Atlassian CEO谈SaaS行业剧变、AI代理与未来趋势

Intuit 是在为你填写税表，还是说Intuit对税法的了解和其他人一样深入？

Is Intuit filling out the tax code for you, or does Intuit know the tax code as well as anyone else can?

他们帮助你的是，利用你的生活数据、你的理解，向你提出正确的问题。

What they're helping is you to take your life data, your understanding, your they're asking you the right questions.

Intuit 几乎更像一家麦肯锡公司。

Intuit's almost more like a McKinsey.

可以这样看待它。

It can be considered that way.

这是他们的流程，他们的独特能力在于如何向你提出正确的问题，从而填写税表，而不是直接帮你填写税表。

It's their process, and their special ability is how to ask you the right questions to fill out the tax code rather than the filling out of the tax code.

是的。

Yeah.

而且

And

所有这些企业都不得不思考：也许我内部有50个流程，我一直认为那是我的核心机密和独特优势。

all these businesses are having to look at, maybe I have 50 processes internally that I think are my secret sauce and unique.

也许只有其中20个是独特的，但现在我必须认真思考，这些流程中哪些才是真正独特的，哪些不是，因为我们以前从未以这种方式思考过。

Maybe only 20 of them are, but now I have to really consider which of those processes are actually unique and which which are not because we haven't had to think about it in that in that manner before.

我认为这其实也是一个问题，即可能存在一个恰到好处的平衡点——自己做是否值得，还是不值得？

I think it's also kind of a question of how like, there's this Goldilocks zone probably of, like, is it worth doing yourself versus not?

比如，把这个所谓的‘非禁忌但独立的变量’考虑进去：我是否应该自己编写某个东西的代码？

Like, If you take this kind of third not third rail, but kind of this independent variable of, should I now Claude code myself Claude code myself some x?

如果这个因素占了我99%的成本，而且因为这家邪恶公司对软件收费过高，我的业务即将失败，那可能就值得自己做。

Well, if it's 99% of my cost and my business is gonna fail because this evil company is overcharging me for software, it might make sense.

但如果一年只花一美元，那大概就不值得了。

If it's a dollar a year, it probably doesn't make sense.

而且，并非所有的记录系统都是一样的。

And then not all systems of record are the same.

我倾向于把记录系统看作是企业中某个事物的基本单元。

I kind of think of a system of record as like the atomic unit of something for a business.

比如，日历可以是时间的记录系统，或者我不知道，ERP系统是库存的记录系统。

Like, it could be calendars or a system of record for time, or I don't know, ERP is a system of record for inventory.

你有这么多不同的记录系统。

Like, you have all these different systems of record.

但我之前举的例子是，如果我在迈阿密有一个办公室，我很少去，那里有一个会议室的记录系统。

But, like, the the example I was giving somebody is if I have an office in Miami that I don't go to very often, and there's a system of record for conference rooms.

有一个会议室的记录系统。

There is a system of record for conference rooms.

就像是谷歌日历。

It's like Google Calendar.

我愿意更换这个记录系统吗？

Like, am I willing to change that system of record?

是的。

Yeah.

因为我的迈阿密办公室，他们一年才聚一次。

Because it's like my Miami office, they only gather once a year.

谁在乎呢？

Like, who cares?

而这是直接影响我收入的事情。

Versus like, this is something that touches my revenue.

这并不贵。

It's not that expensive.

我真的会自己种食物来处理这种事吗？我的意思是，这其实正是农业这个隐喻的有趣之处，对吧。

Am I really going to grow my own food for something where I mean, actually, this is the cool thing about farming, right, as you kind of take that metaphor.

去餐厅吃饭实际上便宜多了。

It's actually a lot cheaper to go to a restaurant.

如果我只是想吃一个汉堡，而不是买一头牛、喂养它、等它长大——其实，很多食物在餐厅吃反而更便宜，因为比较优势和规模经济。

If I just want one hamburger versus, like, get myself a cow and feed the cow and wait to it's just a lot of food is actually cheaper if you consume it in a restaurant because of comparative advantage and economies of scale.

所以，可能确实存在一些系统记录，除了我们讨论的这些因素之外，它们更容易被取代，因为它们定价过高，或者它们所存储和记录的内容本身价值不高。

So there probably are systems of record where it's like there's some where outside of any of the factors that we're talking about, they're more susceptible just because they overpriced or they're just not as valuable in terms of what it is that they're storing and keeping records for.

我的意思是，Carta 为很多公司管理股权结构表。

I mean, like, Carta keeps track of cap tables for a lot of companies.

你多久访问一次你的股权结构表？

How often do you access your cap table?

不太频繁，但它非常有价值。

Not very often, but it's super valuable.

你不能搞砸了。

You can't f that up.

对吧？

Right?

比如，我可能更愿意用Carta来处理，而且他们也没收我多少钱。

Like, it's I'd probably rather use Carta for that than like and they only they don't charge me that much money.

当然。

Like, sure.

我会用Carta。

I'll use Carta.

而且它并不是那种日常使用的工具。

And it's not it's not like a daily use kind of product.

所以甚至都不是这个维度的问题。

So it's not even like that dimension.

我觉得‘振动编码’这个概念特别有趣，因为作为软件行业的人，他们会觉得，哦，人们会用振动编码来替代所有这些工具。

I think the vibe coding thing is so fascinating to me because, yes, as someone in software, they're like, oh, people are just gonna vibe code all these replacements to tools.

我觉得，要是让我用振动编码来自己写一个Workday然后运行它，简直太可怕了。

I'm like, the idea I would vibe code my own Workday and then run it is terrifying.

我有一些非常聪明的工程师。

I have some really smart engineers.

首先，我还有其他事情要他们去做。

Firstly, I have other stuff for them to do.

其次，我想说，等等。

Secondly, I'm like, wait.

我觉得这对我的利远小于弊。

I feel like that has way more downside than upside for me.

不过，这正是所谓的替代理论。

However and so that's the sort of replacement theory.

但我们在内部确实看到了使用振动编码等方式提升软件可扩展性的巨大优势。

There is a great gain we are seeing internally in extensibility of software using things like Vibe coding.

所以这些应用程序大多高度可配置、可定制，在我们的情况下，这通往真正的可扩展性。

So most of these applications are highly configurable, customizable, you know, in our case, the way through to true extensibility.

你可以编写运行在我们平台之上的软件应用，涵盖各种不同的领域。

You can write pieces of software, apps that run on top of our platform that have all sorts of different areas.

很多客户确实这么做了，但这些客户需要投入一个技术团队来完成这项工作。

And lots of customers do, but those customers need to put a technology team on doing that job.

他们能够所谓的‘随性编码’扩展、定制，开发出针对特定使用场景的非常个性化的应用。

Their ability to quote unquote vibe code extensions, customizations, very tailored applications with a very specific use case of something.

