

今年上千次采访了。哎呀这该死的狗。好了。很高兴你能来。

thousand interviews this year. Here we the damn dog. We go. It's great to have you.

谢谢。

Thank you.

我们很快会再见的。

We'll we'll we'll see you soon.

回头见

We'll see

很快。干杯。祝你好运。我们有些突发新闻要报道。Sam Altman已经回复了他关于ChatGPT中情色内容的帖子。

you soon. Cheers. Have a good wagon. We have some breaking news that we gotta cover. Sam Altman has replied to his, his post about erotic content in ChatGPT.

他说，好吧。这条关于ChatGPT即将进行更改的推文在情色内容这一点上引起的反响比我预想的要大得多。

He says, okay. This tweet about upcoming changes to ChatGPT blew up on the erotica point much more than I thought it was going to.

你没想过你没想过发布情色内容会

You didn't think you didn't think posting e r o t I c a was gonna

有那么多其他词汇你可以用来，比如，你知道的，达到那个意思而不必说那个词。没错。我们我们说过。在我们整个讨论过程中，Grock，我们喜欢这是一个干净的节目。我们喜欢我们

There are there are so many other words that you could have used to kind of, like, you know, get there without saying that companion. Exactly. We we have said that. When the entire time we were discussing, Grock, we like that this is a clean show. We like that we

我从未我从未想在节目上说那个词

I've never I never wanted to say that

没错。所以当我们讨论这个时，我们说成人内容，成人浪漫伴侣。我们不会用其他词汇。但是，是的，他用了那个词。

word on the show. Exactly. So when we talk about this, we say adult content, adult romantic companion. We don't go to the other words. But, yes, he went there.

他说了情色。他在他的X应用中打出来，发送了推文，他说这只是我们允许成人用户更多自由的一个例子。这里是为了更好地传达这一点。我要读一下，你告诉我，这是更好的沟通吗，Jordy？他说，正如我们之前所说，我们决定优先考虑青少年的安全而非隐私和自由，并且我们没有放宽任何与心理健康相关的政策。

He said e r o t I c a. He typed it out into his ex app, sent the tweet, and he said it was meant to be just one example of us allowing more user freedom for adults. Here is an effort to better communicate it. I'm gonna read it, and you tell me, is this better communication, Jordy? He says, as we have said earlier, we are making a decision to prioritize safety over privacy and freedom for teenagers, and we are not loosening any policies related to mental health.

这是一项新的强大技术，我们认为未成年人需要重要保护。我喜欢这一点。我们也非常重视将成年用户视为成年人的原则，随着AI在人们生活中变得越来越重要，允许人们以他们想要的方式自由使用AI是我们使命的重要部分。当然，这并不适用于所有情况。例如，我们仍然不允许那些对他人造成伤害的事情，并且我们会以不同于非心理健康危机用户的方式对待那些正处于心理健康危机中的用户。

This is the new and powerful technology, and we believe minors need significant protection. I love that point. We also care very much about the principle of treating adult users like adults as AI becomes more important in people's lives, allowing a lot of freedom for people to use AI in the ways that they want is an important part of our mission. It does it doesn't apply across the board, of course. For example, we will still not allow things that cause harm to others, and we will treat users who are having mental health crises very different from users who are not.

在不采取家长式作风的前提下，我们将尝试帮助用户实现他们的长期目标，但我们并非世界选举产生的道德警察，就像社会区分其他适当界限一样。例如，R级电影，我们想在这里做同样的事情。让我们为日益流行的MPAA评级系统喝彩。是的，我非常渴望这个。

Without being paternalistic, we will attempt to help users achieve their long term goals, but we are not the elected moral police of the were of the world in the same way that society differentiates other appropriate boundaries. R rated movies, for example, we wanna do the same thing here. Let's hear it for the MPAA rating system that's gaining popularity. Yes. I want this so badly.

我只是想在App Store里，它应该标明是R级。

I just wanna in the App Store, it should say r rated.

R级。哦，但那不是NC-17级，那是不同的东西。所以好吧。我们一直在争论这个。所以你是电影爱好者？

R rated. Oh, but that is not NC 17 rated, which is a different thing. And so Okay. We were debating this. Is So you're a movie buff.

不，不是。我的意思是，亚马逊Kindle上的情色内容，那是X级吗？那和OnlyFans一样吗？那和成人内容视频网站一样吗？

Is no. No. I mean, is is is Amazon's erotic content on the Kindle, is that is that x rated? Is that the same as OnlyFans? Is that the same as a a an adult content video site?

就像，感觉确实有些不同。所以我不知道。也许这是对保持在R级类别而不进一步的认可。我不知道。我们拭目以待。

Like, it does feel somewhat different. And so I don't know. Maybe this is a nod to staying in the r rated category, not going further. I don't know. We'll see.

Jordy，你对这篇帖子有什么反应？这是有效的沟通吗？戴上你的Lulu帽子。

What's your reaction, Jordy, to this post? Is it effective comms? Put on your Lulu hat.

是的。我的意思是，问题是昨天有近1400万人看到了这篇帖子。你再加上，你知道的，成千上万的其他帖子，然后，你知道的

Yeah. I mean, the problem is that 14 almost 14,000,000 people saw yesterday's post. You add that up with, you know, thousands of other posts and, you know

那确实爆炸了

That really did blow up

第14个帖子肯定会获得大量浏览量，但我认为社区已经接纳了它。

14 post is gonna get certainly get plenty of plenty of views, but I think the the community has already taken it.

我不知道。我我我觉得我觉得十二月的时间线，从现在到十二月之间还有很多天。AGI可能在那之前就实现了，但在政策出台之前，有很多事情可能会调整它，真的把它孤立起来。我不知道。我们我们得看看。

I don't know. I I I think I think the the the December timeline, there's a there's a lot of days between now and December. AGI might hit before that, but there's so much that could happen to tweak that policy before it goes out, really silo it. I don't know. We'll we'll have to see.

