Polymarket’s Regulatory Hurdles, Pre-IPO Betting Boom on Prediction Markets, AI Increasing Workloads

欢迎各位收看《The Information》的TITV节目。

Welcome everyone to The Information's TITV.

我是阿卡什·帕什里卡。

My name is Akash Pasricha.

今天是5月27日，星期三。

It is Wednesday, May 27.

在进入今天的节目之前，先告诉大家，《The Information》年度AI峰会的门票现在已经开售了。

Before we get into today's lineup, tickets are now on sale for The Information's annual AI Agenda Live Conference.

这场活动将于9月23日在旧金山的现代艺术博物馆（SFMOMA）举行。

It is taking place in San Francisco on September 23 at SFMOMA, the Museum of Modern Art.

现在预订还能享受早鸟优惠价。

You can now lock in early bird pricing.

更多详情请访问我们的活动页面：theinformations.com/events。

We have more details on our events page at theinformations.com/events.

今天的第一条重磅消息是，《The Information》独家披露了Polymarket正面临的监管问题，不仅在美国，全球多地都在关注。

First up in today's show, The Information published exclusive reporting about the regulatory questions that Polymarket is facing, not just in The United States, but all around the world.

我请我们的财经记者来谈谈，客户身份识别对这家公司来说是个核心问题。

I'll bring on our finance reporter to talk about how customer identification is an issue that is front and center for the company.

接下来，我会和我们的加密货币记者聊聊，Polymarket 和 Kalshi 是如何抓住 SpaceX 首次公开募股的机会的。

I'll then be chatting with our crypto reporter about how Polymarket and Kalshi are capitalizing on the SpaceX IPO.

我还会采访客户服务平台 Freshworks 的首席执行官，探讨为什么人工智能反而让一些人工作更多了。

I'll also then speak with the CEO of IT and customer services company Fresh works about why AI is creating more work for some people rather than reducing it.

节目最后，我们会独家披露 Baseten 的一轮大额融资，以及开发者为什么在悄悄跟他们的 AI 助手说话。

And we'll wrap the show with some exclusive reporting about a big funding round from Baseten, and a look at why developers are whispering to their AI agents.

这期节目内容很精彩，咱们马上开始。

It's going to be a fun show, so let's get right on into it.

Polymarket 近期面临一些关于平台可能滋生内幕交易的质疑，而《The Information》今天独家发布的新报道指出，该公司在客户身份识别上的做法，长远来看也可能威胁到它的运营。

Polymarket has run into some questions around insider trading that might be growing on its platform, and a new exclusive story from The Information Today looks at how the company's approach to customer identification rules could also be a threat to the company's operations in the long run.

我请我们的财经记者迈克尔·罗丹来分享他发现的情况。

I want to bring on our finance reporter, Michael Roddan, to share more about what he found.

迈克尔，欢迎再次做客我们的节目。

Michael, welcome back to the show.

很高兴你能来。

It's great to have you here.

你好，阿卡什。

Good day, Akash.

好的。

Okay.

迈克尔，你是怎么决定要追这个故事的？

How did you decide to pursue this story, Michael?

Polymarket目前

Well, Polymarket's

正面临不少问题。

got a lot of problems at the moment.

它在市场份额上正被主要竞争对手Kalshi甩开，而Kalshi已经推出了U。

It's losing a lot of ground to its main rival, Kalshi, which has this U.

S。

S.

它的运营非常明确，清楚知道客户是谁、在做什么以及身在何处。

Operations where it knows exactly who its customers are, what they're doing, where they're located.

Polymarket 的情况就完全不同了。

Polymarket is very different.

它离开了美国。

It left The U.

美国。

S.

2022年的时候，政府基本上把它给禁了。

In 2022 when the government essentially banned it.

所以人们一直在用VPN访问Polymarket的全球平台，以此来隐藏自己的身份。

So people have been accessing Polymarket's global platform using VPNs that sort of shroud their identity.

这点很重要，因为Polymarket有很多必须遵守的规则。

This is important because Polymarket has a lot of rules that it needs to abide by.

比如第一条，不允许美国人在它的全球平台上进行交易。

Like one, not letting people in The US trade on its global platform.

还有像制裁法律这样的问题。

But also things like sanctions laws.

比如俄罗斯或伊朗这些地方，根本不能向那里的客户提供服务。

So people in say for instance Russia or Iran where it's it's no no to be able to serve customers in those places.

我发现，从这些地区开始在Polymarket上交易其实很容易。

What I found was that it's quite easy to start trading on Polymarket from these jurisdictions.

很多人在自己搭建交易机器人，专门服务俄罗斯客户。

A lot of people are setting up their own trading bots to run and serve Russian customers.

还有一些伊朗背景的人在指导人们如何访问这个平台。

There's Iranian influences that are sort of coaching people about how to access the platform.

这正逐渐成为一个大问题。

And this is becoming a big problem.

因此，Polymarket正在采取非常细微的措施，来识别使用其全球平台的交易者。

And so Polymarket is taking very tiny little steps to identify the traders who use its global platform.

目前有

There are

那Polymarket到底在采取哪些细微的措施呢？

some What are these what are these tiny little steps that Polymarket is taking?