我想为迈阿密团队开发一个会议室预订应用，而迈阿密有一些奇特的人力资源政策，所以这个应用需要对接Workday和其他系统。

I want an app for the Miami team to do conference room booking, and Miami has some weird HR policy, so that app needs to look at Workday and this and that.

这个应用只有20个人使用。

It's used by 20 people.

我原本可能负担不起让内部IT团队专门去开发这个应用，因为成本太高了。

I probably wouldn't have been able to afford to put the IT team internally on building that because the bill would have been too big.

但现在，也许我自己就能开发出这个应用。

But now maybe I can build that.

对吧？

Right?

但这使用了Workday在全球范围内的数据和规则，只是为迈阿密前台的员工提供了一个非常定制化的界面，以完成他们特定的需求。

But that uses Workday's data and rules around the world underneath, but it it just gives me a very custom interface for, I don't know, the person on the front desk in Miami to do something very specific to what they need.

这非常强大，但它并不能取代Workday。

That is super powerful, but it's not a replacement for work poor Workday.

我觉得阿尼尔经常成为这些概念性例子中的笑柄。

I feel like Aneel is like the the butt of a lot of these conceptual examples.

这真的非常强大。

That's really powerful.

对吧？

Right?

这实际上让Workday在企业中更具粘性、更有价值，因为你可以在其上构建所有这些应用程序，这正是AI、低代码和创造力的力量，能让系统更贴合我的需求。

That actually makes Workday stickier in the enterprise and more valuable because you can build all these applications on top, which is the power of AI and vibe coding and creativity to make it more tailored for what I need.

但我们必须非常谨慎地处理稳定性、规则和流程与定制化之间的这些层次关系。

But we're gonna have to be really careful about these sort of layers of stability and rules and process versus customization.

对吧？

Right?

你可以说，像OpenCloren这样的东西就是为我个人打造的专属应用的例子。

And you could argue, I don't know, OpenCloren stuff is an example of building very personal apps just for me.

这些人大部分都不是软件开发者。

Most of those people aren't software developers.

他们只是在Gmail或其他平台之上，为自己构建一些只适用于自己的应用。

They're building apps that work just for them on top of their Gmail or something else.

对吧？

Right?

但它仍然以Gmail作为基础框架。

But it still uses Gmail as the rails.

他们依然会去Gmail收发邮件，但会为自己构建某个特定的功能来解决只有自己才有的问题。

They still go to Gmail to read their email and do their email, but they build some specific thing for themselves to solve a problem they have and probably only they have.

其中少数几个可能最终发展成公司。

A couple of them maybe turn into companies.

他们中的大多数只是在解决自己需要的问题。

Most of them are just solving some stuff that they needed themselves.

就是这样。

That's it.

这很棒。

And that's great.

这非常强大。

That's really powerful.

这就是为什么我对所谓的定价公平性第二类感到好奇，即后端不等于前端。

That's why I'm curious about, maybe I'd call it my bucket two of this pricing fairness where the back end is not the front end.

以Salesforce为例，他们按许可证收费。

So if you think of Salesforce, they charge for licenses.

我想我们公司有600人，可能就有600个Salesforce许可证。

Like, I think we have 600 people at our firm, might have 600 Salesforce licenses.

我从未登录过Salesforce，但我敢打赌我们也在为我付费。

I've never logged into Salesforce, but I bet we pay for me.

但我有时会使用它的输出，因为它确实是我们的主数据系统，虽然我不太想过度使用这个词，但它存储了我们所有的关系。

But I I use the output of it sometimes because it actually is the system of record, not to overuse that term, but it stores, like, all of our relationships.

而我，就像是关系型数据库中的一张表里的一个条目。

But I am, like, part of a table in a relational database.

所以，你知道，我是用户ID 422。

So it's like, you know, I'm user ID number 422 here.

每当我与一家公司接触时，哦，用户ID422在另一个数据库里被匹配到了，但我们其实只是想为这个数据库付费。

And then whenever I meet with a company, oh, well, like, user ID four twenty two is matched in this other database, but we really just wanna pay for a database.

所以在一个前端和后端不一致的世界里，这正是关键所在。

So, like, in a world where the front end is not the back end, I mean, that that's the thing.

对于Workday，我觉得他们想出了一个非常巧妙的定价策略。

It's like, for Workday, I kind of think they've come up with a very clever pricing trick.

用‘技巧’来形容它都低估了它的精妙。

The trick trick undersells it.

我的意思是，这是一种强大且令人感到公平的定价模式。

Mean, I think it's a it's a it's a powerful pricing paradigm that feels fair.

这就像员工越多越好，为什么这样才公平？

It's like the more employees that you have and why is that fair?

因为通用电气的利润比一家十人公司要高。

Because GE has more profits than a 10 person company.

通用电气会为这个东西支付更多钱。

GE's gonna pay more for this thing.

这仍然是九牛一毛。

It's still a drop in the bucket.

这完全处于定价的黄金区间。

It's totally within the Goldilocks zone of pricing.

我不认为有任何人在期待他们会带来所有这些AI收入。

And I don't think anybody's in a vibe code that they're gonna add all this AI revenue.

但最重要的是，他们的定价感觉很公平。

But most importantly, their pricing feels fair.

而对于那些前端与后端 somewhat 脱节的产品，我不知道什么样的定价方式才算公平。

Whereas for these things where it's like the front end is somewhat divorced from the back end, that that that one is I I I don't know what's the fair what's the fair format for pricing.

那么，软件定价会发生什么变化？

Like, what will happen to software pricing?

而且，显然，如果没人会自己写 vibe code，也没有任何竞争，那么定价就会保持不变。

And and, obviously, like, if nobody's gonna vibe code their own thing and there's not gonna be any competition, then pricing will stay unchanged.

但你可以想象一个世界，人们会在上面构建应用来读取数据库。

But you can imagine a world where people are building things on top to read from the database.

对吧？

Right?

因为，我的意思是，一个记录系统背后都有一个数据库，这就像一切之下的抽象层。

Because, I mean, a system of record has a database represent that that's like the the abstraction layer beneath everything.

定价会在这些类别中受到任何压力吗？

Will the pricing will there be any pricing pressure on on any of these categories?

对我来说，我认为如果前端和后端是分离的，那么相比它们高度紧密交织的情况，更容易受到定价影响。

And I I for me, I think it's like if the front end is not the back end, there's more susceptibility than if they're like very, very tightly like, intertwined.

比如，QuickBooks 被小企业广泛使用。

Like, QuickBooks is used by small businesses.

他们没有座位。

They don't have seats.

就像是企业的老板直接用QuickBooks。

It's like the owner of the business just lies into QuickBooks.

