在人工智能领域，谁将成为房间里的成年人？

欢迎收听《深度视角对话》播客。

Welcome to the Deep View Conversations podcast.

希巴尼，很高兴见到你。

Shibani, so good to see you.

再次见到你真好。

Nice to again.

对于不了解你的人，介绍一下你自己以及你在Salesforce的职位吧。

For those who don't know, talk about who you are and what your role at Salesforce is.

嗯，我在Salesforce的职位大概一个月前刚变了。

Well, my role at Salesforce just changed about a month ago.

我记得我们上次聊天的时候，我负责企业IT战略，主要就是跟全球的C级高管，特别是CIO们见面，给他们提供AI转型方面的建议。

So I think when we last chatted, I was leading enterprise IT strategy, and what that meant was I was meeting with c suite leaders, particularly CIOs around the world, advising on AI transformation.

结果呢，我给出的建议效果不错，公司就跟我说，你给别人的那些建议，也可以在公司内部用起来。

Well, turns out I gave some really good advice, and the company said, that advice that you've been giving to other folks, you can do it for us within the company.

所以现在我在负责Koa的数据和AI战略。

So now I'm leading data and AI strategy for Koa.

所以Koa涵盖了我们所有的CFO和COO职能。

So Koa is all of our CFO and COO functions.

所以现在我所做的就是应用所有这些建议，我们正在推动构建代理型企业。

And so now what I'm doing is all that advice, I'm applying, and we are driving we're building the agentic enterprise.

我和我的团队正在为Salesforce构建代理型企业。

I'm me and my team are building the agentic enterprise for Salesforce.

哇。

Wow.

好的。

Okay.

你知道吗，我本来想说，自从我们二月份聊过之后，发生了一些事情。

You know, I was gonna say I I was gonna start this by saying, you know, since we talked in February, a couple things have happened.

确实发生了一些事情。

Couple things.

但这不在我的预料之内，所以真是个惊喜。

But that wasn't even in my purview, so Surprise.

那挺好的。

That's good.

我确实在领英上看到过。

I do remember seeing it on LinkedIn.

我记得看到你在领英上有些变动。

I remember seeing, you know, something change on LinkedIn.

但那可是件大事。

But, but that's that's huge.

我是说，你想想，从二月份到现在，你们都已经推出了自己的AI模型。

I mean, think about but you think about even since February, like, you all have launched your own AI model.

是的。

Yes.

对。

Yep.

你们确实做到了。

You have, yeah.

这些智能体已经发展得飞快了。

The agents have had, like, the have gone parabolic.

就像囚犯越狱了一样。

The inmates have escaped the prison.

别这么说嘛。

Come on now.

AI智能体就像越狱了一样。

The AI inmates have escaped the prison.

就像，呃，

Like, a

发生了很多事，对吧？

lot has happened, right?

我们还建立了一个很好的框架，确保安全、保障和治理。

We've also built a beautiful harness ensures safety, security, governance.

我们已经从概率性、不可预测的智能体，发展到了概率性加确定性的智能体。

We've evolved from probabilistic and unpredictable agents to probabilistic plus deterministic.

听我说，Jason，我们得把情况说得全面一点。

Think, listen, Jason, we've got to paint a balanced picture.

好的，我理解。

Okay, I appreciate it.

这周很多人跟我说过这个，挺好的。

A lot of people have told me this week, which is good.

你的团队在这件事上口径很一致。

Your team is very on message with this.

因为我用了“囚犯”这个比喻，这周不是第一次用了，每次得到的回应都一样，就是说“我们控制得住”。

Because I've used the inmates Not the first time I've used that joke this week, and I've gotten the same thing, like, we've got it under control.

这就是我们的职责所在。

That's what we're here for.

不过说句公道话，我的标题是Salesforce可能是AI领域里最靠谱的那个。

I will say that my, to be fair, my headline was that Salesforce might be the adult in the room on AI.

所以

So

杰森，我完全同意。

Jason, I'm down for that.

确实如此。

It tracks.

确实如此。

It tracks.

谢谢。

Thank you.

那是什么？

What is it?

对你的角色来说，这看起来是什么样的？

What does it look like for your role?

因为上次我们聊的时候，你一直在和首席信息官、很多IT负责人以及各种企业交流。

Because last time we talked, you'd been talking to CIOs, a lot of IT leaders, a lot of different enterprises.