我我我几乎可以保证，当它推出时，当天时间线上会有疯狂的截图，人们让它做疯狂的事情，因为人们喜欢越狱这些东西。他们已经这么做了。不管怎样，让我告诉你Turbo Puffer。搜索每一个字节，无服务器向量和全文搜索，基于对象存储从头构建，快速，便宜10倍，并且极其可扩展。你可以把你所有的色情内容加载到Turbo Puffer中。

I I I can almost guarantee that when it rolls out, there will be crazy screenshots on the timeline that day of people getting it to do crazy stuff because people love jailbreaking this stuff. They've already been doing it. Anyway, let me tell you about Turbo Puffer. Search every byte, serverless vector, and full text search, built from first principles on object storage, fast, 10 x cheaper, and extremely scalable. You can load all your erotic content into Turbo Puffer.

搜索它便宜10倍。是的。而且我们有我们的

Search it 10 times cheaper. Yeah. And we have our

嗯，我们的第一个

Well, our first

节目的第一位现场嘉宾。我们请到了Adam Ryan

in person guest of the show. We got Adam Ryan

亚当·瑞安。

Adam Ryan.

来自Workweek。传奇人物。传奇人物。快过来。这是专业社交的平台。

From Workweek. The legend. The legend. Bring it in. It's the platform for professional networks.

欢迎来到节目，亚当。你好。第二次，第三次，我都记不清了，但非常感谢你的加入。我要把麦克风转到这里。

Welcome to the show, Adam. Hello. Second time, third time, I've lost track, but thank you so much for joining us. I'm gonna turn this microphone over here.

我们要讨论的是

We're be talking about

你将情色内容引入专业社交网络的策略。

Your strategy to bring erotica to professional networks.

我一直在等待，我一直在等待

I've been waiting I've been waiting

这一刻的到来。

for this moment.

是的。是的。

Yes. Yes.

这正是全部内容的核心。

Exactly what it's all about.

实际上，让我们重新开始

Actually, kick us back off

从论文开始

with thesis

整个公司的理念是人们在职场工作中渴望情色元素。

of the whole company is that people wanted erotica in their work in the workplace.

让我们重新开始，比如，实际的公司概述，你如今是如何自我介绍的。

Kick us back off with, like, the actual, the actual overview of the company, how you introduce yourself, these days.

还有还有，为那些错过了上次访谈的人快速回顾一下历史。

And and the the quick history too for anybody that that met missed the last interview.

是的。四年前创办了这家公司，这个人是公司创立故事的一部分。一路走来我们学到的一点是，大家都关注消费者行为——人们更信任个人而非机构。是的。在美妆、时尚、游戏领域，都有像凯莉·詹娜这样的代表人物。

Yeah. Started the company four years ago, and this guy was part of the the founding founding story of that. And today, you know, one of the things that we learned along the way that everyone has focused on consumer behaviors of, hey, I trust individuals over institutions Yep. And beauty, fashion, gaming. There's a Kylie Jenner out there.

嗯。但在职场领域却没有人。是的。如果你是市场副总裁，如果你运营医院或医疗系统，你会找谁？嗯。

Mhmm. There's nobody in work. Yeah. If you're VP of marketing, if you run a hospital or health system, who do you go to? Mhmm.

如果你关注美妆界的凯莉，那个领域就没有其他人了。

If you're into into beauty of Kylie, don't have anyone else in that space.

是的。所以那个

Yeah. So that

是我们想要解决的第一个问题。然后我们想可能是因为金钱、带宽问题，这些人很忙，他们已经赚了很多钱，所以不创作内容。是的。

was the first problem we wanted to go solve. And then we thought it was potentially because of money, bandwidth, like this what these people are busy. They already make a lot of money. That's why they don't create content. Yeah.

我们意识到实际上缺少一个平台。人们不会上领英说：'嘿，我正在准备即兴演讲，有人能帮我吗？' 嗯。

We realized that there's there's an actual just platform missing. People aren't going on LinkedIn being like, hey. I'm prepping for a riff. Can someone help me? Mhmm.

所有这些平台都是开放网络。你未必能如愿与同行进行对话。其他选择大多都是模拟式的。

And all these platforms are open networks. You can't necessarily have the conversations with your peers that you want to. All the other options are mostly analog.

嗯。

Mhmm.

所以，用我的话说，你实际上是在建立一个专业网络，但它是高度特定的人群，基本上可以说是——我不知道用‘层级’这个词是否准确——但如果你是市场副总裁，并且只想与其他市场副总裁互动，那么Workweek就是你可以实现这一点的平台。

So so putting it in in, my terms, you're building, you know, effectively a professional network, but hyper specific to people, like, within basically certain I don't know if the right way to say it is bands, but if you want if you're a VP of marketing and you want to interact only with other VPs of marketing, like, Workweek is the platform that you can do that on.

是的。我们建立了一个横向平台，然后按职业进行了垂直细分。嗯。所以我们有医疗保健、市场营销、人力资源、电子商务等领域的平台。如果你是这些领域的高级领导者，我们会对你进行验证，以防止机器人和不良信息的干扰，同时确保你可以进行那些值得信赖的讨论和对话。

Yeah. So we've built a horizontal platform and then we verticalized it by the profession. Mhmm. So we have one for in health care, marketing, HR, e commerce, etcetera. And so if you're in if you're a kind of senior leader in one of those spaces, we verify you so that prevents the bots, the erotica as well as make sure that you can kind of have those those trusted discussions and conversations.

