Practical AI
Less about Models; More about Architecture
In this Practical AI episode, the guest, Rackspace's chief AI officer, traces the evolution of industrial AI from predictive maintenance and defect detection to LLMs and physical AI. He argues that enterprise adoption now depends less on individual models and more on architecture, including governance, assurance, data sovereignty, cost and tokenomics, and the ability to swap models as technology changes. The conversation also compares robotics and AI adoption across Japan, China, and North America.
本期 Practical AI 的嘉宾是 Rackspace 的首席 AI 官（transcript 未给出姓名），他回顾了自己从数学、Hitachi 到 Rackspace 的职业经历，并讨论工业 AI 从预测性维护、缺陷检测、deep learning 到 LLM 和 physical AI 的演进。他认为，企业落地 AI 的关键已不只是选择模型，而是架构：包括治理与 assurance、data sovereignty、tokenomics/cost，以及能否随技术快速变化替换模型。他还谈到 physical AI 和机器人可能在商业化上率先落地，并比较了 Japan、China 与 North America 在机器人和 AI 采用上的不同节奏。整体上，节目强调在模型快速变化以及外部监管、供应链等因素影响下，企业需要从架构层面做出可治理、可扩展的 AI 决策。
00:002756
00:01
Welcome to the Practical AI Podcast, where we break down the real-world applications of artificial intelligence
欢迎收听 Practical AI Podcast，我们会拆解 AI 在现实世界里的实际应用，
00:08
and how it's shaping the way we live, work, and create.
以及它如何改变我们生活、工作和创造的方式。
00:12
Our goal is to help make AI technology, practical, productive, and accessible to everyone.
我们的目标是让 AI 技术变得实用、高效，并且对每个人都触手可及。
00:18
Whether you're a developer, business leader, or just curious about the tech behind the buzz,
无论你是开发者、企业管理者，还是单纯对这股热潮背后的技术感到好奇，
00:22
you're in the right place.
你都来对地方了。
00:24
Be sure to connect with us on LinkedIn, X, or Blue Sky
记得在 LinkedIn、X 或 Blue Sky 上找到我们，
00:27
to stay up-to-date with episode drops, behind-the-scenes content, and AI insights.
第一时间获取节目更新、幕后花絮和 AI 洞察。
00:32
You can learn more at practicalai.fm. Now on to the show.
更多内容请访问 practicalai.fm。好，让我们正式开始。
00:41
Welcome to another episode of the Practical AI Podcast.
欢迎收听Practical AI Podcast的又一期节目。
00:44
This is Daniel Whiteknack, I am CEO at Prediction Guard
我是Daniel Whiteknack，Prediction Guard的CEO。
00:48
and I'm joined as always by my co-host Chris Benson, who is a principal AI and autonomy research engineer at Bucket Martin.
和往常一样，和我一起的是我的联合主持人Chris Benson，他是Bucket Martin的首席AI与自主性研究工程师。
00:56
How you doing, Chris?
Chris，你怎么样？
00:57
Hey, doing great today, Daniel. Looking forward to a conversation here.
嘿，Daniel，我今天状态很好。很期待今天的对话。
01:01
Yeah, excited to chat about all sorts of things, both in terms of background and current work with Chetan Gupta today
是的，很高兴能和Chetan Gupta聊各种各样的事情，包括他的背景和现在的工作。
01:10
who is chief AI officer at Rackspace.
他是Rackspace的首席AI官。
01:13
Welcome, Chetan, how are you doing?
欢迎你，Chetan，你最近怎么样？
01:15
Hey, I'm doing good. Thanks, Daniel and Chris for having me.
嘿，我挺好的。谢谢Daniel和Chris邀请我。
01:17
Quite excited about the conversation.
对于这次对话非常期待。
01:21
Yeah, yeah, well, like I say, we had bonded even yesterday when we were chatting about our backgrounds
对对，是的，就像我说的，我们昨天聊天聊到彼此背景时就很有共鸣。
01:29
and physics and mathematics.
还有物理和数学。
01:31
I know you started out in mathematics and spent a bunch of time at Hitachi.
我知道你最开始是学数学的，而且在Hitachi待了挺长时间。
01:35
Do you want to give us just an idea of a little bit of your background and what you've been involved with over the years?
你能稍微给我们介绍一下你的背景，以及这些年你都做过些什么吗？
01:42
Sure, sure.
当然，当然。
01:44
My background is actually quite diverse.
其实我的背景相当多元。
01:46
It's sort of a typical of people in my role.
这在我们这种角色的人里挺典型的。
01:50
I started off with a PhD in mathematics.
我一开始拿的是数学 PhD。
01:52
Then I joined Hillet Packet Labs with the research scientist working on data mining, machine learning.
然后我加入了 Hillet Packet Labs，以研究科学家的身份做数据挖掘和机器学习。
01:57
Then I joined Hitachi as a principal researcher for November correctly.