他们都在努力理解AI，而且你知道，他们的进展和困惑程度各不相同，而你一直致力于帮助他们。

They're trying to get their arms around AI, and they were, you know, with varied rates of success and confusion and all And, of you know, you are dedicated to helping them.

我觉得你就像，我聊过的人，我都记不清了。

I think what you're like, I've talked to, I can't even remember.

那真是太多了。

It was a lot.

587。

587.

有意思，我本来想说超过500，但我觉得那不可能。

It's funny, I was gonna say over 500, and I thought that can't be right.

那只是

That's just

太多了。

too many.

确实。

Was sure.

是587，就像一对一是一回事，但这是一对多。

It five eighty seven, like it would be one to one was one thing, but it was like one to many.

那会是一大堆

It would be a lot of

会议，

conferences,

而且还是在第三方会议上，不只是我们自己的。

and it was also at third party conferences, not just our own.

那还是，那是一个

That's still, that's a

很多。确实很多。

lot It is a lot.

那是很多人。

That's a lot of folks.

好的，所以你跟超过500位，差不多600位CIO聊过AI。

Okay, so you talked to over 500, almost 600 CIOs about AI.

你的角色是什么时候变的，从那以后又是什么样的？

When did your role change, and what's it looked like since then?

所以，大概一个月前我转到了这个新岗位，还挺有意思的。

So it's about, about a month ago when I shifted into this new role and it was funny.

我一直在提醒自己，要记住我一路以来给那些CIO的建议，也就是你刚才说的那种感觉——确实有点让人应接不暇。

I'm trying to remind myself of the advice that I've been giving to CIOs along the way that we're saying exactly what you were saying is that like, this is a little bit overwhelming.

每个人都在推销自己的AI方案。

I've got everyone that's got an AI solution.

我还有一堆技术债务要处理。

I've got tech debt.

还有数据需要整合统一。

I've got data that needs to be harmonized.

我知道自己必须得做点什么。

Like, I know I've got to do something about that.

我也知道得搞定那些智能体。

I know I have to figure out agents.

一下子涌过来太多东西了。

It was a lot coming at

对。

them.

当然。

Sure.

我记得当时我会怎么说，现在我觉得我甚至进一步调整了，就像我在Salesforce学到的东西一样。

And I remember back then what I would say now, and I think now I've tweaked it even further in terms of like, as I'm, what I'm learning about what we are doing at Salesforce.

当时，我会说，你知道，深呼吸一下。

Back then, I would have said, you know, like, take a breath.

我想我们讨论过智能体成熟度模型。

We, I think we talked about the agentic maturity model.

是的。

Yes.

我当时说，Matt，你可能有一百个，一百七十个关于智能体的用例和想法。

I had said like, Matt, you probably have a hundred, hundred and seventy use cases, ideas of agents that you have thought about.

拿这些去对照智能体成熟度模型，那个模型是我们构建的，或者说是一个框架，一边是聊天机器人，然后是副驾驶或生成式AI，再然后就是智能体AI的整个范围。

Take that and map it against the agentic maturity model, and the agentic maturity model was a model that we had built or it was a framework that we had built that showed chatbots on one side, then you got copilots or, like, regenerative AI, and then you got the spectrum of agentic AI.

我会说，把你所有能想到的用例都对照那个成熟度模型来梳理，然后从第一级开始。

And I would say just map all the use cases you can think of against that maturity model and then start at level one.

因为我觉得，那种不知所措的感觉，就是人们会想，我怎么才能构建一个能和其他智能体对话、能穿越虚拟主机的理想智能体呢？

Because I think that that was a bit of the overwhelm is people are like, how do I build this utopian agent that's gonna speak to other agents that's gonna traverse the virtual host?

我当时就说，不不不。

And I was like, no, no, no.

从第一级开始。

Start at level one.

所以我现在又回到那些基本原则，来问我们自己，我们已经在哪些地方有了第一级的用例，应该去理解我们做了什么，我们是怎么取得那些成功的？

And so I'm I'm kinda coming back to those basic principles for us to say, where do we already have level one use cases where we should be understanding what we've done, how we've had those successes?

所以这就是我回到的地方，就是说，让我们看看那些第一级、第二级的用例，然后倒推着来想，好吧，要让那个智能体完美运行，我们需要哪些最关键的数据元素？

And so that's what I'm coming back to, to say, let's look at those level one, level two use cases and then work backwards to say, okay, what's the minimum critical data elements that we need to get perfect for that agent?