我的意思是，我猜关于不良信息的讨论，人力资源小组和支持小组可能已经炸开锅了。所有财富五百强公司的人力资源领导者都在那里。

I mean, I I imagine that with the with the erotica discussion, the the HR group, support group is probably going wild. All the HR leaders across Dowser up there. The Fortune five hundred.

是的。是的。这正是我们要

Yeah. Yeah. That's exactly what are gonna

怎么处理？因为我们把ChatGPT给了所有人，然后他们就能用它来做各种事情，甚至超出

do about this? Because we've given ChatGPT to everyone and then they're gonna be able to use it for over the

是的。

Yeah.

我猜在商业计划方面，你可以直接关掉那个。是的。

And I'm I'm assuming on business plans, you can just turn off that. Yeah.

像这样，你能吗？也许

Like This is Can you? Maybe

所以它是值得的

So it's it's worth

40棋。你现在必须升级到企业解决方案，因为你需要能够微调整个组织，这样我们才能让实习生泰勒远离成人版。

40 chess. You gotta upgrade to the to the enterprise solution now because you gotta be able to fine tune your entire organization so we can keep intern Tyler off of the off of the adult version.

泰勒在整个系列中都一直在咧嘴笑。

Tyler's been grinning this whole series.

他简直笑翻了。太搞笑了。

He's just cracking up. It's funny.

是的。我的意思是，我听说在这些小众领域里有疯狂的价值故事。有句谚语，好像是，小众里有财富。小众里有财富，类似这样的话。我听说过一家公司只做纯粹的Slack社区。

Yeah. I mean, I've heard I've heard there's, like I've heard crazy stories of value in these, like, niches. There's some phrase, like, there's riches there's riches in the niches. There's riches in the niches, something like that. I heard about one company that was just doing pure Slack communities.

所以他们会去找遍所有不同初创公司的办公室主任，把他们全部拉进一个需要付费加入的Slack社区。但当他们公司有需求时，比如需要修理暖通空调系统，他们就能直接到这个Slack频道发问：'嘿，还有谁在管理办公室，需要安排过滤水的定期配送，或者找更好的Wi-Fi供应商之类的？'这确实很有道理。市场副总裁的情况也一样，还有很多其他类似的群体。

So they would go and find all of the chief of staffs across a whole bunch of different startups, put them all in a Slack community that they had to pay to be part of. But then when they got a request from their company, oh, I need to, you know, fix up the HVAC system, they'd be able to just go to this Slack channel and say, hey. Who who else is, you know, managing an office and needs to figure out a regular delivery schedule of filtered water or, you know, a better Wi Fi provider or something like that. And that makes a lot of sense. Same thing with VP marketing, a whole bunch of other stuff.

嘿，我们在讨论...你觉得这个连续谱该怎么看？比如一边是LinkedIn这种超级宽泛的类别，另一边可能是像Lenny这样的Substack创作者，拥有自己的节目，像个独立个体存在于这个世界，再对比这些精心策划的社区？你觉得普通用户的'茶水间'会存在于哪个位置？

Hey. We're we're we're talking. How do you think about the the like, if there's this continuum between, like, LinkedIn's this super broad category or you have someone like, you know, a Lenny, a Substack Yeah. Creator with a show that's like, you know, this individual off in this world versus, like, these like curated communities? Like, where do you see like the water cooler existing for the for the average user?

我觉得有传统的社交网络，都是免费加入的。对。然后你有那个办公室主任社区，Silva拥有那个，是订阅制的。

I think there's there's the traditional social networks which are all free to join. Yeah. And then you have the chief of staff one. Silva owns that one. It's a subscription.

好的。

Okay.

大部分那种社区，你知道的，就像是'X领域的YPO（青年总裁组织）'。他们都创建了那种社区，但都是订阅制。是的。真正做过身份验证的只有一家公司

Most of those, you know, it's the YPO but for X. They've all created those, but subscription. Yep. There's really only been one company that's done verification

嗯。

Mhmm.

通过让它免费...有意思。那就是Doximity，好的，是医生的专业网络。

With making it free Interesting. Which is Doximity Okay. Professional network for doctors.

哦，他们大概有10分左右。

Oh. They're like a 10 ish.

140亿美元市值的公司？什么？今天。自成立以来产生的现金流比爱彼迎还多，不可能吧。而且只融资了50美元，真是一家了不起的公司，去买它的股票吧。

$14,000,000,000 company What? Today. Produces more cash flow than Airbnb since starting No way. And only raised $50 Hell of a company, go buy the stock.

那很好。不是财务建议。他说买股票的时候，不是财务建议。

That's good. Not financial advice. When he said buy the stock, it was not financial.

我们确实在遵循很多这样的蓝图。如何构建一个平台，在其中创建像社交网络那样的参与循环，是的。但你仍然拥有验证机制。

And we're really following a lot of that blueprint. How do you build actually a platform where you create engagement loops like a social network Yeah. But you still have the verification.

明白了。所以因为

Got it. So Because the

因为如果你有一堆营销副总裁

because if you have a bunch of VPs of marketing

是的。

Yeah.

在一个网络中，针对那个受众的展示价值可能比LinkedIn上的平均展示价值高出数千倍。

In a in a on a network, that the impressions against that audience are are worth thousands of times, potentially, the average impression on a LinkedIn.

是的。

Yep.

是的。而且我认为，昨天节目中的Alexis谈到了死互联网理论。这就像是，是的。尝试解决这个问题的完美时机是，比如，你如何真正知道谁在做什么？而我使用Reddit。

Yep. Well, and I think it's to the Alexis on the show yesterday talked about the dead Internet theory. It's like Yeah. The perfect time for trying to solve that is, like, how do you actually know who's doing what? And I use Reddit.

我喜欢Reddit。但我过去能通过回复判断你是否聪明。而现在有了Chatuchibi Tea，我就觉得，我不知道了。

I love Reddit. But I used to be able to tell, are you smart by your reply? And now with Chatuchibi Tea, I'm like, I don't know.