几天前，他们在网站上开通了一个入口。

So so a couple of days ago, they they opened a portal on their website.

这个入口很低调，但上面写着：如果你想获得最快的交易速度，比平台上其他交易者快几毫秒，那就提交你的身份信息。

It's very sort of discreet, but it says, you wanna access, you know, our our the the fastest trading times, so that'll get you milliseconds ahead of your competitors who are also trading on the platform.

你需要提供你的身份证件。

You're going to want to submit your ID.

你需要告诉我们你是谁。

You're going to want to tell us who you are.

你还需要说明你身在何处。

You're going to want to say where you are.

如果你是企业客户，比如那些基于Polymarket构建的交易应用或接口，它们为这个平台带来了大量交易量，他们就需要了解你的投资者是谁，你的公司是哪种法律实体。

And if you're a business customer, like any number of the trading applications or interfaces that sort of build on top of Polymarket that is sort of driving a lot of volume to this exchange, they want to know who your investors are, what kind of entity your company is structured by.

目前来看，这还是一种自愿性质的措施。

And for now, this seems like a voluntary measure.

他们想慢慢引导人们提交身份信息，因为长期以来，这个交易平台一直像法外之地，这才吸引了不少用户。

They're trying to coax people to give over their ID because for a long time this exchange has been the Wild West and that's drawn a lot of customers to it.

很多人不愿意告诉Polymarket他们是谁，具体原因嘛，你得去问他们本人。

People don't want to let Polymarket know who they are for, I don't know, you'd have to ask them.

不过，当初这个消息公布后，网上一些论坛的评论提到，有人可能是想躲税。

But, you know, some of the comments in online forums when this was announced suggested that people were sort of wanting to dodge tax.

他们不希望国税局发现他们在做交易。

You know, they didn't want the IRS finding out that they were trading.

当然，还有大约三十个国家的用户被禁止使用Polymarket，但他们依然想登录平台，交易各种标的——比如天气、马斯克会发什么推文，还有伊朗局势的发展等等。

And, of course, there are people in, you know, 30 or so countries that are essentially banned from using Polymarket that still want to access there and trade on everything from, you know, the weather to what Elon Musk is going to tweet to the war in Iran and what happens there.

所以

So

KYC是‘了解你的客户’规则的缩写。

KYC is this acronym, Know Your Customer Rules.

这是银行常用的身份验证方式，我想很多交易平台也都在用。

This is the method of identification that banks use it, I think trading platforms use it.

听起来好像Polymarket并没有在用这套系统。

It sounds to me like So Polymarket is not using it.

它正在向这种身份验证靠拢吗？还是说你刚才说的这种情况，属于客户识别规则中的哪个阶段？

It's getting closer to that type of identification, or where does what you just described sort of fall in the spectrum of customer identification rules?

是的。

Yeah.

如果我们回溯得更早一点，这些规定最初是为了打击几十年前南美贩毒集团的洗钱行为而设立的。

So it's like if we want to go way back, these rules were brought in to fight sort of drug laundering from South American cartels decades ago.

政府希望银行能清楚地知道谁在使用他们的服务，谁在存钱，谁在取钱，以及他们这么做的原因是什么。

The government wanted banks to essentially really identify who was using their services, who's putting money in, who's taking it out, why are they doing that.

现在，这套规则几乎适用于所有类型的金融机构。

And this now applies to pretty much every financial sort of institution under the sun.

而Polymarket目前正在朝着更正规化的方向发展。

And Polymarket is in this process of going more legit.

它是由一群年轻小伙子在区块链上运营的。

So it's run by like these young guys who are operating on the blockchain.

它基于加密货币，而且是匿名的。

It's in cryptocurrency, it's anonymous.

这正是它的吸引力所在，非常有趣。

And that's part of the appeal, it's very fun.

但现在他们想成为一家大型受监管的机构，跟体育博彩公司和赌场竞争，就必须遵守KYC这类规定。

But now that they're trying to be like this big regulated institution that is competing with sports books and casinos, they also have to comply with things like KYC laws.

他们得清楚自己的赌客和投注者是谁。

They have to know who their gamblers and bettors are.

Polymarket 正在努力向一家正规金融机构靠拢。

Polymarket is trying to edge that way to try and become like a bona fide financial institution.

它正在筹建一个受监管的美国交易所，必须真实识别客户身份，了解他们到底在交易什么。

It's in the process of setting up this regulated US exchange where it has to actually identify its customers and know who they are and know what they're trading.

如果它想让全球平台在这些监管压力下继续生存，就必须把同样的规则应用到全球业务上。

And if it wants its global platform to survive these sort of regulatory problems, then it'll actually need to apply those rules to it as well.

因为现在越来越多的国家正在试图封禁 Polymarket。

Because you see increasingly a number of countries around the world are trying to ban Polymarket.

他们不喜欢这样。

They don't like it.

所以我们看到荷兰昨天或者前天刚禁止了它。

So we've seen The Netherlands ban it just yesterday or the day before.

西班牙也出台了新的禁令。

Spain put a new ban on it.

澳大利亚已经禁止了它。

Australia has banned it.