我们需要连接哪些技术，我用简单的话说，才能让那个智能体真正跑起来？

What's the tech that we need to connect, and I'm using simple terms, for that agent to become active?

在这个过程中，你也在整合和清理那些能带来影响的数据元素。

And in the process, you're harmonizing and cleansing the data elements that are gonna drive impact.

在这个过程中，你会开始审视过去的技术架构，可能需要更新和改造。

In the process, you're starting to look at technical architecture from the past that you might need to, like, refresh and revamp.

所以就是

So it's

不是那种要花两年的项目，你现在是分阶段进行，每一步都能看到价值和收益。

not like a two year You're now doing it in pieces and you're starting to see value and benefit at every increment.

对吧？

So Right?

你的项目不再是什么大工程了。

Your projects are not much of a thing anymore.

对。

No.

这就像是从瀑布式项目转向敏捷交付。

It's like it's it is it's from going from project like waterfall to agile delivery.

就是冲刺。

It's sprints.

就是把团队聚在一起。

It's bringing teams together.

技术、业务用户，把这些人聚在一起，共同完成这件事。

Technology, the the business users, bringing all these folks together to be able to do this collectively.

每个人都在桌上互相学习，这些跨职能小组。

Everyone is learning from each other at the table, these cross functional pods.

那作为内部要去各个团队帮助他们的人，是什么感觉？

So what is it like being the person internally who has to go teams to help them?

因为我猜你的工作有一部分是帮助赋能，可能在某些地方，还要稍微加快进度。

Because I suspect some of your job is helping empower, maybe in some places, like, hastening a little.

那具体是什么样的情况？

So What's that look like?

比如，我最早发现的一件事，因为我觉得我们得先真正定义什么是AI。

Like, one of the earliest things that I've discovered because I think we've gotta actually define AI.

一方面，大家都在说，我们必须让AI普及化。

Everyone's like, on one hand, you're hearing everyone say, we've got to democratize AI.

让组织里的每个人都能用上AI。

AI for everyone in an organization.

但另一方面，这也是我一直在和客户讨论的话题——因为我才上任一个月——就是关于AI的成本和token费用的问题，对吧？

And then on the other hand, and this is still a conversation I'm having with, because remember, I've been in my job for a month, I'm having with customers is about like the cost of AI and token, token costs, right?

所以，这就像是光谱的两端。

Like, so those are two ends of the spectrum.

所以我内心也在反思，在我们组织内部，我该怎么处理这个问题？

So I'm reflecting on that internally for me to say, like within our organization going, how am I, do I want to handle this?

我停下来想了想，我觉得我有机会像定义成熟度模型那样来定义AI，那我该怎么定义AI呢？

And I'm pausing to say, I think I have an opportunity to define AI Similar to the maturity model, how do I define AI?

成熟度模型主要是让技术人员理解消费、技术栈、数据清洗这些内容。

And the maturity model is really for technologists to understand consumption, tech stacks, data cleansing, all that good stuff.

但我在我们组织里把AI定义为一级AI或模式一AI，

But I'm defining AI in our organization as like level one AI or mode one AI,

好的。

Okay.

就是我们给每个人提供了Gemini，还有Slack和Slackbot。

Is where we've given Gemini, we've given Slack and Slackbot.

Slackbot给所有人用。

Slackbot to everyone.

我们正式把AI普及了，因为每个人都能用上。

We have officially democratized AI because everyone has access to it.

整个公司吗？

The whole organization?

所有人。

Everyone.

因为我们在用Slack，所以每个人的Slack里都内置了Slackbot。

Everyone has Slackbot built into Slack because we use Slack.

所以现在每个人都能用上AI。

So everyone is able to leverage AI today.

而且它是个人辅助工具。

And it's an individual assistive.

这是辅助型AI。

It's assistive AI.

好的。

Okay.

所以在这个阶段，每个人都能用上它。

So in that one, I have everyone's got it.

而且我们并不是期待那种实实在在的投资回报。

And it's not like we are expecting ROI in a tangible way.

对。

Okay.

它是在帮助和支持同事。

It is assisting and aiding a colleague.

第二阶段就是，我们现在有了像Claude CoWork和Claude Code这样的工具。

Level two is where we're saying, okay, now we've got things like Claude CoWork and Claude Code.

那是更高级的建模。

That is more advanced modeling.

所以不一定整个组织里每个人都需要这个，但我正在跨部门查看，问他们需要这个权限吗？

So not necessarily like everyone in the entire organization needs that, but I'm going across the corporate functions and saying, needs this access?