或者像是，你是否付出了努力

Or like, have you put effort into

如果你付出了，我就会觉得，好吧。现在我却觉得，我完全不知道这是谁。因此，整个互联网的信任度我认为正在转变，特别是在工作方面，比如非常在意是谁在给你建议。

And your if you did, I'm like, alright. Now I'm like, I have no idea who this is. And so the kind of whole trust of the Internet's I think shifting, particularly when it comes to work, like really care about who's giving you advice.

是的。节目的朋友Isaac几天前发帖说，做Raya做过的事，但是为一个社交网络，只限人类，人工验证，难以加入，不知道如何实现，但直觉上感觉可能行得通。我喜欢看到这个，因为这多年来一直是Workweek的论点。你们一直在朝这个方向努力，所以人们终于意识到了。你对X最近的一些变化有什么看法？

Yeah. Isaac, a friend of the show, posted a couple days ago, do what Raya did, but for a social network, humans only, human verification, hard to get into, no idea how to do it, but feels intuitively like it could work. And I love seeing this because work this has been Workweek's thesis for years now. You guys have been building towards it, so people are are finally realizing. What do you think about some of the recent changes on X?

他们正在增加显示用户所在地的功能。这可能有诸多好的理由，也有一些不好的理由。我认为一些担忧来自像英国这样的地区，那里对言论自由的容忍度较低。

They're adding functionality that shows where the user is based. There's probably a lot of reason good reasons for this and then some bad reasons. Think some of the concerns were people that are in areas like The UK that don't tolerate free speech.

哦。

Oh.

你可能会成为目标，或者当地当局可能会试图追查账户。但这实际上是一个如此有趣的平台，因为我们实际上都把它当作一个专业网络，一个讨论我们行业或广泛各行业的地方。但同时，就像约翰昨天把它比作潜水酒吧一样，对吧？就像人们总是在那里打架。

You could be targeted or or or authorities there could be trying to come after accounts. But it's like such a actually such an interesting platform because we effectively all use it as a professional network and a place to talk about our the industry or or various industries broadly. But at the same time, it's like John compared it yesterday to being a dive bar. Right? It's like it's the people are getting in fights all the time.

就像，你知道，那种潜水酒吧。

There's like, you know The dive bar.

那个角落里可疑的家伙，你知道，他有前科。

That that shady guy in the corner that, you know, he's got a history.

他有前科。是的，没错。但是，是啊，我不知道。

He's got a history. Yeah. Exactly. But, yeah, I don't know.

我的意思是，这就是尼基塔正在做的事情。如果你是一个开放的网络，而且是免费的，你就会有机器人。是的。然后你还会遇到，你知道，有人滥用它们。

I mean, that's what Nikita's doing. If you're an open network, that's free. You're gonna get bots. Yeah. And then you also are gonna get, you know, people abusing them.

我们喜欢我一直在推特上。

We love I'm on Twitter all the time.

但是

But the

推特在开始允许更多机器人后就失去了信任身份，然后蓝勾市场也改变了验证流程，你可以购买自己的验证。但俄罗斯机器人仍然有信用卡。

Twitter lost its trust identity when they started allowing more bots, and then also the blue check market changed the the verification process a where you could buy your own verification. But Russian bots still have credit cards.

我记得我得到了验证，然后六周后，他们花了几年时间。我也是。是的。就像是

I remember I got I remember I got verified, and then six weeks later, they did spent the years. Me too. Yeah. It's like

给奶奶发短信和

texting grandma and

终于得到了。我发现以前只需要

finally got got it. I found like old used to just have to

需要大约五篇关于你的新闻文章之类的

Needed like five news articles or something about you

或者类似的事情。我找到了一些旧新闻文章和维基百科页面。流程获批了，然后一周后，就像是，哦，

or something. I found some old news articles and Wikipedia page. Process got approved, and then a week later, it was like, oh,

大家都在想，你对独立媒体的现状有什么看法。你是打造The Hustle的关键人物。曾有一段时间，每个人都说，我想为X打造Morning Brew或为X打造The Hustle。基于此诞生了一些很棒的企业。它们中很多现在还在蓬勃发展吗？

everyone thinks What is give us your take on the current state of of independent media. You were a key player in building the hustle. There was a time that everybody was saying, I wanna build the morning brew for x or the hustle for x. There were some great businesses built off of that. Are they are many of them still thriving?

当前的状况是怎样的，也许你可以先从，比如，新闻媒体业务开始讲起。

What is the current state of the maybe you start with, like, the newsletter media business.

是的。我是说，Substack所做的已经民主化了访问渠道。我认为，归根结底，人们在媒体领域总是归结为：你需要帮助人们赚钱，帮助人们获得关注，或者帮助人们更快地制作内容。如果你解决了其中一个问题，你就有了一个不错的业务；如果三个都解决了，你就有了一个很好的业务。而正在发生的情况是，我认为独立一方正在崛起，因为像Beehive和Substack这样的平台已经开始真正让这成为可能。

Yeah. Mean, think what Substack has done has democratized access. I think, like, ultimately, what people always come down in media is you need to, like, help people make money, help people get eyeballs or help people make content faster. And, like, you solve one of those things, you have a decent business solve all three, you have a good one. And what's happening is I think the independent side is starting to rise because platforms like Beehive and Substack have started to really allow that to be the case.