所以前端某种程度上就是后端，而像Salesforce这样的系统，你可以想象，没人会放弃Salesforce，但可能因为需要的前端更少，所以座位数减少了，但他们依然极度依赖后端。

So the front end kinda is the back end versus, you know, Salesforce, where you can imagine, like, nobody gets rid of Salesforce, but maybe they have fewer seats because they need fewer front ends, but they really still need the back end desperately.

他们不会去消除或对后端做任何改动。

They're not gonna go you know, they're not gonna eliminate or do anything with the back end.

这始终取决于公平性和定价的观感，我认为这非常重要。

It depends on always like I think your fairness and optics in pricing are really, really important.

人们需要清楚自己付钱买的是什么，并且觉得所付的费用与他们的使用量在大体上是相关的。

People understanding what they pay for and feel like what they pay for is relates to their usage in some broad way.

对吧？

Right?

我会说，一家一万人的公司支付Workday的费用，而两万人的公司可能支付两倍的费用，再加上一些折扣，因为他们购买得更多，而且通常他们的系统复杂度也翻倍，他们觉得这样是公平的。

I I would say that a 10,000 person company paying for Workday, the 20,000 person company probably plays twice as much plus on discount because they're buying more because they generally have twice as much complexity of stuff, and they see that as fair.

这就是你所说的，按员工数量为我的人力资源系统付费似乎很合理。

That's what you mean by, like, it seems reasonable that I would pay by employee for my HR system.

我认为很多这类问题的关键在于，当我们谈论前后端时，比如，它并不是一个数据库。

I think the question with a lot of these things is, you know, what what processes when we talk about front end and back end as an example, it's not a database.

它是一个数据库加上一组流程。

It's a database plus a set of processes.

我年轻的时候，我们称之为业务逻辑。

We used to call it business logic when I was growing up.

这些业务逻辑并不是无关紧要的。

The those business logics are not irrelevant.

那么，在为什么企业需要它们的世界里？

So in the world of what why does a business have them?

因为企业运作是一系列流程的集合，他们希望在某种程度上实现流程标准化。

Because it runs as a collection of processes, and they want standardization of process to some level.

对吧？

Right?

所以两个团队的工作方式是一样的。

So the two teams work the same way.

所以有人可以管理它们，理解它们，跟踪产出。

So someone can manage them, understand them, track output.

你知道，如果我拥有一堆汽车工厂，我想持续跟踪每个工厂的汽车进出总量。

You know, I don't know if I have a bunch of car factories, I wanna track the total amount of cars in and out consistently across them.

业务逻辑被嵌入的地方，某种程度上正是价值所在，因为你可能需要——再次说明，a16z可能并不是Salesforce客户的典型例子，毕竟它在传统意义上的销售量并不大。

The business logic where it gets baked in is somewhat where the value is because you may need, and again, maybe a16z is not a great example of a Salesforce customer, right, that actually has a huge amount of sales going on in terms of traditionally.

你为销售团队嵌入的这些流程对你来说非常有价值，你也会觉得这样付费是合理的。

The processes you bake into that for your sales teams are totally valuable to you, and you would think that's a fair way to pay.

问题是，那些与销售相关的支持团队，作为协作者而非核心用户，他们对这些流程的需求有多大，又有多小？

The question is your sales adjacent teams, the sort of collaborator rather than the core user, how much do they need those processes and how much do they not?

所以我不确定。

So I don't know.

我猜我们讨论的是Salesforce Sales Cloud。

I assume Salesforce Sales Cloud, I guess we're talking about.

Sales Cloud 有一个 MCP 服务器。

Sales Cloud has an MCP server.

这个 MCP 服务器并不连接数据库。

That MCP server doesn't go to the database.

它可能涉及你流程中的各种规则和流程。

It probably involves your processes and the rules on the way through.

所以问题在于那些与销售相关的人员。

So the question is someone sells adjacent.

我不确定。

I don't know.

他们可能在市场部，或者在客户成功部，类似这样的部门。

They're in marketing or they're in customer success or something like this.

如果他们需要这些流程、治理、控制和规则，比如，我们只对日本的客户做 x 操作。

If they need those processes and governance and controls and rules and, you know, hey, we only do x for customers in Japan.

我们对这个地区的客户做 y 操作，诸如此类的事情。

We do y for customers in this area, that sort of stuff.

甚至连他们的MCP服务器都需要一个账户。

Even their MCP server is gonna need an account.

客户是否认为这公平，那是另一个问题。

Whether the customer thinks that's fair, that's a different question.

对吧？

Right?

这仅仅是关于如何定价的挑战。

It's just the challenge of like, how does that get priced?

我认为，因为我们经常讨论基于消耗的定价、基于使用量的定价和基于成果的定价。

I'd say, because we get this all the time talking about consumption based pricings, usage based pricing, outcome based pricing.

在很多领域，这种定价方式是合理的。

There are a lot of categories where that makes sense.

我绝对不认为它会成为所有软件、所有SaaS软件的主要定价方式。

I definitely do not believe that it will be the majority pricing manner for all software, for all SaaS based software.

因为当你和客户交谈时，他们讨厌这种定价方式。

Because when you talk to customers, they hate it.

他们真的很讨厌这种模式。

They really hate it.

当——*——这种定价方式与他们认为自己所获得的价值无关时。

Where, asterisk, it is not related to the value they consider that they put in.

所以我对Splunk采用了按使用量计费的模式。

So I have usage based pricing for Splunk.

如果我发送的日志量翻倍，我就要付更多的钱。

If I send them twice as many logs, I pay more money.

我明白。

I get it.

但日志的多少是由我决定的。

But the logging is up to me.

对吧？

Right?

我可以多记录，也可以少记录。

I can log more or I can log less.

我可以对团队大喊：嘿，你们为什么录了这么多日志？

I can yell at teams where I'm like, hey, how come you're logging so much?

这太贵了。

This is expensive.

而且，你知道吗，你们真的在用这些日志吗？

And, you know, are you using these logs?

我可以控制我输入的数据量。

I can control the amount of data I put in.

存储和S3之类的情况也一样。

Same with storage and s three or something canonically.

我输入一GB，或者输入两GB。

I put in a gigabyte, put in two gigabytes.

没问题。

Fine.

对吧？

Right?

问题是，这些相对而言是可以转移且由我作为客户控制的。

The problem is those are relatively transferable and controllable by me as a customer.

人们举出的许多关于结果或按使用量计费的例子，都不是由我作为客户所能控制或可互换的。

A lot of the examples people give of either outcome or consumption based pricing are not in control by me as a customer and not exchangeable.

所以，AI令牌世界、AI积分世界对客户来说真的非常困难，因为他们会想：我不太明白你们给我的这个赌场筹码到底是什么东西。

So the AI token world, the AI credit world is really, really difficult for customers because they're like, I don't really understand what this casino token you've given casino chip you've given me is.

对吧？

Right?