他们为什么需要？

Why do they need it?

但我没有限制它。

But I'm not limiting it.

在那个层面，我们只是在盘点正在构建的技能、代理和功能。

That level, all we're doing is an inventory of the skills and the agents and the capabilities that are being built.

我们在看使用情况，哪里能找到一些很棒的重度用户，想展示他们？

We're looking at the consumption, where are we finding some amazing power users that we wanna showcase?

然后你进入AI的第三阶段，我们会说，好的，想想我们组织内部，比如财务部门的L3流程，我们可以自动化并代理化。

Then you get to mode three of AI where we're going, okay, think that we've got some processes within our organization within say finance at L3 that we could automate and agentify.

我们会说，让财务部门的流程专家来，和我们一起进行两周的冲刺。

We're saying we're bringing the process experts from finance, come sit with us for a two week sprint.

好的。

Okay.

我会带着专家，来自Agent Force运营团队，也就是以前的Regrelo。

I will bring experts with, at agent force operations, what was formerly known as Regrelo.

我们会帮你制定蓝图，帮你实现自动化和流程优化。

We'll help you create the blueprints, we'll help you automate and streamline this.

现在我们可以真正开始量化价值和收益了。

Now we can actually start to quantify value and benefits.

然后就是AI的最后一个阶段，有些组织在后退一步思考，如果我们能重新架构整个公司会怎样？

Then you got the last mode of AI, which is where organizations are, some are stepping back and saying, what if we could re architect our entire company?

比如说，从线索到现金，也就是从你有一个线索或潜在客户，一直到你实际收到现金，我们可以想象一个世界，有代理能够跨多个部门、多个职能，执行某些操作。

What if, what if like lead to cash, the point at which you have a lead or a prospect right down to where you get cash in hand, that we could imagine a world where we've got, agents that could traverse multiple departments, multiple functions, and perform certain actions.

这是一个比较慢的建设过程。

That is a slower build.

治理方式也不同。

There's a different governance.

如果出了问题，影响范围也会不同。

There's a different blast radius should something go bump in the night.

所以我开始从四个不同的层面来看待这个问题，并且合理地理解投资回报率是什么。

So I'm starting to look at this in four different planes and appropriately understanding what is the ROI.

这些智能体应该由谁来构建？

Who should be building these agents?

这些智能体的治理机制是什么？

What's the governance of these agents?

所以我觉得现在这开始变得更有意义了，我们是否在正确的领域进行建设，是否在所有领域都做了一点。

And so I think that this is now starting to make a bit more sense in, we building in the right spaces, and are we doing a little bit in all of

这些类别？

the categories?

所以类别很多。

So that's a lot of categories.

确实

It

是。

is.

那挺好的。

That's good.

确实。

It is.

现在是本周赞助商Deal的广告时间。

Now it's time for a word from this week's sponsor, Deal.

全球招聘和发工资可能是个大麻烦。

Global hiring and payroll can be a serious headache.

我们懂。

We get it.

大多数团队要同时用五六个，甚至十个不同的人力资源工具，就为了在海外招人和发工资。

Most teams juggle five, six, sometimes 10 different HR tools just to hire and pay people abroad.

Deal用一个AI原生的平台，把人力资源、IT和工资管理这些工具都整合在一起了。

Deal replaces those tools with one AI native platform for HR, IT and payroll.

像Shopify、Revolut和Eleven Labs这样的公司，已经在用Deal来拓展全球业务了。

Companies like Shopify, Revolut and eleven Labs already scale internationally with Deal.

还有Air Wallachs，这个领先的AI金融平台，管理并支付着16个国家的1100多名员工。

And Air Wallachs, the leading AI finance platform, manages and pays over 1,100 employees across 16 countries.

要使用Deal建立你的全球团队并开始扩展业务，今天就访问deel.com/deepview。

To build your global team with Deal and start expanding your business today, visit deel.com/deepview.

网址是deel.com/deepview。

That's deel.com/deepview.

我们感谢Deal对《深度视角》的支持。

We thank Deal for their support of The Deep View.

现在回到节目中来。

And now back to the show.

上次我们

When we

聊天时，你说了一句话让我印象深刻。

talked last time, you said this thing that stuck with me.

你说，AI是最小也是最大的词。

You're like, AI is the smallest, biggest word.

你怎么记得这么清楚？

How do you remember that?

太厉害了。

That's amazing.