我认为我们开始在那里为我们更具体的垂直领域建立自己的案例。但最终，消费者才是说出真相的人。就像，Lenny通过订阅赚的钱比大多数存在了五十年的B2B行业出版物的出版商都多。是的。所以消费者的钱包最终总是会说话。

I think we're starting to make our own case there for more specific kind of verticals. But I think consumers in the end are the ones telling the telling the truth. There's like, Lenny is getting more more money on a subscription basis than most publishers in b to b trade publications that have been around for, like, fifty years. Yeah. So consumers' wallets always end up wearing

给我讲讲Trungfan的存在，他的业务。

Walk me through Trungfan's existence, his business.

他是个有趣的人，在

He's the interesting man in

世界。他在Workweek上写商业内容，但不在Workweek工作。他是员工吗？他在公司网络上吗？比如，他有一个通讯简报。

the world. Writes on business with Workweek, Not at Workweek. Is he an employee? Is he on the network? Like, he has a newsletter.

他就是Trung。

He's just Trung.

这是怎么运作的？只是

How does that work? Just

Trung。Trung是最早的创作者之一

Trung. Trung was one of the first one

的

of the

Workweek最早签约的创作者

first creators that Workweek

是的。嗯，我所以是的。在The Hustle签了他。好的。而且认识他有一段时间了。是的。

Yeah. Well, I so Yeah. Signed him at The Hustle Okay. And had known him for a while. Yeah.

实际上他是我们的一大亮点，因为当我们在Hustle聘用他时，他非常有趣，而且比其他人优秀得多。是的。他改变了拥有150万订阅者的新闻简报的互动情况。哇。几乎是一夜之间。

And he actually was a big part of, like, our our shtick is because when we brought him on at the hustle, he was so funny and so much better than everyone else. Yeah. He changed the engagement of a newsletter with 1,500,000 people. Woah. Like overnight.

我的意思是，好像每个人都开始更多地打开邮件。是的。广告效果也提升了。所以这实际上是一个很大的因素。就像一个人可以完全成为那种100倍影响力的创作者，

I mean, was like everyone just started like more and opening more. Yeah. Ads went up. And so we actually that was like a big part of it. It's like one person could totally you have like the 100 x creator,

确实是的。

which Yeah.

大家都在谈论开发者。如今，首先，他是世界上最好的爸爸。你不能提到Trung而不谈论TJ Trung Junior。

Everyone talks about developers. Today, one, he's like the best dad in the world. You can't mention Trung and not talk about TJ Trung Junior.

哦，

Oh,

是的。而且我认为他从未参加过我在的任何公司的全员会议，但你知道，这就是他的风格和他做事的方式，他自带所有特质。但是，是的，他是

yes. And I don't think he's ever been on an all hands call for any company I've ever been a part of with him, but, you know, that's just kind of his vibe of what he does, and he comes with comes with it all. But, yeah, he's

认为他是个有趣的人但是

think he's funny guy But

他是否是一个管理较小专业人士群体的例子，可以被赋予某种称号？

is he an example of someone who's who's stewarding a a smaller community of professionals that could be given some sort of moniker?

他更多是在工作日方面协助我们。所以我们有很多像布莱克·马登这样的人，他负责医疗保健领域。是的，如果你在医疗行业，你一定知道布莱克是谁。

He he works more on the work week side of helping us. So what we have a lot is we have people like Blake Madden who covers health care. Yep. If you're in health care, you know who Blake is.

好的。

Okay.

在覆盖细分医疗领域很难让某人走红。是的。Trung是帮我们做到这一点的人。

Very hard to get someone viral in covering niche healthcare. Yep. Trung's the guy to help us do that.

好的。所以他更像是

Okay. So he's a little bit more like

他是光环。他是光环创造者。

He's a halo He's halo creator.

明白了。好的。然后还有特定的细分领域。是的。我想象中你们这些不同的社区是否存在某种幂律分布。

Got it. Okay. And then there's specific niches. Yeah. Is there a power I imagine there's some sort of power law to those to the various communities that you have.

比如，最顶尖的是什么？你真正在哪个领域，比如，占据主导地位？

Like, what's at the top? Like, where are you really, like, dominating?

大约，比如，30%的收入来自人力资源，9%来自医疗保健。但如果我们看总目标市场，它实际上分布得相当均衡。好吧。相当平均地，比如，我们预计每个网络能获取多少人。是的。

About, like, 30% of revenue is HR, nine percent's health care. But if we look at Sure. Like TAM, it's actually pretty indexed Okay. Pretty equally of, like, how many people we think we can get per Yeah.

每个网络。然后在人力资源内部，例如，是有多个社区、多个创作者，还是只有一个大池子？你怎么看

Per network. And then within HR, for example, are there multiple communities, multiple creators, or is it all just one big pool? How do think

这个问题？

about that?

所以我们更倾向于把它看作一个网络。当然。所以SafeSpace是人力资源网络，我认为这是其中的一部分。你必须使用行业术语的行话。

So we think about it more like a network. Sure. So SafeSpace is the HR network, and I think this is part of it. You have to use the lexicon of lingo.

比如，当然。

Like Sure.

Blake使用的语言与人力资源非常不同。是的。只是讨论一下。所以每一个都发展得与之匹配，并且有趣。然后在网络内部，我们有创作者。

Blake uses very different language than HR. Yeah. Just talk about. And so each one's developed to match that, and Interesting. And then within the network, we have creators.

是的。

Yep.

所以我们相信，如果你想为HR领导者写作，嗯。我们是获取关注的最佳平台。嗯。然后我们有自己的广告平台来支持这一点，是的。他们可以从中获得收益。

And so we believe if you wanna write for h write for HR leaders Mhmm. We're the best place to get the eyeballs. Mhmm. And then we have our own ad platform that helps fuel that Yep. That they that they can get money off of.

还有其他类型的服务或活动，比如，我不知道，是在整个平台上分摊或共享的吗？显然，广告销售似乎是其中之一

Are there other are there other kind of services or or activities that are kind of, I don't know, amortized or shared across the entire platform? Obviously, ad sales seems like one of

是的。

them Yeah.