我可以从AWS拿一GB数据，然后放到Azure上，我知道它们会怎么收费，因为一GB是相对恒定的。

I can take a gigabyte from AWS and go put it in Azure, and I know how much they're gonna charge me because the gigabyte is kind of constant.

当我拥有这些AI积分时，我不清楚你们的积分和你们的积分是不是一样的。

When I have these AI credits, I'm like, I I don't know if your credits are the same as yours or the same as yours.

顺便说一下，你们不断添加新功能，而这些功能会消耗我的积分，因为我的用户在使用它们。

And by the way, you keep adding features which chew up my credits because my users use them.

然后我就想：等等。

And I'm like, wait.

我不清楚他们是怎么使用这些积分的。

I don't I don't know what they're doing with those credits.

这不是公司自己决定去使用的。

Like, it's not the company choosing to use them.

而是供应商添加了一些功能，让软件变得更好，但这些功能似乎悄无声息地就发生了。

It's the vendor adding, like, features that make the software better that seem to just happen.

对吧？

Right?

我可以通过添加一大堆新功能，比如‘我为你做了这些很棒的摘要’，一夜之间就把客户的积分使用量增加十倍。

I can 10 x my customer's credit usage overnight by adding a whole bunch of stuff like, hey, I built these great summaries for you.

而他们却说：等等，这可不是我干的。

And they're like, wait, I didn't do that.

所以我认为，当跟客户聊基于成果的使用模式时，他们想要的其实是席位。

So I think the outcome based usage, when you talk to customers, they want seats.

大概是因为今天他们能理解这个概念。

Probably because today they understand it.

其次，他们被这种基于消耗的模式坑了，账单突然飙升，他们不禁问：我该怎么控制这个？

And secondly, they're being burned by a lot of this consumption base that the bill just goes up massively, and they're like, wait, how do I control this?

对。

Right.

需要做一些调整。

Is that make some adjustment.

是的。

Yeah.

这在许多类别中都会普遍存在。

It will be certainly present in a lot of categories.

你知道，我们在Atlassian的业务中有一些领域，你可以认为是基于消耗的定价，或者干脆就是纯粹的按消耗计费。

You know, we have a bunch of areas of our business at Atlassian that are, you would argue, consumption based pricing or literally just consumption based pricing.

但我们尽量坚持那些客户使用量翻倍的领域。

But we try to stick to areas where customers do twice as much stuff.

他们获得的价值也翻倍。

They get twice as much value.

他们支付两倍的钱，但这一切都在他们的掌控之中。

They pay twice as much money, and it's in their control.

许多其他事情他们却无法掌控。

A lot of these other things aren't in their control.

基于成果定价的最后一个例子是，这些成果也是动态变化的。

And the last example of outcome based pricing is those outcomes are also dynamic.

所以，以客户服务为例，我帮你节省了开支——你过去每月在客户服务上花费20美元。

So the problem with, say, customer service where I've saved you you know, you used to spend $20 on customer service.

使用我们的工具后，你只需花费10美元。

With our tool, you'll only spend 10.

这在第一年是个很棒的销售说辞。

That's a great sales pitch in year one.

到了第二年，客户会说：但我现在只花了10美元。

In year two, the customer goes, but I only spend 10.

现在我想只花5美元。

Now I wanna spend 5.

否则，你就没有带来任何价值。

Otherwise, you didn't deliver any value.

而供应商会说：‘如果你不用我，你还是会花20块。’

And the vendor goes, well, if you took me out, you'd be spending 20.

但你会说：等等，我可没花20块。

And it's like, wait, But I don't spend 20.

我只花了10块。

I spend 10.

所以，从成果角度来看，我每年帮你省钱的能力很难衡量。

So, like, my ability to save you money each year is difficult from an outcome basis.

对吧？

Right?

我是在消除任务。

I'm eliminating tasks.

我觉得从销售的角度来看，我也创办过两家支付公司，我之所以了解Workday，是因为我曾经很羡慕他们，我会跟我的销售团队谈论Workday，因为他们从外部清楚地知道他们从通用电气那里赚了多少钱。

I think also, like, from a sales perspective, I I've started two payment companies, and it was really I used to this is why I know Workday is I envied them, and I would talk to my sales team about Workday because they know from the outside in how much money they make from GE.

他们说，好吧。

They're like, okay.

GE 使用 PeopleSoft。

GE uses PeopleSoft.

他们有 33 万名员工。

They have 330,000 employees.

也许我们每月收他们 4 美元，但更可能是每位员工每月 5 美元。

Maybe we charge them $4 a month, but probably $5 per employee per month.

这就是你从这个客户那里赚的钱。

This is how much money you make from that account.

而且，如果你销售的是软件产品，或者任何其他东西，组建销售团队要容易得多，顺便说一句。

And it's so it's so much easier to scale a sales team if you're selling a software product or anything, by the way.

如果，你知道，这家公司愿意付我们 300 万美元，而不是像我们当初创业时，签下了 100 家 Flowers 公司。

If, you know, that company will pay us $3,000,000 versus like, you know, we when when we were starting a firm, we signed up one-eight 100 Flowers.

我们完全不知道能从他们那里赚多少钱。

We have no idea how much we're gonna make from them.

结果发现，真正让这个生意成功的是什么？

And it turned out like, you know what really made the business work?

Casper 矩阵公司。

Casper the mattress company.

什么？

It's like, what?

就是这个愚蠢的地图？

Like, this stupid map?

但你就是不知道。

Like but it's like, you just don't know.

你以为你搞到了沃尔玛这样的大客户，但一开始其实没怎么成功。

And you think like you get like a big like, we got Walmart, didn't really work out that well in the beginning.

我们搞到了 Casper 矩阵公司。

We get Casper the mattress company.

天哪。

Oh my god.

不可思议。

Incredible.

Workday 在两个方向上都具有可预测性。

Workday has the it's predictability in both directions.

对吧？

Right?

对于花钱的一方——也就是客户来说，它是可预测的；同时，对于管理团队来说，也知道应该把时间花在争取通用电气这样的大客户上，而不是去签一家十人公司，因为通用电气比十人公司大得多。

It's predictably for the spender of the money, which is the customer, but it's also the predictability for the management team knowing that you should spend your time trying to sign up GE and not sign up a 10 person company because GE is bigger than a 10 person company.

而在互联网领域，情况却很疯狂，比如 Stripe 可能从一家十人公司赚的钱比从通用电气还多。

Whereas it's crazy in Internet land where it's like Stripe might make more money from a 10 person company than GE.

我想，你在那里或许能达到更高水平的可预测性。

And I guess you you can get to, like, higher levels of predictability there.

但当你采用基于成果的定价或基于使用量的定价时——我的意思是，基于使用量的定价本身并不坏，但如果你从外部无法预知一个客户能带来多少收入，那么扩大销售和营销团队就会变得异常困难，作为创业者，你根本无从下手。

But, like, when you have outcome based pricing or consumption based pricing or something I mean, consumption based pricing is not bad per se, but if you don't know from the outside in how much you can make from an account, it just becomes exponentially harder to scale a sales and marketing team because you just, as an entrepreneur.