所以现在的问题是，如何在全球环境中运营这个平台，因为很多国家的政府根本不想让他们继续运营。

So it's like they're facing a lot of problems about how to operate this exchange in a global environment where a lot of countries, governments don't actually want them to operate.

那么，Polymarket 最终不得不实施客户身份识别规则吗？

So is it inevitable then that Polymarket will put in place these customer identification rules?

我是说，KYC，我不确定这是否只是这里的问题，你刚才提到

I mean, KYC, I'm not sure if it's just here, and you asked

我面临一个两难的处境。

me little a dilemma.

对。

Right.

所以我想问的是，这难道不可避免吗？

So guess what I'm asking is, is this inevitable?

我的意思是，Polymarket 要想继续扩大业务，是不是非得实行这些正式的身份验证规则？

I mean, Polymarket need to implement these formal identification guidelines to scale their business going forward?

现在还很难说。

It's really not clear.

他们得在让大量匿名交易者满意和合规之间走钢丝。

They have to walk a fine line between wanting to keep this huge rump of anonymous traders happy.

他们需要保持这笔交易量。

They want to have that trading volume.

靠这个赚取手续费。

They want to earn fees off that.

如果他们太快推行身份验证，那些人肯定会不爽，直接走人。

And if they go if they move too fast in instituting, you know, KYC, then these people are just gonna get pissed off and leave.

对。

Right.

但是

But

但另一个选择就是直接被关停，你什么都做不了。

But but the other option is that you just get shut down and you do any You

被关停。

get shut down.

但目前法律到底怎么规定、谁来执行，根本说不清楚。

But it's very unclear what the law is or who is going to enforce it at the moment.

你知道吗，实际上监管Polymarket的机构是商品期货交易委员会（CFTC）。

Know, it's like if you look at the actual regulator for Polymarket, it's the CFTC.

而这个机构对预测市场一直非常友好。

And that has been extraordinarily friendly to the prediction markets.

所以Polymarket和Kalshi都这样。

So Polymarket and Kalshi.

他们几乎是敞开大门欢迎了。

They basically rolled out the red carpet for it.

特朗普政府时期，政府基本上对这些公司的合规问题撒手不管了。

And the government under the Trump administration has essentially sort of washed its hands of sort of enforcing the law on these companies.

很多

A lot

美国各州政府曾试图通过法院来关闭Polymarket和Kalshi。

of the state governments in The US have tried to use the courts to shut down Polymarket in Kalshi.

而监管机构CFTC却站到了Polymarket和Kalshi这边，替它们打这些官司。

And the regulator, the CFTC, has actually come in on the side of Polymarket and Kalshi and is fighting those cases for them.

所以目前来看，法律最终会怎么定，还很不清楚。

So it's very unclear where the law is going to settle at the moment.

是的。

Yeah.

我有个问题，Mike，你一直关注银行业、金融机构、加密货币和预测市场，对吧？

Can I ask, Mike, you've been covering banking and financial institutions and crypto and prediction market?

你差不多把所有相关领域都覆盖到了。

You've sort of covered everything under the sun here.

根据Polymarket透露给客户的那些信息，你认为这件事最终会怎么发展？我在想加密货币，以及加密货币当初是怎么成长起来的。

What is your best guess here as to how this plays out based on what Polymarket is telling you know, the indications that they're giving to their customers, I'm thinking about crypto, and I'm thinking about the ways in which crypto did grow up in some ways.

这倒是有争议。

Debatably.

对，说得对。

Yeah, okay.

我应该说，他们其实是被逼着成长的。

I should say, they were forced in some ways to grow up.

他们当时遇到了不少危机。

They had crises on their hands.

里面还掺杂着欺诈行为。

There was fraud in the mix.

所以，会不会是某个突发事件，直接推动了这个领域的监管进程？

So, I mean, it likely that we have some sort of acute event here that basically pushes regulation forward in this arena?

归根结底，这些公司就是赌徒，他们在拿自己未来的生意做一场豪赌。

At the end of the day, these are gambling companies, and they're taking a massive gamble on the future of their business.

他们大概只剩下两年半的时间，等到特朗普政府任期结束。

They've got, what, two and a half years left of the Trump administration.

之后要是民主党上台，很可能会对这些公司严加管控。

You know, there's a chance that a democratic government will come in afterward and they will most likely move to really rein in these companies.

所以，他们只有两年半的时间，要么赶紧把事情理顺，要么就赌一把，觉得之后四年还能高枕无忧。

So, you know, they've got two and a half years to either get their house in order or just take a bet that things are going be hunky dory for another four years after that.

所以说，未来的事谁也说不准，但也许Polymarket和Kalshi上的人能猜得出来。

So yeah, can't predict the future, but maybe the people on Polymarket and Kalshi can.

这倒是个不错的收尾。

Well, it's a good place to end it.

这个故事讲得真精彩，迈克尔。

That is a great story, Michael.

非常感谢你来分享你的报道。

I want to thank you for coming on and sharing with us your reporting.

这是迈克尔·罗丹，我们《信息》杂志的财经记者。

That is Michael Roddan, our finance reporter here at The Information.