但实际的软件呢？还是你们会说，嘿，Trung，比如，没有比在X上发帖更好的产品了，所以我们不打算构建自己的短文本

But actual software? Or do you say, hey, Trung, like, there's no better product than just posting on x, so we're not gonna try and build our own short form text

是的。应用。我们让他们自己做自己的事。

Yeah. App. We just let them do their own thing.

是的。用你最适合的工具。如果你想用Substack或Slack

Yeah. Use your use whatever tool is best for the job. If you wanna be on Substack or Slack

我们使用自己的新闻通讯。就是Newsletters。好的。特别是Newsletters，就是这样。这是我们利用内容将人们引入网络的方式。

We use our own newsletter. So Newsletters. Okay. Newsletters, particularly, is is that. And that's our way that we use content to drive people into the networks.

是的。然后你可以将人们在不同的网络之间进行调配。

Yep. And then you can shuffle people around through the different networks that

我们使用的。与Substack现在为他们做的方式没有太大不同，但我们的更垂直。

we use. Not that different than how Substack's now doing it for theirs, but ours is more vertical.

对，对。他们的更像是新闻驱动的。是的。有道理。

Yeah. Yeah. And theirs is more, like, news driven. Yeah. Makes sense.

政治的，是的。

Political and yeah.

想听听你对大卫·埃里森最近在派拉蒙和天舞影业方面的行动的看法。很多人认为他们为《自由新闻》支付的价格很疯狂。我认为当我们从人才收购的角度来看，并考虑到拥有一个非常高价值的受众群体时，我们为其进行了辩护，但我还是想听听你的想法。

Wanted your take on on David Ellison's recent actions with Paramount and Skydance. A lot of people thought that the price they paid for the free press was crazy. I think we came in the defense of it when you look at it from a talent acquisition standpoint and having a very high value audience, but I wanted wanted to hear how you thought about it.

合并本身，我的意思是，这使他们能够竞争。想远一点，每个人都想赢得流媒体市场。我的意思是，这就是游戏的名字。所以这最终让派拉蒙拥有了更多IP来与Disney+和Netflix竞争，而他们都在...Netflix今天刚宣布，我认为，与Spotify合作，开始引入更多播客。就像，这是一场流媒体的IP竞赛，而派拉蒙正在输掉，我认为这使他们能够做到这一点。

The the merger itself, I mean, it allows them to compete. Think, like, taking a big step out, everyone wants to win streaming. I mean, that's, like, the name of the game. And so it allows Paramount to now finally have more IP to compete with Disney plus and Netflix while they're all and Netflix just announced, I think, today, a partnership with Spotify to start bringing more podcasts on. Like, it is an IP race of the streaming, And Paramount was losing, and I think this allows them to do that.

所以我认为更大的图景实际上是那样的。

So that's I think the bigger picture is actually that.

是的。自由媒体在其中就像是一个四舍五入的误差。

Yeah. The the free press is like a rounding error as part

总共是1.5亿，而他们购买的，你知道，只是其中的1200万。这不是世界末日。我认为如果你要

of this It's a 150,000,000 total, and they're buying, you know, 12 revenue of that. It's not the end of the world. I think that was also if you're going

那是他们的收入数字吗？

Was that their was that their revenue number?

1200万到1500万，我

12 to 15, I

认为是的。

think. Yeah.

好的。

Okay.

是的。

Yeah.

我以为

I thought

它高于20，但是

it was higher than 20, but

我不知道。是的。大概是10倍左右，我想。当然，10到12倍。

I don't know. Yeah. It was around like 10 x multiple, think. Sure. 10 to 12.

但我也认为，当你创办自己的公司时，基本上就像大卫·埃里森那样，你会想要自己的人。当然。是的。而且你会确保你建立了正确的基础，那就是

But they I also think when you're starting your own company, which basically David Ellison is, like, want your own people. Sure. Yeah. And that's you're gonna you're gonna make sure you you build the right foundation, and that's that's

是的。每个人都专注于他们为自由出版社支付了什么？你应该关注的是巴里·韦斯会让CBS变得更有价值多少。

Yeah. Everybody's focused on what is what what did they pay for the free press? You should be focusing on how much more valuable will Barry Weiss make CBS.

这没什么不同，就像，在一个完全不同的世界和非常不同的数字中。但扎克说，比如，我需要一个AI团队。我要去得到我想要的人。他说，我想要拥有最好的编辑团队。我要去得到我想要的人。

This is no different, like, in a totally different world and very different numbers. But then Zuck saying, like, I need an AI team. I'm gonna go get the people that I want to get. He's saying, I wanna have the best editorial team. I'm gonna go get the people I wanna have.

是的。你觉得那些大型平台、公共平台在那些小众专业网络中变得不那么相关了吗？我感觉如果回到十年前在推特上，会有像#营销星期二这样的标签，我会看到一群营销人员在星期二讨论营销。对，基本上就是那样。

Yep. Do you think the the bigger platforms, the public platforms have kind of become less relevant in those niche professional networks? I feel like if I went back like ten years on Twitter, there would be like hashtag like marketing Tuesday and I'll and I would see a bunch of marketing people talk about marketing that On Tuesday. Yeah. Basically, it was like that.

疯狂吧？那不是挺有意思的吗？也许是营销星期一之类的？

What crazy Wasn't that something? Maybe it marketing Monday or something?

周一显化日？那是当时的热门。

Maniface manifestation Mondays? That was the thing.