有一件事我想回头再聊聊，就是你们在这个时代是如何适应的。

One thing I want to go back to of dealt with how you guys are adapting in this era.

你能再多分享一下，这对你们来说最显著的表现是什么吗？以及它如何促使你们改变业务？

Can you share more about the biggest ways in which that's manifested for you and you know, how it's made you, you know, change your business?

听好了。

Look.

我认为我们看待这个问题的方式是，我们关注的是……

I think the the way that we think about it is we look.

我们销售的是解决人类协作问题的协作工具，对吧？这些工具应用于许多不同领域，比如服务团队、广泛的业务团队、人力资源、财务、软件团队等，各种类型的团队都在使用我们不同的应用组合。

We sell collaboration tools that solve human collaboration problems, right, in lots of different areas, service teams, broad business teams, HR finance, software teams, like lots of different types of teams by different sets of apps from us, collections and and sets of apps.

从根本上说，这些都是涉及大量文本的协作问题，这对我们非常有利。

Fundamentally, they're all collaboration problems that involve a lot of text, so this is really good for us.

那些人究竟在做什么，这可能是关键所在。

What are those people doing is probably the important part.

对吧？

Right?

科技界常常倾向于认为：我们要重新发明一切，这才是未来的方向。

The technology world often runs to, we're gonna reinvent everything, and that's the way of the future.

在中长期的时间尺度上，这通常是成立的。

And that generally is true in the medium to long arc of time.

我们的挑战始终在于，有许多客户仍在使用当下的工作方式、工作流程和应用程序，但他们非常聪明。

Our challenge is always we have a lot of customers that work in today's manner, today's workflows, in today's set of apps, and they're not they're very smart.

他们希望迈向未来，但同时也必须推动大量人员的转变。

They wanna get to tomorrow, but they also have to move a lot of people.

因此，当我们构建AI功能时——我可以举这些方面的任何例子——我们需要理解这项技术是什么，以及它如何帮助我们。

So when we're building AI features, and I can give examples of any of these, we need to understand what that technology is, how it can help us.

这是我们首先考虑问题的方式。

That's how we think about it, firstly.

其次，为了迎接未来，我们需要构建哪些基础平台组件？

Secondly, what fundamental platform componentry do we need to build for whatever that future will be?

因为这些技术的发展速度正在急剧加快。

Because this stuff's accelerating so fast.

对吧？

Right?

因此，我们由此发展出了AI网关、团队协作图谱以及企业合规与控制功能。

So that's how we got to our AI gateway and the Teamwork Graph and the enterprise compliance and controls.

你必须将这些与你在特定应用中为客户构建的功能区分开来。

You have to separate that out from the features you're building for customers in a given app.

然后，我需要为客户提供他们实际会使用的功能。

Then I have to build features for customers that they use.

对吧？

Right?

那么，这些功能该放在哪里？

So where do you put those features?

这些功能具体是什么？

What are those features?

其中很多都集成在现有工作流程中，帮助客户更快、更好、更高质量、更高效地完成这些现有流程。

A whole bunch of them are in existing workflows to help the customer do that existing workflow faster, better, higher quality, more efficiently.

从吸引人的角度来看，这些功能可能并不炫酷，无法像在X平台上那类三十秒的动画GIF那样吸引眼球，但对客户而言却极其重要，因为他们今天就能使用。

Those tend to be very unexciting from a magic point of view in terms of what sells a, you know, a thirty second animated GIF on on X, but they're incredibly exciting from the customer because they can use them today.

他们的现有工作方式只是变得更好了。

Like their existing way of working just got better.

他们说，这太棒了。

They're like, this is amazing.

他们对这些东西赞不绝口。

Like they rave about that stuff.

在AI领域，我觉得这其实挺简单的。

And in the AI world where I'm like, but that's pretty simple.

但事实上，它今天能给他们带来巨大的帮助。

And it's like, but it actually helps them today in a massive way.

不过，我内部跟人说的时候，可以举个服务领域的例子，这还不够，因为你还需要将他们的现有工作流程与新应用结合，或者审视新的工作流程，并能应对这些情况。

I tell people internally though, and you can give an example in service, that's not enough because you also need to use their existing workflows with new apps or look at new workflows and be able to handle that as well.

对吧？

Right?

所以我们必须做所有这些事情。

So we have to do all of these things.

所以如果你看一下，比如Jira就是一个典型的例子，在我们的服务集合中，尤其是在我们的HR和IT服务管理产品中，总结工单是我们现在能做得远比以往更好的事情。

So if you look at, you know, Jira's a canonical example, you know, in the service collection, in our in our HR and IT service management products, summarizing a ticket is something we can do way better than we ever could.

因为企业中已经存在大量现有的工作流程。

Because there's a lot of existing workflows we have in an enterprise.

可能有四到五个人，甚至六个人内部协作处理同一个工单来解决问题。

Maybe a four or five, six people work a ticket internally to try to resolve the problem.

当第四个人接手时，会看到一大堆附件文件。

The fourth person that shows up, there are a whole lot of attached files.

还有很多对话内容。

There's a lot of conversation.

还有各种各样的其他信息。

There's a lot of different things going on.

他们通常需要花三十分钟来通读所有内容，理解发生了什么，然后才能运用自己的专业知识来解决这个问题。

They would normally have taken thirty minutes to, like, read it all and understand what's going on so then they can bring their expertise to bear on the problem.

仅仅通过把内容直接丢进大模型来生成摘要，这根本不是简单的事。

Literally just summarizing that, and it's not a simple stick it into, you know, an LLM and get back to summary.

你必须非常小心地处理上下文，这对他们来说非常强大，但他们的工作流程却一点都没变。

You have to be very careful about the context, is so powerful for them, but they haven't changed their workflow one iota.

仍然是亚历克斯说：嘿，埃里克，你能来帮我看一下这个工单吗？

It's still Alex saying, hey, Erik, can you come help me with this ticket?

埃里克出现了。

Erik shows up.

埃里克不得不重新加载他大脑里所有相关信息。

Erik has to bootload his brain with all the things.

所以这是一种现有的工作流程，我们可以用大模型来极大地改善它，他们非常喜欢。

So that's like an existing workflow where we can use LLMs just to make that customer way better, and they love it.

对吧？

Right?

他们对这类功能赞不绝口，但这些功能其实都很简单。

They rave about all these types of features, but they're very simple.

它们通常并不是自主代理的。

They're usually not agentic.

那我们可以这么说，不错。

Then we can say, cool.

但这个服务流程，我们需要在各个节点上引入代理。

But that service workflow, we need to put agents in at various spots.