想了解更多关于预测市场的报道，我请来了我们的加密货币记者杨悦琪，她曾深入报道过Polymarket如何利用SpaceX上市的机会拓展业务。

For more predictions markets coverage, I want to bring on our crypto reporter, Yueqi Yang, who covered how Polymarket is using the SpaceX IPO as an opportunity for its business.

她今天在财经简报里写了这篇文章。

She wrote about that in today's finance newsletter.

悦琪，欢迎再次做客我们的节目。

Yueqi, welcome back to the show.

很高兴你今天能来。

It's great to have you here.

嗨，阿卡什。

Hey, Akash.

好的。

Okay.

我们刚和迈克尔聊完Polymarket面临的监管问题，这正是你作为加密货币记者一直关注的领域，我知道你对此非常熟悉。

So we just finished talking with Michael about the regulatory questions that Polymarket is facing, and that is certainly something I know that you are well versed in, given your crypto reporting career.

但我还想跟你聊聊SpaceX，这可不是你平时报道的领域。

But I want to talk to you about SpaceX, which is not a company that you typically cover on your beat.

预测市场是怎么提前押注SpaceX上市的呢？

How is it that prediction markets are looking ahead to the SpaceX IPO?

预测市场，尤其是Polymarket，正在提供各种投注选项，让投资者可以押注SpaceX、OpenAI、Anthropic这些即将上市的热门初创公司的估值。

Prediction markets, especially Polymarket, are offering bets that allow investors to bet on the valuation of SpaceX, OpenAI, Anthropic, all these hottest startups that are gearing up to go public.

我觉得这特别有意思，因为你知道，过去普通投资者很难接触到私营公司的股权，主要是因为这些公司信息披露非常有限。

And, I think this is very interesting because historically, as you know, retail investors have a hard time accessing, equity in private companies because of the limited disclosures that these companies make.

而Polymarket提供了一种替代方式，让用户能够表达对一家公司估值的看法，甚至从中获利。

And Polymarket is making this alternative way to effectively allow their users to express a view on the valuation of a company and to potentially benefit from it.

所以我深入研究了这个话题，也梳理了它们可能遇到的监管限制和障碍。

So I look into this topic and some of the potential regulatory restrictions and limits that they will run into.

目前这些投注大多只对美国以外的用户开放。

Most of these bets are only available outside of The US.

美国用户无法参与，因为这些合约太像证券了，如果在美国提供，很容易触犯美国的证券法规。

They're not available to US users, and that's because the contracts resemble too much as securities, and they can't really offer that in The US without running into troubles with US securities laws.

对。

Right.

监管问题确实是其中一面。

So the regulatory questions are one side to this.

我更想深入看看，预测市场所创造的这种关于IPO交易的平行世界。

I kind of want to dig a little bit deeper into sort of this parallel universe of trading on IPOs that prediction markets have created.

你能给我们简单讲讲吗？

Can you just give us a flavor?

举个例子，你在Polymarket上能对SpaceX或OpenAI的IPO下什么注？

What is a sample bet, I guess, that you could place on Polymarket as it relates to the SpaceX or OpenAI IPO?

就是你猜它上市时会值多少钱？

It literally what you think it's going to open at?

你可以赌Adobe到六月底的市值会是多少。

You can bet on what's the market cap for Adobe by the end of June.

上面有各种概率和不同的数字供你下注。

And there are all these probabilities and different numbers that you can bet on.

比如说，最受欢迎的押注之一是认为Adobe在六月底前市值会突破一万亿美元。

For example, I think the most popular one is showing that Adobe will reach more than $1,000,000,000,000 of market cap by the end of the June.

你也可以押注SpaceX在IPO后首个交易日结束时的市值是多少。

And you can also bet on, for example, what's the market cap of SpaceX at the end of the first day trading after its IPO?

不同的数值都有相应的赔率。

So there are odds for different numbers as well.

那这和直接买股票有什么区别呢？

So how is that any different than from like buying a stock then?

区别可大了。

There's a big difference.

如果你买股票，你就是真正的股东。

If you buy a stock, you're the actual owner of a Right.

我更想知道的是，这两种方式各自有什么优势？

And I guess maybe more what I get is, what's the benefit of one versus the other?

预测市场是不是更容易参与？

Prediction markets as easier to buy?

哪个更好呢？

Is it better?

这里的好处到底在哪？

What's the benefit here?

如果你买的是真实股票，你就真正拥有公司股份，享有所有投资者保护和权利，比如投票权和分红权。

If you buy a real stock, you get the actual ownership, which comes with all the investor protections and investor rights, such as the right to vote, the right to receive dividends.

但如果你在预测市场上下注，你只是在赌这家公司的估值走势，并不享有任何投资者权益。

If you bet on prediction markets, you are really just speculating on the price index for this company's valuation, and you don't have any investor rights.

不过，这些市场对海外投资者，尤其是那些过去无法投资私人公司股份的散户来说，更容易参与。

And, however, these these markets are accessible to offshore investors and especially retail investors offshore who historically do not have means to get access to private company shares.

所以各有优缺点。

So there's pros and cons to it.

不过你发的通讯里提到一个有意思的地方，比如纳斯达克这样的公司，已经开始参与这些与股票相关的预测市场产品了。

Now the interesting part that you pointed out though in your newsletter is that, I mean, you've got companies like NASDAQ, which are actually getting involved in the prediction markets products that are associated with these stock trades.