那时候有很多像标签这样的东西，所有老套的标签。围绕特定话题的聊天，你会看到很多人在谈论他们特定的领域，但现在X感觉像是政治新闻茶壶，它处于比任何小众话题都高得多的层面。

There was a thing around, like, hashtags All the corny hashtags. Chats around certain things, and you'd see a lot of those people that were just talking about their specific, like but now, x feels like teapot political news, like there's there's there's it's it's at a much higher level than any sort of niche

讽刺的是，我认为当他们全面转向算法优先时，就失去了社区感。嗯。以前你关注的人就是你的圈子，对。而算法完全改变了产品的动态，但因为品牌和标识没变，没人意识到它们已经不同了。

one the irony, I think, was, like, when they went all algorithms first, they lost the sense of communities. Mhmm. And when you used to follow people and that's who you did, it was like your people. Yeah. And when it's algorithms, it totally changed the dynamic of the product, but because it was the same brand and logo, no one thought about them

它是慢慢发生的。

differently. It happened slowly.

我认为这让信息流更有趣了。是的。或者说我们确实失去了

It's I think it makes the feed more interesting Yes. Or we definitely lost

而且他们赚更多钱。我的意思是，你就是为了参与度而存在的。完全正确。随着尼基塔优化算法，你的互动率也会上升。所以，他只关心这个。

some And they make more money. I mean, like, you're you're there for engagement Totally. And your dial mile goes up as Nikita makes that algorithm better. So, like, that's all he cares about.

是的。是的。

Yeah. Yeah.

我认为这最终就是他们的动机所在。我确实觉得像领英，你知道，我知道你们在那很厉害。是的。并且真的很依赖领英的核心。没错。

And I think that's, like, ultimately what their incentives are. I do think like LinkedIn, you know, I know you guys are huge there. Yeah. And really rely on the center of LinkedIn. Exactly.

所以虽然不想这么说，但你知道，我觉得大概十年前，如果有人像这样，嘿，我在领英上看到你的联系人。你能帮忙介绍一下吗？我会说，哦，好的。我认识他。今天，

So hate to hate to say this, but, you know, I think like ten years ago, somebody like, hey, I saw your connect on LinkedIn. Can you make that introduction? I'd be like, oh, yeah. I I know it. Today,

我

I'm

就像，我有

like, I have

完全不知道是谁

no clue who

就是这样。

that is.

我认为大多数社交媒体都存在负面网络效应。有趣的是，用户越多，体验反而越差。

I think there's a negative network effect happening across most social. Interesting. The more people added, the worse it gets.

是的。不。这非常有道理。

Yeah. No. That makes a ton of sense.

我们私下讨论过OnlyFans为何一直无法出售自己。他们已经尝试了很长时间。昨天的新闻，显然，你知道，其他公司也在涉足通过浪漫聊天变现这一大类业务，

We were talking off air about how OnlyFans hasn't been able to sell itself. They've been trying for a long time now. The news yesterday, obviously, know, other companies leaning into that broad category of monetizing romantic chatting,

这个

which

OnlyFans在这方面可能比世界上任何公司都做得更好。今年早些时候xAI涉足浪漫伴侣领域时，我们认为他们这样做是有道理的。他们需要一个策略来，你知道，推动新用户注册、参与度和订阅等等。但为什么你会对OnlyFans找不到买家感到惊讶呢？

OnlyFans has done better than probably any company in the world. When xAI was leaning into romantic companions earlier this year, we were like, it made sense for them to do. They needed a strategy to, drive, you know, new user sign ups, engagements, subscriptions, etcetera. But why do do you think have you been surprised that there's no buyer for OnlyFans?

是的。按人均现金流来看，我认为它是世界上最好的公司。

Yeah. It's like the on a a cash flow per head count, I think it's the best company in the world.

直到什么是什么？Tether？

Until what what is it? Tether?

我想是90亿美元吧，OnlyFans每年现金流就这么多，大概只有70名员工左右。这简直太离谱了。我知道老板每年给自己开5亿美元的分红支票，银行里还留着一大笔钱。但他们从未能上市，也找不到买家，因为内容的问题。

9,000,000,000, I think, OnlyFinds does in cash flow a year with, like, 70 employees or something. It's, like, absurd. I know the owner takes, like, a he pays himself a 500,000,000 dividend check a year, and then they leave, like, plenty in the bank. But they've never they can't take themselves public. They've not been able to find a buyer because of the content.

而现在的讽刺是，大部分内容只是虚拟面孔，但他们允许——他们开发了工具。OnlyFans的前CEO，我的前同事，她太棒了。她为模特们创建了很多工具，让她们能以安全的方式制作内容并更好地变现。现在他们更进一步，发展到像是聊天机器人那样，基本上已经不再进行真实对话了。

And now the irony is most of it's just a face, but they allow they've created the tooling. The c old CEO of OnlyFans, former colleague of mine, she's amazing. She created so many tools for the models to make content in a safe way and then to monetize it better. And now they've advanced further to where, like, it's a chatbot. They're essentially, like, not even having real conversation anymore.

这和我们在讨论的OpenAI情况没太大不同。是的。这只是品牌问题。我认为这就是，你知道，X想成为万能应用，你会觉得，我有点能理解。这，我认为，就像你之前提到的苹果类比，这对OpenAI来说有点玷污了这个类比。

It's not that different than what we're talking about for OpenAI. Yeah. And it's just branding. And I think that's the, you know, x being the everything app, you're like, I can kinda see that. That's that's, I think, to your point earlier about the Apple comparison, this this is a little bit of a tarnish on that comparison, I think, for OpenAI.

是的。是的。

Yeah. Yeah.

好吧，在下一位嘉宾加入之前，我们有些突发新闻要报道。但非常感谢你的光临。

Well, we have some breaking news we gotta cover before our next guest hops on. But thank you so much for stopping by.

什么突发新闻？

What's the breaking news?