我，我，因为你让人印象深刻。

I I because you're memorable.

哦。

Oh.

但这话说得特别有智慧，对吧？因为大家对它的定义太不一样了。

But it it was it was such a it was such a a wise thing to say, right, because people define it so differently.

因为对它的期待和恐惧，就像光谱的两端。

Because the expectations of it and the fear of it are like these two ends of the spectrum.

没错。

Exactly.

然后争论就被这两极定义了，而在这之间其实有很多可能的结果，一边是AI会毁灭人类，另一边是AI能治愈所有疾病、消除贫困、带来无限的富足。

And then the argument gets defined by these two very there's a lot of outcomes between AI is gonna kill humanity and AI is gonna cure all disease and bring an end to poverty and have unlimited abundance.

就是说，在这两者之间其实有很多可能的结果。

Like, there's a large set of outcomes, potential outcomes in between that.

所以我很期待问你这个问题，就是Salesforce是怎么看待这个问题的，因为我觉得现在挑战的一部分就是帮助人们意识到，事情不是非黑即白的，而且坏结果和好结果也不是互斥的，对吧？

And so I've been looking forward to asking you about this, like how Salesforce thinks about this because I think that's part of the challenge right now is helping people realize like it's not and also these two things are not mutually exclusive, the bad outcomes and the good outcomes, right?

AI是工具，是人用的。

AI is a tool that people use.

Salesforce是工具公司。

Salesforce is a tool company.

你们在和客户合作的时候，是怎么帮他们建立对AI的愿景的？

You know, how do you all, when you're working with your customers, help them have a vision for what AI is?

因为我觉得这肯定是其中的一部分。

Because I'm sure that's part of it.

你看，现在有各种困惑，还有恐惧。

Like, there's all the confusion, there's the fear, you know.

我记得我们之前聊的时候，你提到过一点，就是你觉得你的工作的一部分就是去帮助人们理解这件事，把它当成一个思维模型来掌握。

Think and one of the things I got from you when we talked before is, that was what you saw as part of your job was to go and help people, you know, get their arms around this, like, as a mental model.

比如，我怎么理解这个，以及它对我们意味着什么？

Like, how do I understand this and what it means for us?

是的。

Yes.

你们内部也是这么做的吗？

Is that what you're doing internally too?

那看起来怎么样？

How's that look?

你说AI是最小又最大的词，这太有意思了，因为我也正是这么说的。

It's so funny that you say AI is the smallest, biggest word because that's exactly what I'm saying.

即使我在跟你聊这些模式，我为什么会有这种模式的概念，正是因为你说的那个原因。

Even as I talk to you about these modes, why am I even thinking in this notion of modes is for exactly what you're describing.

我们是不是期望，当我们给某人提供Slackbot或Gemini时，我们就能算出投资回报率，并且我们需要能够治理和控制它？

Are we expecting that when we give someone, Slackbot or Gemini that they are gonna we are gonna be able to calculate the ROI and we need to be able to govern and control this?

然后你会说，不，其实不是这样。

Like, you're going, no, actually, no.

在某个阶段，比如模式一，别指望有那种回报。

At a certain point, at at, at, at, mode one, don't expect that return.

那是你可以坚持的东西。

That is something that you can hold on to.

而且也别过度管控，因为日常辅助型AI能做的事情是有限的。

And also don't over govern it because there's a limited amount of things that folks can do with like everyday assistive AI.

所以我觉得要合理设定预期，就像你说的，不管是ROI，还是护栏、安全、信任、治理，这些都要考虑。比如，如果我在野外部署一个模式四的智能体，我肯定想知道有合适的护栏，有职责分离，如果出了问题，能明确谁负责那个智能体。

And so I think like right sizing expectations, like you said, whether it's on ROI or, the guardrails or the security, the trust, the governance that you put in place in terms of, if I put on a Mode four agent in the wild, I certainly want to know that there are appropriate guardrails and that there is separation of duties and that if it goes bump in the night, there's clarity of who owns that agent.

但这不是对待面向大众的辅助型AI的方式。

But that's not how you treat assistive AI that is meant to be for the masses.

好的。

Okay.

所以，这正是我们在做的，这就是为什么我有一个思维模型。

So it is, that is exactly what we're doing is that's why I've, a mental model.

我为我们创建了一个思维模型，确保我们在看待公司里AI的整个光谱时，采取合适的步骤。

I've created a mental model for ourselves to make sure that we are taking the appropriate steps, as we look at the spectrum and the full spectrum of what AI is in our company.