你能再详细说说，纳斯达克是怎么参与进来的吗？

Can you expand a little bit on how Nasdaq is getting involved?

纳斯达克正在提供私营公司估值的数据，Polymarket 将用这些数据来结算合约。

Nasdaq is providing the data on private company valuations that Polymarket will be able to use to settle their contracts.

这是一项他们最近达成的合作，让 Polymarket 能够引用纳斯达克的数据，比如说明年六月底 Anthropix 或 OpenAI 的估值是多少，并以此作为判定合约输赢的依据。

So this is a partnership that they entered recently, and that allows Polymarket to point to the Nasdaq data and say, this is the price for Anthropix valuation or for OpenAI's valuation at the end of June, and that's the price that they will use to decide who is winning the contract.

明白了。

Okay.

那纳斯达克为什么要这么做呢？

So why is Nasdaq doing this, though?

这会不会实际上是在引导人们不去买股票，而是转向预测市场交易？

I mean, this is conceivably a way to dissuade people from actually buying a stock, more towards just trading on a prediction market?

你觉得他们这么做的商业逻辑是什么？

What's the business logic here for them, you think?

我认为这并不会对纳斯达克的主营业务造成冲击，因为他们其实是在开拓一个全新的市场——主要是面向海外散户，而这些人过去根本没法投资私营公司股票。

Yeah, I don't think it necessarily cannibalizes with Nasdaq's business, because I think they're kind of tapping into this new market that's offshore, that's maybe retail oriented, that historically couldn't really get access to private company shares.

如果你是合格投资者。

If you're accredited U.

嗯。

S.

如果你作为投资者有正规渠道可以买这些股票，大概就不会去预测市场下注了。

Investors and you have proper means to buy these stocks, you probably wouldn't go to prediction markets to make the bet.

对。

Right.

所以你提到的那些预测市场上的投注，比如押注SpaceX上市之类的，

And so the key there is that so the bets that you're referring to on the prediction markets, betting on the SpaceX IPO, etcetera.

这些产品都只在美國以外的地方才有。

These are all products that are only available outside of The United States.

是的。

Yes.

这一点非常关键，值得强调。

And that's a very important part to highlight.

我觉得这也能解释Kalshi和Polymarket在合约设计上的差异。

And I think it shows up in the differences of contracts that Kalshi and Polymarket offers.

Kalshi 也提供与 SpaceX 和 OpenAI 相关的合约，但这些合约关注的是 OpenAI 会在哪个月宣布 IPO，或者会聘请哪家投行，而不是直接押注 OpenAI 或 SpaceX 的估值——那才更像证券。

Kalshi also offers contracts related to SpaceX, to OpenAI, but these are contracts that are about which months OpenAI will announce their IPO or which investment bankers they will hire, they're not exactly contracts on the valuations of OpenAI AI or SpaceX, which will make it more like a securities.

这说明，在美国运营的预测市场，很难推出让用户直接押注这些公司股价的产品，不管是上市公司还是私营企业。

So it shows that for prediction markets operating in The US, they're limited in their ability to push out products that allow people to bet directly on the prices of these companies, whether it's public company or private companies.

因此，它们在冲击美国股市方面的能力也受到了限制。

And as a result, they're limited in their ability to disrupt The US stock market.

所以你其实又把话题带回到你来之前我们和迈克尔·罗登聊过的那个点——预测市场面临的监管问题确实很多。

So you actually then connecting us back to this discussion that we had with Michael Roddan before you came on, the regulatory questions that predictions markets face are plentiful.

顺便说一句，这不光是美国的问题，其他国家也一样。

And by the way, this is not just in The US, this is outside of The US as well.

他还跟我们讲过，欧洲有些国家已经明令禁止预测市场了。

He was telling us about the countries in Europe, for example, that have banned prediction markets.

你可是已经追踪加密货币好几年了。

I mean, you've been covering crypto for quite a few years.

你觉得这类产品，也就是预测市场提供的这些，能撑得住吗？

Do you think that these products last insofar as prediction markets offering them?

我说的不只是美国，而是全球范围。

And I'm not just talking about The US, I'm talking about around the world.

我的意思是，会不会有一场清算，迫使它们收紧那些追踪股市、私有公司估值的特定产品类型？

I mean, is there a reckoning coming where they will have to rein in these specific types of products that are tracking the stock market, private company valuations?

你觉得接下来会怎么发展？

How do you see it unfolding?

对，我认为监管方面的动向将是我们重点关注的指标。

Yeah, I think regulatory development will be a big metric that we'll be watching out for.

在美国以外，

Outside of The U.

你会发现不同国家的监管机构开始对预测市场进行打压。

S, you start to see regulators in different countries to start to crack down on prediction markets.

有些人认为这些是未注册的赌博平台，而赌博在大多数国家都是非法的。

Some of them think these are unregistered gambling platforms, and gambling is not legal in most countries.

即使在美国，我也提到过，这些合约的表述方式有时可能会引发潜在的法律不确定性。

And even within The US, I mentioned that the way these contracts are phrased sometimes can run into potential legal uncertainty.