对吧？

Right?

大多数人都是在梳理流程时，发现你知道什么吗？

And most people are taking a workflow and finding you know what?

这个步骤经常让我们卡住。

This step trips us up a lot.

这个步骤耗费了我们大量时间。

This costs us a lot of time.

我们能不能让这个步骤更快一些？

Can we make this step faster?

这绝对是需要我们自己提供代理框架的事情。

And that's absolutely something that we have to provide agent frameworks ourselves.

我们有一个非常棒的代理框架，它能利用所有的时间图和你拥有的所有上下文信息。

We have a pretty great agent framework that uses all the timograph and all the context you have.

它非常简单。

It's pretty simple.

它也很便宜。

It's pretty it's very affordable.

或者你可以使用自己的代理框架。

Or you bring your own agent framework.

对吧？

Right?

我认为，大多数企业内部都会运行三到五个大规模的代理平台。

Most businesses, I think, will have three to five large scale agent platforms running internally.

他们会说，嘿。

And they say, hey.

我用Agent Force来做这个，或者我用Gemini来做这个。

I use Agent Force for this, or I use Gemini for this.

很好。

Great.

把这个代理引入进来，我们会把它插入到这个工作流中，让它正常运行。

Bring that agent, and we'll pop it into the workflow here and we'll make that work.

对吧？

Right?

我们必须能够做到这一点。

We have to be able to do that.

但你仍然处于现有的工作流体系中。

But you're still all in the existing workflow world.

你只是在完成旧的任务，同时再执行一个新而高效的任务，但仍在现有工作流内。

You're just doing the old task and then doing kind of a new and efficient task, but in the existing workflow.

然后你会听到有人问：如果服务工单根本不存在呢？

Then you get people like, what if the service ticket didn't exist at all?

对吧？

Right?

因此，你正在重新构想整个类别的软件以适应新的工作流程，而我们必须帮助客户跨越这一鸿沟，因为他们通常没有单一的服务团队。

So you're reimagining whole categories of software to new workflows, and we have to help our customers make it across that gap because they don't generally have one service team.

他们有数百个。

They have hundreds.

对吧？

Right?

如果他们有数百个不同的服务台在运行，他们可能会说，这20个将采用这种新方式运作。

And if they have hundreds of different service desks running, they might say these 20 are gonna work in this new way.

但其他的仍需全部管理。

These but they have to manage them all.

所以，我想我们正试图将团队协作图中的数据与此结合，同时从客户驱动的角度出发。

So I guess we're trying to bring data in the teamwork graph together with this and also from a customer driven lens.

我认为这一点常常被忽略了。

I think that often gets left out here.

对吧？

Right?

我们正试图将他们带入五年后的未来。

We're trying to take them five years into the future.

我们的职责实际上是让他们同时进入一年、两年和五年后的未来，这正是我们正在努力做的。

It's our job to actually get them one year and two years and five years into the future simultaneously, which we we're trying to do.

最后我想说的是，我们在设计上投入了大量资源。

And the last thing I'd say is we're investing a lot in design.

我认为这种NII对话常常被忽略，因为在这套系统如何运作方面，还有很多基础性设计要做。

And I think that always an NII conversation gets left out because there's a lot of foundational design to do in how this works.

对吧？

Right?

我们已经看到了这方面的初步成果。

We're seeing the first elements of this.

但如果回顾移动时代，第一批应用只是把桌面或网页上的东西照搬进手机里。

But if I look at the mobile era, the first set of apps were kind of just canonically taking desktop or web things and sticking them in a phone.

然后我们才逐步发展出新的交互和体验模式。

And then we evolved new patterns of interaction and experience.

对吧？

Right?

甚至不是视觉方面。

Not even the visuals.

我们该如何使用这些东西？

How do we use these things?

推送通知是做什么用的？

What push notifications for?

它们在一开始并不存在。

They didn't exist at the start.

对吧？

Right?

下拉刷新就是一个非常明显简单的例子。

Drag to refresh is like a very obvious simple example.

这是一种相当典型的設計模式，通常在这里很成功，并被推广到其他地方。

That's a pretty canonical design pattern that generally it's successful here and it gets moved across.

但整个来说，我该怎么把手机和电脑结合起来用呢？

But the whole, like, how do I use my mobile and my desktop together?

我该怎么在两者之间来回切换？

How do I move back and forth?

我们面临这么多设计挑战，其实都是为了帮助人们理解眼前的东西。

We have so many design challenges to solve that actually help people to understand what's there.

我们普通的客户、普通的用户，并不想去理解这些原理。

The average customer we have, the average user they have, they don't want to understand.

如果AI对他们来说不存在，那也没关系。

If if AI doesn't exist for them, that's fine.

但他们想要的是AI带来的结果。

But they want the outcomes of it.

对吧？

Right?

他们不需要了解所有技术细节。

They don't need to know all of the technical detail.

我们的职责是隐藏这些复杂性，只提供他们想要的答案，或让任务更高效、更有效。

It's our job to hide them and just give them the answer they're looking for or make a task more effective or efficient.

我觉得在科技界，我们有时过于痴迷于模型的质量。

I feel like in the technology world, sometimes we get so obsessed by, like, model quality.

现在说模型的能力远远超过了它们实际带来的价值，这几乎都成老生常谈了，因为那些未被充分利用的能力实在太庞大了。

You know, it's it's almost trite now to say the models are far ahead of the the actual value they're delivering now that the the underutilized capabilities are so big.

这个等式的一部分实际上是设计和体验。

A part of that equation is actually design and experience.

对吧？

Right?

我该怎么实现这一点？

How do I get this?

给人们一个功能无限的聊天框，他们却只会说：给我讲个爸爸笑话。

Give people a chat box that can do unlimited power, and they're like, tell me a dad joke.

这就像拥有无限的力量，但要帮助他们有效利用这种力量却非常困难，而这正是我们面临的巨大挑战所在——如何将智能代理及其全部能力融入工作流程和协作循环中，让人与代理协同工作。

Like, it's like unlimited power, but it does it's very hard to help them utilize that power, which is where a huge amount of our challenge goes in terms of bringing agents and all the the power of them into workflows and collaborative loops and and having humans and agents work together.

我非常喜欢关于拟物化设计的观点，你知道，最初你有的只是一张张纸。

I I I love the skeuomorphic point on both you know, it's first, it's like you had pieces of paper.

早期的网页就只是网页。

The early web was just like a web page.

这就是为什么它叫网页。

That's why it's called a web page.

就像8.5乘以11英寸那样。

It's like eight and a half by 11.

对吧？

Right?

到了移动时代，哦，他们得到的只是一个小小的网页。

And then mobile, oh, well, they get a tiny web page.

但结果发现，如果你不局限于拟物化设计，而是从第一性原理出发，充分利用设备的功能，就能做出各种其他创新。

And then it turns out if you don't just go into the skeuomorphic world, but you just think from first principles and take advantage of the power of the device, you do all sorts of other things.