太好了。

Excellent.

好的。

All right.

你们推出了自己的模型。

Y'all launched your own model.

很有意思。

Very interesting.

非常酷。

Very cool.

Koa 是 Salesforce 基于 NVIDIA 的 Nemotron 构建的模型。

Koa is the model that Salesforce did, built on NVIDIA's Nemotron.

这

This

真的很有意思，因为这是过去几个月里我和企业界进行的最有意思的对话之一。有些企业突然灵光一现，觉得他们需要拥有自己的 AI。

is so interesting because it's one of the most interesting conversations I've had over the past couple months with enterprises is enterprises there's a few of them that have, like, had this light bulb moment where they're like, we need to own our own AI.

就是说，这是我们自己的智能。

Like, this is our own intelligence.

这是最有价值的，这是我们的镇店之宝。

This the most valuable this is our crown jewel.

这是我们业务中最有价值的部分，或者将来会是，而且我们不能就这么把它外包给别人，尤其是那些我们的竞争对手也在用的，而且我们也不一定知道谁在控制数据之类的。

This is the most valuable part of our business or will be, and, like, we can't just sort of outsource that to somebody where the to somebody that also our competitors are using and that, you know, we don't necessarily know who controls the data and all of that.

所以，我一直在聊这个话题。

So so I've been having this conversation.

我就觉得，哦，对。

I'm like, oh, yeah.

那真的很有意思。

That's really interesting.

比如，汤森路透就做了自己的AI模型。

Like, Thomson Reuters made their own AI model.

CrowdStrike，几周前，他们也宣布了自己的AI模型，也是基于Nemotron的，跟Salesforce的一样。

CrowdStrike, a couple weeks ago, they announced their own AI model also based on Nemotron, just like Salesforce's.

然后，当然，我到了这里，听到的第一件事就是我们正在推出自己的AI模型。

And then, of course, I get here, and one of the first things I hear, we're we're launching our own AI model.

所以我非常想听听你对这件事的看法。对。

And so I I would love your thoughts on on that Yeah.

因为这太及时了。

Because, it is so timely.

而且，如果你六个月前问我，比如我们二月份谈话的时候，七个月前，如果我问你或者你问我，我们自己搞一个AI模型怎么样？

And also, if you would have asked me six months ago, like in our conversation in February, seven months ago, if I would have asked you or you would have asked me, you know, like, what about us doing our own AI model?

那简直是不可能的，根本就不会在我们的考虑范围之内。

That would have been like we would have never it just would not have even been in the realm of possibility.

但是很多事情都变了，我觉得这显示了事情加速发展的趋势。

But a lot has changed, and I think it shows the acceleration of things.

而且你构建的Nemotron Super，这个模型所基于的，当时甚至还没发布，这说明了现在的时间线有多疯狂。

And and Nemotron Super that you're built it wasn't even released yet that the this model is built on, which sells how bananas, like, the timeline is right now.

所以，你觉得这对Salesforce来说意味着什么？能够拥有一个专门为你们、你们的客户和你们的业务定制的模型，你觉得怎么样？

So your thoughts on what does that enable for Salesforce, you know, being able to have its own model that is specialized for you and your customers and your business?

我是这么看的，这其实跟我们所有产品的做法是一样的，就是那种无头化的概念。

The way I'm looking at it is it's kind of following a pattern of what we're doing with all of our products, like this whole notion of headless.

我们是在提供选择。

We're creating options.

对。

K.

但这就是一个自己选路线的游戏。

But it's a it's a choose your own adventure.

就像我们之前说的，选你自己的湖，选你自己的仓库，选你自己的应用，选你自己的界面。

No different than we've said choose your own lake, choose your own warehouse, choose your own applications, choose your surface.

这不过是又多了一个，就是选你自己的模型。

This is just one more that we're saying choose your own model.

而且我们也，你可能也看到了，我们推出了一系列的智能体，对吧？

We also have, like you also probably saw, the spectrum of agents that we've introduced, right?

我们有那种你可以自己搭建的智能体，也有现成的，比如Qualified或者Fin这样的。

We've got agents that you can you've got we've got the platforms where you can build agents as you wish, or you can get agents like, like Qualified or Fin.

这些就是专门的智能体。

Like, these are specialized agents.

所以现在我们有了专门的模型，让客户有更多选择。

So now we have specialized models that will allow customers to just have optionality.

这才是我们真正想做的事情。

That's really what we're trying to do.