而且，这些合约到底算不算已注册的证券，目前还存在很大的模糊地带。

And there's a a huge gray area in terms of whether this contract is under registered securities or not.

比如，Kalshi 允许人们押注特斯拉每季度的交付量，但这本身并不直接等于押注特斯拉的盈利或股价。

For example, Kalshi allows people to bet on the quarterly car delivery numbers of Tesla, And that itself is not necessarily a bet on the earnings of Tesla or the stock price of Tesla.

不过，我们都清楚，投资者对汽车交付量的关注度非常高。

However, as we all know, investors watch the car delivery number very closely.

这个数据确实有可能影响到他们的盈利表现和股价走势。

This is a metric that could potentially affect their earnings and their stock prices.

所以，这些都属于法律上的灰色地带，得看监管机构最终怎么界定，这类合约是否允许在预测市场中出现。

So these are legal gray areas where we'll need to see what regulators say in terms of whether these are even allowed to be offered on prediction markets.

很好。

Great.

Yueqi，非常感谢你今天来做客。

Well, Yueqi, I want thank you for coming on.

这位是《信息》杂志的加密货币记者杨悦琪。

That is Yueqi Yang, our crypto reporter, here at The Information.

一家名为Freshworks的IT与客户服务公司发布了一份新报告，指出AI在某些方面反而让工作变得更复杂了。

A new report from IT and customer service company Freshworks suggests that AI in some ways could be making work more complicated rather than less.

我想请Freshworks的首席执行官丹尼斯·伍德赛德来深入聊聊这家公司的一些观察。

I want to bring on Freshworks' CEO, Dennis Woodside, to unpack some more of the company's observations.

丹尼斯，欢迎来到TITV。

Dennis, welcome to TITV.

很高兴你能来。

It's great to have you here.

嗨，各位。

Hey, guys.

见到你们真好，谢谢邀请我。

Good to see you, and thanks for having me.

你们今天发布了一份挺有意思的报告，探讨了AI带来的任务复杂性问题。

So you guys put out a pretty interesting report today looking at the complexity tasks that have come up with AI.

报告里有不少有趣的统计数据。

And look, there were a number of interesting stats in there.

让我特别在意的一点是，你们提到管理AI反而增加了本应减轻的工作负担，团队得不断修正AI输出的错误，还要应对越来越多的AI工具带来的管理混乱。

The line that jumped out to me that I want to get your color on here is you guys wrote that managing AI is now adding to the workload it was meant to reduce, with teams fixing flawed outputs and governing tool sprawl across a growing stack of AI products.

也就是说，事情变得比预期复杂得多，而不是简单了。

So this idea that things are getting much more complicated, not much less complicated.

在你们对IT负责人做的这项调查中，这个结果让你们感到意外吗？

Did this surprise you in this survey that you did with IT leaders?

确实没想到。

It really did.

我每周都会和CIO们交流，拿一家大约五千人的中型企业来说，他们可能正在使用二十到三十个不同的AI应用，或者在现有系统里整合了多个AI模块，全都要去管理和协调。

I talk to CIOs every week, and the average CIO in a mid market organization, think of a company with about 5,000 employees, they can have 20 or 30 different AI applications or elements of AI in existing applications that they're trying to manage and trying to to deal with.

想想看，围绕这些工具的治理工作有多繁重。

Think about the governance tasks around that.

再想想，还得培训团队，让他们真正理解这些工具能带来什么价值。

Think about just training your teams to understand what all this actually could do for you.

AI的出现当然带来了巨大的潜力。

So with AI coming in, there's a huge amount of promise.

AI带来的潜力巨大，但这种复杂性是实实在在的。

There's a huge amount of upside from AI, but this complexity is real.

我们的目标就是减轻这种负担，打造一款高度整合的产品体验，无缝衔接他们已经在用的员工服务系统、IT管理工具或客户支持平台，让AI真正变得更容易使用、更快上手、更快产生价值。

And what we seek to do is to reduce that and to offer, you know, a a a product experience that's very well integrated with what they already are using for either their employee service operations, managing IT, or their customer support operations so that the AI is actually easier to use, faster to get up and running, you get value from it much faster.

那这种复杂性到底从何而来？

So where is this complexity coming from?

是管理层施压，要求必须用上这些AI工具吗？

I mean, is this pressure from management saying, Hey, we have to find a way to use these AI tools.

我们愿意花钱。

We are willing to spend.

我们听说过token消耗的波动，还有需求起伏的情况。

We've heard about token maxing, the ways in which that's fluctuated in terms of appetite.

这种复杂性到底从何而来？

Where is the complexity coming from?

是因为大家根本不知道怎么用这些工具吗？

Is it people just not knowing how to use these tools?

你试着站在首席信息官的角度想想。

Well, put yourself in the shoes of a CIO.

他们上面有来自CEO和董事会的压力，要求他们采用人工智能，提升效率，像其他公司那样从AI中获益，提高产出，增强为客户提供服务的能力。

They're getting pressure from above, from their CIO and their board, to adopt AI, drive the efficiency that we're seeing others get from AI, improve our output, improve what we actually can deliver for our customers from AI.

这种压力来自高层，是一种必须马上行动的硬性要求。

That's coming from the top, and that's like a do something now type of imperative.