我必须知道。这类事情太多了。MrBeast已经申请了商标，要推出他自己的金融服务，将命名为MrBeast金融。他显然是看到了我们关于Erebor的帖子，然后就想，必须掺和进来。开个玩笑。

I gotta know. There's a ton of these. MrBeast has filed a trademark launching his own It will be called MrBeast Financial. He clearly saw our post about Erebor and was like, gotta get it on the action. Just kidding.

申请是两天前提交的。

Application was filed two days ago.

哦，但是

Oh, but

这是真的。今天早上发生的。

it's real. Happened this morning.

看到了。你对此有什么看法？感觉如果你想从广泛的年轻受众那里变现，也许创建一个可以货币化存款的银行是一种相当契合的方式？这有点像

Saw What's your what's your take on this? It feels like if you want to monetize an extremely broad audience of young people, maybe just creating a bank that you can monetize deposits is a fairly aligned way to do that? It's kind of like

我猜是吧。

I guess.

是的。我的意思是，我只是

Yeah. I mean I just

不知道我不知道那件事

don't know that I don't know that

就像一条信任的通道，迟早你会搞砸的。比如，你知道，为什么我会想要一个Mr. Beast银行？我

like a lane of trust sooner or later that, like, you're just gonna screw it. Like, you know, like, why do I want a Mr. Beast bank? I

不知道。我的意思是，从Mr. Beast酒吧或

don't know. I mean, the the the lesson from the Mr. Beast bar or

汉堡的教训

the Hamburgers.

汉堡的教训是它完全没有价格歧视。是的。所以如果你的观众里有富人，你只能向他们收取5美元买一个

Hamburgers is that it it has zero price discrimination. Yeah. And so if you get a rich person in your audience, like, you're only gonna be able to charge them a $5 for a

糖果。但是让我们来分析一下。

candy But but let's break it down.

银行你实际上可以得到

A bank you can actually get

这是一种非常不同的事情，就像

It's a highly different thing like

基本上就是物流运输。

shipping, basically.

但是没错。不过想想看，有钱人不会看到这家银行推出就决定我需要这个，我要转移一大笔钱过来。

But yeah. But but think about, like, wealthy people are not gonna see this bank launch and decide, I need this. I'm gonna move a bunch

要以十年为单位思考。要以十年为单位思考。一个孩子看MrBeast的视频就会想，没错，我需要我的第一个支票账户，我就开个MrBeast支票账户吧。

of Think in decades. Think in decades. A kid watches MrBeast is like, yeah. I need my first checking account. I'll set up MrBeast checking.

然后二十年后，他们会想，嗯，我可能应该换一个更正经的银行。现在还用在野兽金融上有点可笑，但你看，我还是在这里。

And then twenty years later, they're like, yeah. I should probably switch to something more serious. It's kind of ridiculous that I'm still on Beast Financial, but like, here I am.

我赚了一百万美元。没错。他实际上有银行特许经营权，这意味着他可以开展抵押贷款之类的业务，是的。

I made a million dollars. Yeah. He's actually a bank charter charter, it means he can get into doing mortgages and like Yeah.

硬核的是，我觉得这种事情在董事会上听起来总是很美好。是的。但然后你并不关心产品本身，也没有真正做出差异化，最后你就会想，好吧。当分销是你唯一的护城河时，那根本不算护城河。

Hardcore I think like that all it always sounds good in like a board meeting. Yeah. And then you like don't care about the product and you don't actually make it differentiated and then you're like, cool. When distribution's like your only moat, it's not a moat.

是的。我猜，我猜如果他算一下数字，比如，好吧，如果我观众里的普通人大概是16岁

Yeah. I so so I guess I guess if he runs the numbers on like, okay, if the average person in my audience is like 16

是的。

Yeah.

他们一个月花多少钱？是的。我打赌，我打赌青少年会觉得这很酷，会想用它。嗯。你大概可以算一下数字，计算一下，比如，好吧，普通青少年花费，他们的父母会给零花钱，或者他们可能有工作，把钱存入账户，可能一个月花几百美元。

How much money do they spend a month? Yeah. I bet I bet teenagers will think this is cool and will wanna use it. Mhmm. And you can probably just run the numbers and calculate like, okay, the average teenager spends, their parents are gonna send their their allowance or maybe they have a job, they send it into the account, maybe they spend a couple $100 a month.

我不知道现在的青少年都花多少钱。我记得在我那个年代，零花钱大概是20美元。然后钱存进去，他们花掉，他可以给他们奖励，这很酷，还有现金返还，他可以从奖励中赚一些钱。我觉得你刚刚把它变得很真实，就像我成年后，我想用真正的银行。我是个成年人了，你知道。就像我记得我从一个信用社

I don't know what teenagers spend these days. I I was remember back in my day, was like, you know, allowance was like $20. And and so it gets deposited in there and then they spend it and he can give them rewards and that's cool and cash back and he can monetize some of that that The rewards I think you just turned it really believe like, I became an adult and I was like, I wanna use a real bank. I'm an adult, you know. Like, I I just remember I I migrated from like a credit union

是的。

Yeah.

在我家乡转到，我想你一直在用卡。是的。即将到来。就像，你觉得你像个上大学的孩子，是的，我还在用我的Mr. Beast卡，就像是的。

In my hometown to like, I wanna You're using a card all the time. It's Yeah. Coming up. Like, you think you're Like, a a a kid that goes into college is like, yeah, I'm still using my Mr. Beast like it's like Yeah.

一个Lunchables的竞争者和一家银行。稍微了解你的受众一点。

A Lunchables competitor and a bank. Like, know your audience a little bit.

是的。但我想，从乐观的角度看，如果你能让数百万孩子仅仅使用 Beast Financial 作为他们的金融工具……

Yeah. But I so so I guess, like, from from the bull case is that if you can get millions of kids to to just be using Beast Financial as their as their

我的意思是，从技术上讲

I mean, as a technically