大家好。

Hey, everybody.

感谢收听本期节目。

Thanks for listening to this episode.

简单说两句，然后我们继续回到对话。

Quick note, and then we'll get you back to the conversation.

我们每周都很喜欢为大家带来这些内容。

We love bringing you this content every week.

我们一直在想办法，怎么给大家带来最大的价值，帮助大家理解AI如何改变商业、改变世界。

We're always trying to figure out how we can deliver you the most value to help you understand how AI is transforming business and transforming the world.

如果你喜欢这个节目，有个简单的方法可以回馈我们，同时也能帮助更多人了解这个节目。

If you're enjoying the show, there's an easy way for you to give a little value back and help others learn about the show as well.

如果你在Apple Podcasts或Spotify上，给我们打个分，留个评论吧。

If you're on Apple Podcasts or Spotify, drop us a rating and leave us a review.

如果你在YouTube上，点个赞、订阅一下，或者留个评论。

And if you're on YouTube, hit like, subscribe, or leave us a comment.

就这样。

That's it.

只需要一分钟，但对我们帮助很大。

It only takes a minute, but it's a huge help.

所以提前感谢你的支持。

So thanks in advance for pitching in.

现在，我们回到节目正题。

And now back to the show.

其中一个

One of

想到这一点，我觉得很有意思，也很有力量，是的。

the interesting things that's come up thinking about that that well, that I think is really powerful Yeah.

显然，AI的能力在最近几个月里进步巨大，对吧。

Is, obviously the capabilities of AI have advanced enormously, right, in recent months.

你看到越来越多的人在实际工作中使用它，这很棒。

You see more people using it, like actually using it in business, which is great.

这个周末我和一位老师、一位非营利组织的负责人、一位教授，还有一位艺术家开了个会。

I was in a meeting this weekend with a teacher, a nonprofit, head of a nonprofit, a professor, and an artist.

我根本没提AI的事，一个字都没提。

And I didn't bring up I did not even bring up anything about AI.

其中一个人提到，哦，嘿，你不是搞AI的吗？

One of them mentions like, oh, hey, you do something with AI, right?

对。

Yep.

然后所有人都开始跟我讲他们用AI做的各种事情。

And then it was like all of them wanted to tell me all the things that they've been doing with AI.

我邻居也是这样。

That happened with my neighbors too.

我就觉得，这太疯狂了。

I'm like, this crazy.

这已经是主流了。

This is It is mainstream.

它已经在这里了。

It is here.

这是每个人都想聊的话题，对吧，

It's the topic, right, that everybody wants to talk about, which

我觉得，我们经历了一个变化，一开始人们会说，我不知道这是什么，

is It's like, I think that there were folks at first, we've gone through this change where at first people were like, I don't know what this

然后变成，我觉得用这个像是在作弊。

To then going, I feel like I'm cheating by, by using it.

现在变成，我简直不敢相信，这太不可思议了。

To now going, I can't like, this is incredible.

我离不开这个。

I can't live without this.

就像，我离不开Slackbot。

Like, I can't live without Slackbot.

Slackbot让我保持清醒。

Slackbot keeps my head on.

就像，Slackbot，我接下来要去哪儿？

Like, Slackbot, where am I going next?

嘿，我接下来要和谁说话？

Hey, who am I speaking to next?

就像，Slackbot就是一切。

Like, Slackbot is everything.

Slackbot为这次会议给你准备了什么？

What did Slackbot prep you for this meeting?

我不需要任何准备，Jason。

I didn't need any prep, Jason.

我当时就跟Slackbot说，我很期待和Jason见面。

I was like, Slackbot, I'm excited about meeting with Jason.

我就是这么说的。

That's what I said.

你太棒了。

You're so good.

你太棒了。

You're so good.

你说过我会让人印象深刻。

You said that I was memorable.

得保持让人印象深刻才行。

Have to stay memorable.

没错。

That's right.

好的。

All right.

非常好。

Very good.

非常好。

Very good.

这能看到你自己的模型，也许，这好像是我人生中第二次看到这种情况。

This able to see your own model, maybe, isn't showing for like the second time in my life.

所以拥有自己的模型很酷的一点是，当你把很多东西放出去的时候，会有很多问题。而且，现在我们到了这个地步，它非常有用，但现在也有一些大的问题。

So the cool thing about having your own model, is when you put a lot of the stuff out there, there's big Also, now we're getting to the point, it's so useful, but now there's sort of big questions.