而与此同时，底层也在施压——很多AI工具，员工自己就能注册使用，比如Cloud Cowork或者其他类似产品。

And then they're getting pressure from the bottom where a lot of these AI tools, you can just go and sign up for a Cloud Cowork or a number of these other products.

所以员工们会说：我需要一个能帮我把工作做得更好的工具。

So employees are saying, Hey, I need something to do my job better.

我不想被甩在后面。

I don't want to be left behind.

所以我干脆自己刷卡，买个能用的解决方案先用起来。

So I'm going to go out and use my credit card and buy whatever solution I can find.

这对首席信息官来说确实很头疼。

So that's tough for a CIO.

然后所有的供应商都跑来说，我们的产品现在也内置了AI功能。

Then all the vendors are coming in and saying, Hey, our products now have AI as well.

所以我觉得，压力是从四面八方涌来的。

So I think it's coming in all directions.

我们现在正处在一片混乱但充满创新的领域，事情太多，CIO们必须把这一切都管理得井井有条。

We're in that messy, innovative space where there's so much going on that CIOs have to manage all those things really well.

这确实是个很难应对的局面。

It's a tough spot to be in.

最出色的那些公司，都在仔细审视眼前的一切，专注于到底想通过AI实现哪些具体的业务成果。

The best organizations really are taking a look at everything that's going on and focusing on what business outcomes are we trying to drive.

我们是想减少创意人员在非创意工作上花的时间吗？

Are we trying to reduce the time that our creative people spend on non creative tasks?

还是说，我们单纯想直接降低成本？

Are we trying to reduce costs outright?

他们正在制定实实在在的业务计划，明确使用哪些工具来推动业务目标的实现。

And they're creating real business plans with specific tools in mind for how they advance business goals.

他们先从商业目标出发。

They're starting with the business goal.

他们不是一上来就说：咱们先搞个AI吧。

They're not starting with, Hey, let's just do AI.

没错。

Right.

那我再确认一下。

So I just want to confirm then.

根据你们的调查，这种复杂性主要是针对IT负责人和CIO的。

So in terms of what your survey found, this is complexity for the IT leaders and the CIOs.

并不是指普通员工，也就是基层员工的工作负担，以及他们如何

This is not referring to workloads for the employees, sort of the lower level employees, and how they're

那对员工来说也不容易。

Well, that's hard too.

这项调查的对象明确是那些身处AI变革核心的IT决策者。

I mean, the survey specifically was of IT decision makers who are really in the middle of this AI revolution.

但确实，对员工来说，用哪个工具做哪种工作、应对什么场景，这也是个问题。

But it's true that for employees as well, which tool do I use for what work for what use case?

这同样是个挑战。

That's a challenge too.

确实如此。

Yeah.

所以我们特别关注的是中型市场和敏捷型企业。

So where we're particularly focused is what we call the mid market and agile enterprises.

我们认为这类公司尤其面临风险。

We think those companies in particular are really at risk.

它们不是那种超级大型企业。

These aren't the super giant corporations.

而是员工规模在五千人左右，最多两万人的公司。

They're 5,000 person companies, maybe up to 20,000 person companies.

比如我们的客户新百伦。

So think about New Balance, which is a customer of ours.

他们有七千名员工。

They've got 7,000 employees.

他们要和拥有九万名员工的耐克竞争。

They compete against Nike, which has 90,000 employees.

对新百伦来说，他们得跟大公司抢货架空间、抢用户注意力、抢客户。

So for New Balance, they have to compete for the same shelf space, the same mindshare, the same customer.

他们必须更加灵活敏捷。

They have to be much more nimble and agile.

这种复杂性可能会严重拖累他们。

And this complexity potentially really hurts them.

所以他们需要找那些能帮他们简化流程的供应商和合作伙伴，这正是我们的工作。

So they need to seek vendors and partners that are going to try to reduce that complexity, and that's what we do.

我们的产品就是为这个目标打造的。

That's what our products are designed to do.

我们把人工智能整合到了他们已经在用的产品里。

We've built AI into the products that they're already using.

上手和使用都非常简单，这才是真正的价值所在。

It's super easy to get up and running and use, and that's where the value really is.

报告里你还提到了另一个概念，就是影子AI。

Now the other thing you talked about in the report is this concept of shadow AI.

那影子AI到底是什么？

What what is shadow AI?

举个例子，比如员工觉得与其走公司流程，不如自己掏钱买一个专业版的OpenAI GPT或者Claude，用来提升工作效率。

Well, that's the example where a an employee decides I'm gonna just, you know, improve my how I'm doing my job by buying my own AI, by going and buying a professional edition of OpenAI GPT or Claude.

这种情况也很常见。

And you see that a lot too.

不能怪员工。

You can't blame employees.

他们想加快进度，把工作做得更好。

Want to move fast and do their jobs better.

但这给IT部门带来了挑战，他们需要管理好这些情况，确保组织的信息得到保护，确保AI只能访问公司允许它访问的内容。

But that creates challenges for the IT department in terms of governing that, ensuring that the organization's information is protected, ensuring that they have access, the AI has access only to what the company wants it to have access to.

所以这又是一个落在CIO桌上的新挑战。

So that's just another challenge that lands on the desk of the CIO.