比如，下拉刷新。

It's like, know, the the the scroll to refresh.

对吧？

Right?

比如，下拉刷新。

Like, the the pull pull down to refresh.

这是移动端带来的一种新概念。

That was a new concept that came from mobile.

对吧？

Right?

前几天我一直在想这件事。

So I was thinking about this the other day.

你试过 Nano Banana 2 吗？

I'm have you tried Nano Banana two?

试过了。

Yes.

真的很棒。

It's really good.

对吧？

Right?

所以，我一个同事刚刚说，嘿。

So, one of my colleagues just said, hey.

对于一位访问日本的美国游客，制作一张关于该做什么和不该做什么的信息图。

For an American tourist visiting Japan, make an infographic about what to do and not to do.

这就像一箭双雕，太棒了。

And it's like, it one shot something that's amazing.

你怎么编辑这个输出结果？

How do you edit that output?

对吧？

Right?

这就让人觉得，嗯，你可以编辑文字。

And that's where it's like, you know, it feels very it's like, well, you could edit the text.

你可以编辑图片。

You could edit the graphics.

你可以直接一次性生成一个新的东西。

You could just one shot something new.

或者，你知道的，我想问你的是，现在的情况如何？

Or, you know, what what is the state of I guess this is my question for you.

我的意思是，你认为当前的技术水平是什么样的，或者应该是怎样的？因为你提到过关于如何设计来编辑AI输出的问题。

It's like, what do you think the state of the art is or should be, and how have you been thinking about this just because you mentioned design for editing the output of the AI output?

对吧？

Right?

因为它们就像是经典的那样，哦，我用个图形界面，点这里改那里。

Because they're they're like the they're the classic it's like, oh, I'll use a GUI and click here and change that.

但感觉这种做法非常拟物化。

But it feels like that's very skeuomorphic.

对于这个问题，我会把视角拉远两层来回答，因为这是个很好的问题。

I would I would I would zoom out two levels from that to answer that question because it's a great question.

首先，在这些领域里，建立客户信任真的很难。

First is customer trust is really hard in these areas.

对吧？

Right?

当你去和用户交谈时，你会坐下来，和他们做研究，问他们问题，问五个为什么。

When you go talk to users, you sit down, you do research with them, you sit, you ask them questions, you ask the five whys.

他们对人工智能感到害怕，不是因为它的强大，而是因为它会主动做事。

They're very scared of AI, not because of its power, because it does stuff.

他们就会说，嘿。

And they're like, hey.

我怎么知道那是对的？

How do I know that was right?

它做了什么？

What did it do?

对吧？

Right?

就像是说，别担心。

It's like the idea that, oh, don't worry.

我的AI机器人已经发了15封邮件，帮你管理收件箱。

My AI bot's gone and sent 15 emails and manage your inbox.

你的收件箱空了。

Your inbox is empty.

然后你就会想，好吧。

And you're like, okay.

它到底做了什么？我还不信任它。

Did it I don't trust it yet.

我对AI这么快就自动做事这件事，存在信任问题。

Like, so I have a trust question on generally AI doing things really quickly.

要建立信任，它必须回来告诉你：我打算这么做。

To gain trust, it has to come back to you and say, here's what I'm about to do.

你确定要我这么做吗？但又不能烦人到非要你确认才行，直接去办不就好了。

Are you sure you want me to do this without being annoying that, like, just effing go and do it.

所以，这完全是一个设计问题。

So, like, that that's a whole design question.

你多久会遇到一次这种情况？你如何建立对这些工具的信任？

How often is it how do you build trust with any of these tools?

第二个问题是，它是否有足够的数据？

The second is, does it have enough data?

对吧？

Right?

很多人工智能都是一次性处理事情。

So much of AI is one shotting things.

在X上坐一会儿，你会看到成千上万个类似这样的例子：嘿。

Sit on X, you'll see a thousand like, hey.

这是那个神奇的提示版本，像哈利波特的咒语一样，能帮你运行一个一个人的十亿美元企业。

This is the magical prompt incarnation Harry Potter spell that does this, like, runs you a one person billion dollar business.

只要把这个提示输入进去，然后粘贴就行。

Just put this prompt in and paste it.

这其实有点荒谬，因为现实是，你在数据方面也需要大量的迭代。

And, like, that's, like, kind of ridiculous because the reality is you also have a lot of iteration on the data side.

对吧？

Right?

一次性完成确实很有用，但你常常需要回去修改输出和输入。

One shotting things is really useful, but you often need to go back and edit the output and the input.

对吧？

Right?

我不太擅长。

I'm not very good.

我用过一个例子很久了，就是你让AI写一篇我的作业论文。

I've used this example for a while where you say, hey, go write me an essay for my homework.

它会生成一篇论文。

It'll spit out an essay.

然后你就会说，等等，不对，不对，这是历史课。

And you're like, wait, no, no, it's a history class.

它会说，哦，好的。

They're like, oh, okay.

那我们来写一篇作文吧。

Well, let's draw an essay.

实际上你是在不断修改输入，这有点像聊天式的迭代。

And like, you're actually changing the input and somewhat this is chat like iterations.

但如果你曾经试过用聊天迭代来做图像编辑，就会觉得特别挫败，比如：哦，不。

But if you've ever tried to do that image editing with chat iterations, it's super frustrating where like, oh, no.

你改了我不让你改的东西，然后又回来了。

You changed the thing I didn't want you to change, and you come back.

你就会说：啊。

You're like, ah.

所以这涉及到输入设计和体验的问题。

So like there's an input design and experience problem.

其中一部分是如何掌握恰当的上下文信息？

Part of that is how do I have the right amount of context?

另外还有输出和迭代方面的问题。

And then there's an output and iteration problems.

我们的团队协作图谱可以访问你组织中的大部分知识。

Our Teamwork Graph can access largely all of your organizational knowledge.

它的准确度高得惊人。

It's insanely accurate.

它拥有强大的搜索功能。

It's got great search.

它的相关性非常出色。

It's got amazing relevance.

你会觉得，太棒了。

And you're like, sweet.

我拥有了完整的组织记忆。

I have full organizational memory.

现在，团队协作图谱知道我曾在2002年写过代码。

Now the teamwork graph knows that I used to write code in 2002.

它之所以知道，是因为它拥有惊人的记忆能力。

And it knows that because it has this insane memory.

我觉得这其实没什么用。

And I'm like, it's actually not useful.

除了某一件事之外，别用它来回答我提出的任何问题。

Don't use that to answer any query I give you other than one thing.

迈克以前是个开发者。

Mike used to be a developer.

可能是个不怎么样的开发者。

Maybe a bad one.

对吧？

Right?

现在他哪儿都找不到工作了。

And it wouldn't get hired nowadays anywhere.