比如，好吧。

Like, okay.

那我们的数据呢？他们是在用我们的数据训练吗？

What about our date like, what data is it are they training on our data?

它会不会基于这些数据产生幻觉，你知道吗？

Are the is it gonna hallucinate, you know, based on this?

因为如果它真的那样，这可是非常关键的业务数据。

Because if it does, like, this is really important business critical stuff.

比如说，如果它胡编乱造，那我就麻烦了。

Like, I'm gonna be in trouble if it's hallucinating.

我们都遇到过这种情况，你问它一个问题，下次再问，它给的答案就不一样了。

And we've all had the thing, like, where you ask it something, and then the next time you ask it, it tells you a different one.

你会说，不对，我希望你像上次那样做，结果它又变了。

You're like, no, I wanted you to do it just like you did the last time, and it's different.

所以，通过这个领域专属模型，用三十年的业务流程来训练，也就是所谓的商业智慧、经验教训或最佳实践，这样就消除了很多可能让你陷入麻烦的问题，对吧？

So by having this domain specific model, train on thirty years of business processes, what you might call business wisdom or learnings or best practices, then it's eliminating a lot of this stuff that might get you in trouble, Right?

世界知识，对于商业解决方案来说，你并不需要那些。

World knowledge, you don't need that for sort of business solutions.

但同时，它也注入了很多关于流程的专业知识和最佳实践，这感觉就像是改变游戏规则的关键。

But it's also then injecting a lot of this, like, expertise and best practices around processes and and that that feels like the sort of the the game changer.

你是这么想的吗？

Is that how you think about it?

因为第一，你获得了更高的准确性。

Because you get, one, better accuracy.

它会用更少的令牌，因为它不需要用那种大型的，你知道，企业级或者前沿模型。

It's gonna use fewer tokens because it doesn't have to use a big, you know, enterprise or or a frontier model.

而且呢，它更准确，也不太容易产生幻觉。

And then also, like, it's more accurate and less likely to hallucinate.

你是这么想的吗？

Like, is that how you think about it?

我理解的方式是，大概是一年前还是两年前吧？

The way I would look at it is, was it about, like, a year a year ago, or two years ago?

很多人当时都说，模型才是关键。

A lot of folks came out with, like, the model is the thing.

模型

The model

是最重要的。

is the most important thing.

而我们作为一家机构，我们当时说的是，光有大语言模型是不够的。

And we, us as an organization, what we said was an LLM is not enough.

光有模型是不够的。

The model alone is not enough.

对。

Okay.

我觉得即使对我们来说，如果认为只要自己造一个模型就能解决问题，那是不对的。实际上，关键在于那个“框架”。

I think that even for us, thinking that just us creating our own model is going be the silver lining and silver bullet, no, it's actually like, it's the harness.

你可能听过Rowan提到过这个“框架”。

I think you might've heard Rowan talk about the harness.

就是模型加上框架，两者结合才行。

It is the model in conjunction with the harness.

所以对我来说，框架就像你刚才说的那种概率行为。

So what's the harness to me is like, what you just talked about was like probabilistic behavior.

模型本身天生就带有概率性。

Models by nature on their own have a probabilistic nature to them.

要让它变得确定，就需要把模型、工作流、应用，还有包含元数据和语义的数据层结合起来。这样你才能得到一个可控的系统，在需要的时候可以推理和概率性运作，但在必须确定的时候又能保持确定性。

To bring that determinism is when you get the model plus the workflows, plus the applications, plus the data layer that has the metadata, the semantics, that's when you start to get something that's contained, that can reason and be probabilistic when you need it to be, but deterministic when you have no choice.

我不知道，光靠模型是不够的，得是模型加上配套系统才行。

I don't know that's the model alone, it is the model plus the harness.

现在我们就是在不断微调、再微调，确保客户对我们部署的东西有信心。

Now we're just saying, we're fine tuning and fine tuning and fine tuning to make sure that our customers have confidence in what we are deploying.

太棒了。

Very cool.

接下来呢？

What's next?

你希望AI能为Salesforce和你们的客户做些什么？

What do you wish that AI could do for Salesforce and for your customers?

比如，有什么是你迫不及待想让AI去做的？

Like, what's the thing that you can't wait until AI goes?

Jason，你问得太远了。

Jason, you're going too far.

接下来我希望AI能帮我做的是，规划我的下一个假期。

What's next is what I do I want AI to do for me is plan my next vacation.