在你们公司Freshworks，这种情况严重吗？

How much of this is a problem within your own company at Freshworks?

挺有意思的。

It's interesting.

我们公司里光是AI的应用就有四十多个了。

We have easily over 40 different applications of AI in our organizations.

这些不是单独的软件或大模型，而是真正带来价值的实际应用场景。

These aren't separate pieces of software or LLMs, but actual applications where we're getting value.

我们同时采取了自下而上和自上而下的方式，让各个团队有权自主尝试AI，提升工作效率。

And we've taken both a bottoms up and a top down approach where individual teams are empowered to adopt AI and try to do their jobs better.

我们鼓励他们明确具体的业务目标和预期成果。

We encourage them to have a specific business case, business outcomes in mind.

同时，公司层面也有整体目标，比如在全团队范围内部署AI。

And then we have more company based goals where we're deploying AI across an entire team.

比如，我们的工程团队用AI来写代码。

For example, our engineering organization uses AI to code.

现在我们一半的代码都是AI生成的。

About half of our code now is originated from AI.

我们的设计团队，几乎所有的设计工作都是从AI开始的。

Our design team, almost all of their design work is done in AI to start.

你们主要用的是Claude还是Codex？

That's Are you using more is it Claude code or Codex?

现在哪个更占优势？

One is the winner right now?

我们之前用过

We've used

全都用过。

them all.

我们最早用的是GitHub Copilot。

So we started with GitHub Copilot.

我们用过 Cursor。

We have used Cursor.

我们现在正在评估 Claude。

We're evaluating Claude now.

我们试过所有这些工具，一直在做实验，因为我们有将近两千名技术人员。

So we've we've used all the variations, and we're always experimenting because we have close to 2,000 technical employees.

我们能做到这一点。

We can do that.

但我们会持续为每个团队寻找最适合的工具，因为这对我们业务太重要了。

But that's we're always gonna be looking for the tool for that specific team because it's such an important part of how we do business.

听起来在编程工具上，你还没选出一个明确的赢家，这最让人兴奋了。

That is the most exciting on one tool, sounds like like in terms of coding, you you don't have a winner just yet.

你们还在不断尝试。

You're experimenting

不，还没定下来。

with No, not yet.

不是。

No.

我们确实在用。

We yeah.

短时间内，我们已经取得了很大进展。

In a short period of time, we've really moved.

我们从 GitHub Copilot 转到了 Claude。

We've moved from GitHub Copilot to Claude.

而且，只要有什么新工具出来，我们都会去试，只要能提升工程师的工作效率。

And again, we experiment with whatever is coming out, whatever will make our engineers' lives more productive.

对。

Right.

我想问你个稍微不一样的问题。

I want to ask you a slightly different question.

你曾经在 ServiceNow 的董事会待了几年。

You were on the board of ServiceNow for a couple years.

你几年前离开了，但我还想问问你，在人工智能时代，人才流动是个什么情况。

You left a couple years ago, but I just wanted to ask you broadly about talent transfer in the AI era.

我之所以想到这个，是因为我们看到ServiceNow的首席营销官科林·弗莱明这周跳槽去了OpenAI。

The reason this comes to mind is because we saw ServiceNow CMO, Colin Fleming, left to go to OpenAI this week.

几年前，我记得他们的首席商务官还去了Go2Anthropic。

A couple of years ago, I believe, the Chief Commercial Officer went to Go2Anthropic.

我想请你，丹尼斯，从整体上说说，当有这么多吸引人的AI公司、初创企业都在抢夺有经验的领导者时，传统企业软件公司该怎么留住自己的顶尖人才？

And the broad question I just had for you, Dennis, is how do enterprise software companies keep their best talent working at the enterprise software companies when there are these shiny AI companies, startups even, that are eager for leadership and seasoned executives?

我的意思是，企业软件公司到底该拿什么打动员工，让他们留下来，而不是跑去创业公司之类的地方？

I mean, what is the pitch from the enterprise software sector to keep their talent here as opposed to going to startups and stuff like that?

对我们来说，我们有七万五千个客户。

Well, think for us, we have 75,000 customers.

所以如果你是个工程师，你今天写的代码，其实能间接影响数百万人的生活。

So if you're an engineer, you have the ability to write code today that impacts millions of people indirectly.

另外，你要是了解我们公司，就会知道我们85%的员工都在印度。

We also When you think about our company, 85% of our employees are in India.

在印度，大多数软件公司只做后台支持。

Most software companies in India, you get the back office.

真正写代码、直接影响客户日常体验的核心开发人员，他们并不在这些公司。

You don't get the people who are actually writing the code that's going into production that's impacting the customer every single day.

但我们不一样。

We do.

所以我们能吸引那些希望在全球范围内创造实际价值的人才。

So we can recruit people who want to deliver value at a global scale.

这对我们来说是个巨大的优势。

That's a huge advantage for us.

现在很多人也更愿意加入那些能让他们看到实际成果的公司。

A lot of people also, want to be working with companies now and see the value.

想想软件行业，我们正在做的事情。

You think about the software industry and what we are doing.

我们正把人工智能融入到每一项业务中。

We're betting AI into everything that we do.