但也许这有助于向我解释一些事情，比如，你有计算机科学学位。

But maybe that helps in explaining something to me in a way that, oh, you have a computer science degree.

我可以这样向你解释，但我并不想了解所有这些信息。

I can help explain it to you in this way, but I don't wanna know all that information.

为什么这是一个输入上的挑战？

Why is that an input challenge?

你现在看到的这些框框，一会儿说要搜索网页，一会儿又说别搜索网页。

You kinda see all these boxes at the moment where it's like, search the web, don't search the web.

搜索你的组织，别搜索你的组织，你这是在让用户做一堆他们根本搞不懂的选择。

Search my organization, don't search my organ like, you're asking the user to make all these choices they don't quite understand.

这不符合设计流程，对吧？设计流程应该是：嘿，这个问题，我猜你是希望我做这个和那个。

That's not in a design flow, right, where it says, hey, this question, I suspect you want me to do this and that.

对吗？

Is that correct?

在深度研究中你偶尔能看到一点这样的情况，但有点让人沮丧。

You see that a little bit in deep research, but it's a bit frustrating.

这就导致了一种情况：天啊，我有17个不同的代理在各自跑腿干活，就像有一堆实习生在忙活。

And it leads to this whole, like, man, I've got 17 different agents running off and doing stuff, and I'm like, it's like the the problem of having a lot of interns.

拥有50个实习生的问题是，你确实能完成很多工作。

We're like, the problem with having 50 interns is you get a lot of work done.

拥有50个实习生的问题在于，他们每分钟会问你50个问题，所以你唯一在做的事就是回答实习生的问题。

The problem with having 50 interns is they ask you 50 questions a minute, and so like, all you're doing is answering questions for interns.

因此，你确实需要解决体验层面的输入问题。

So there's an input problem of experience that you really need to solve.

然后你就会遇到迭代问题，这在公司里要困难得多。

Then you get to the iteration problem, which in a corporation is much more difficult.

对吧？

Right?

因为我们举过一个很好的例子，比如头脑风暴，通常不是一个人独自进行的。

Because, we gave this great example of, you know, brainstorming where it's not usually one person brainstorming.

所以在我们的白板和Confluence里，你可以引入代理并说：嘿。

So in our in our whiteboard and Confluence, you can bring in agents and say, hey.

我想就这个主题进行头脑风暴。

I wanna brainstorm about this topic.

它们非常擅长通过团队协作图从你的组织知识中获取所有信息，然后回来提供一个出色的头脑风暴方案。

They are really good at going off and getting all the information from your organizational knowledge through the teamwork graph and coming back with a really good brainstorm.

我们不断改进，把卡片画得更好，放在正确的位置，等等。

And we get better and better drawing it and putting the cards in the right places and everything else.

如果你只是随意地拿走这些，说‘去吧’，就会失去人类的输入和信任。

If you just take that randomly and say, go, you lose human input and trust.

所以实际上，通常会发生的是，我们拥有一大堆数据。

So actually, usually what happens then is we've got a bunch of data.

我们要开个会。

We're gonna have a meeting.

我们要把大家聚在一起。

We're gonna get people together.

我们要问：大家觉得怎么样？

We're gonna go and say, well, what do we all think?

加入我们的直觉、大脑的判断，这些哪些是有用的，哪些是没用的？

Add our intuition, the the the the brain matter, which of these are useful or not useful?

然后这些信息必须回到另一个智能循环中，说：太好了。

And then that information has to go back into some other agentic loop to say, cool.

现在我们已经投票了，尽管这个投票实际上是人类过程的输出，接下来你就要采取行动。

Now we've kind of voted, although the voting is like the output of a of a human process, then you're gonna go and do something.

然后我们要决定该做什么，以及我们是否做对了，还有所有这些事情。

Then we're gonna work out what to do, and did we do it correctly and and all these things.

正如你所说，输出的质量是非确定性的，但我认为这需要一个人类代理循环。

It's, as you said, it's very nondeterministic in the quality of output, but it requires, I think, this human agent loop.

对吧？

Right?

而正确设计这个流程是一个设计问题。

And and getting that right is a design problem.

循环太多，会让人感到沮丧。

Too many loops, it's frustrating.

循环太少，就会失去信任，事情就只是自动发生了。

Not enough loops, you lose trust, and it just it just happens.

因此我们看到，实际上已经在Jira中上线了很多代理功能。

And so we see that we just shipped, you know, agents in Jira in a lot of ways.

所以你可以把工作分配给一个代理，然后它就会去执行任务。

So you can, like, assign work to an agent, and it goes off and does stuff.

当我们用真实用户测试时，他们会问：它到底在做什么？

And when we test it with people, they're like, well, what's it doing?

你是想让我们走一千步吗？

Like, do you wanna give us a thousand steps?

他们还说：你干嘛跟我讲这么多废话？

And they're like, why are you telling me all this crap?

我说：等等。

I'm like, wait.

因为你明明知道我在做什么。

Because you said you know what I was doing.

所以，把它们融入工作流程和业务流程中，会面临很多设计挑战，比如我不知道。

And so there are lots of design challenges with just bringing them into workflows And back to the business processes, like, the I don't know.

安全部门在很多地方都参与其中。

The security team is involved in a of places.

会计团队、财务团队，有很多地方都需要涉及。

The accounting team, the finance team, there's lots of places.

比如在销售中，通常需要财务部门批准交易，或者由财务人员处理。

Like, even in sales, finance usually has sign off on a deal or someone in finance does.

你如何做到这一点，并让这个工作流程更顺畅，仅仅通过分配给代理来实现？

How do you do that and make that workflow better where you're just assigning to agents?

你必须非常谨慎地对待用户体验。

You need to be very careful about the experience.

它如何返回？

How does it come back?

它什么时候返回？

When does it come back?

这让人感到沮丧吗？

Is it frustrating?

它是否以一种新的方式返回？

Does it come back in a new way?

我能了解一下它现在正在做什么吗？

Can I interrogate what it's doing right now?

比如，我们的内部或第三方代理在Jira中运行时，如果他们在处理任务，你可以在他们工作时与他们聊天，问他们在做什么？

Like, our agent, first or third party agents were running in Jira, if they're off doing a task, you can chat to them while they're doing the task and say, what are you doing?

这有助于我们在短期内建立信任，我们相信是这样的。

Which helps you build trust in the short term, we believe.

但从长远来看，如果你信任这个代理执行这项任务——毕竟它过去20次都做对了。

But in the long term, if you trust it, this particular agent doing this task, man, it's got it right the last 20 times.

概率是对的。

The odds are right.

它做得很好。

It's it's good.

我就直接忽略它了。

I'm just gonna ignore it.

我认为，这些都是根本性的设计和体验问题。

These are all, I would argue, a fundamental foundational design and experience problem.