然后我加入了 Hitachi，担任首席研究员，负责 November correctly。
02:01
And then I sort of, you know, through the management ladder was when I left Hitachi in 2026,
然后我，你知道，就是一路沿着管理阶梯往上走，到 2026 年离开 Hitachi 的时候，我已经在领导 Hitachi 全球所有的 AI research 了。
02:09
I was leading all of AI research at Hitachi globally.
而且那是一个非常强的团队。
02:12
And this is a very strong team.
这就是我做过的事。
02:15
And that's what I did.
这就是我所做的。
02:16
And we focused a lot.
我们当时非常专注。
02:17
We started focusing on industrial AI as we defined it at that time.
我们开始专注于当时我们所定义的工业AI。
02:24
And as the sort of industry matured, we should have started looking at a broader spectrum of things.
随着行业逐渐成熟，我们本应开始着眼于更广泛的领域。
02:31
And that's what I was doing.
而那也是我正在做的事情。
02:32
And then I joined RAC space.
然后我加入了RAC space。
02:34
And I then have four months.
之后我有了四个月的时间。
02:36
So it's been sort of a very, very fascinating journey.
所以这真是一段非常非常精彩的旅程。
02:40
And in some sense, if I try to this podcast, I was thinking about my career.
某种程度上，如果让我回顾这期播客，我其实也在思考自己的职业生涯。
02:46
Typically, you don't think about these things.
通常来说，你不会去想这些事情。
02:48
And I realized that if you, in some sense, I have followed the trajectory of the whole community at a broad level.
然后我意识到，从某种意义上讲，我一直在宏观层面跟随着整个 community 的发展轨迹。
02:57
If I remember, if you remember guys, you were all doing machine learning, making small models, solving specific problems.
如果我没记错的话，各位应该记得，你们当时都在做 machine learning，做小模型，解决具体问题。
03:04
And there was a community in Bay Area that was focused primarily on cell accommodation systems.
当时 Bay Area 有一个 community，主要关注的是 cell accommodation systems。
03:09
And as Chris, you would know, in companies like Lockheed Martin or Hitachi, we focused industrial problems.
Chris，你也知道，在 Lockheed Martin、Hitachi 这样的公司里，我们关注的是工业问题。
03:15
And so that's what I was doing for the first half of my career.
所以我职业生涯前半段做的就是这些。
03:18
And then more than the half.
后来这件事其实超过了职业生涯的一半。
03:20
And then, obviously, deep learning became important beside looking at vision models, language models that have been blossomed in total, as language models.
再后来，显然 deep learning 变得非常重要，除了 vision models，language models 也在整体上全面开花。
03:29
And in some sense, a lot of the traditional machine learning problems today are much more solvable with automated tools, with sort of vibe coding and so on and so forth.
从某种意义上说，现在很多传统的 machine learning 问题，用自动化工具、vibe coding 之类的，已经好解决多了。
03:38
And the new challenge now is, yes, you can do all this in machine learning and AI.
而现在的新挑战是，没错，这些你在 machine learning 和 AI 里都能做到。
03:42
But how do you, how do you make it accessible to more people?
但你怎么，怎么让它对更多人来说更可用？
03:46
How do it make it actionable?
怎么让它变得更可操作？
03:47
How do you make it much more sort of safe?
怎么让它变得更安全一些？
03:50
And that's where sort of RAC space comes in.
而这就是所谓的 RAC space 发挥作用的地方。
03:53
So in some sense, you should look at the trajectory of problems.
所以某种程度上，你得去看问题演变的轨迹。
03:56
That's how I have traveled in some sense, like looking for trouble.
我就是这样一路走过来的，某种程度上，像是去找麻烦。
03:59
So I love that.
我很喜欢这一点。
04:02
So I've got, I want to go back for a second.
我想稍微往回聊一下。
04:05
I'm going to drag you back as you kind of went through your timeline for a second.
我要把你拉回去一点，就像你刚才回顾自己的经历那样。
04:08
Because as you were talking about kind of coming up through the ranks, the managerial ranks at Hitachi.
因为之前你谈到在 Hitachi 一步一步往上走，进入管理层。
04:16
And you look at, you know, with kind of the AI world exploding, you know, in terms of volume and importance and budgets and all that stuff.
你也知道，现在 AI 世界正在大爆发——无论是规模、重要性，还是预算，全都一样。
04:25
And so many times organizations will go to, you know, kind of an outside expert and bring them in to fill a particular key position and stuff.
很多时候，组织会去请一个外部专家，专门来填补某个关键职位，诸如此类。
04:34
And yet you kind of came through the ranks at Hitachi.
但你是从 Hitachi 内部一步一步升上来的。
04:38
And I'm wondering if you have any thoughts about like, what, you know, because there are other people out there that are watching this and listening to this right now that are in their careers.
我想知道你怎么看这件事——因为现在正在看这个、听这个的人里，有很多都处于自己的职业发展中。
04:48
And they are aspiring to move up through the ranks themselves.
而他们自己也渴望能一步步往上爬。
04:53
And what were some of the things that you brought to bear that made you able to kind of move up through that and take on the leadership role at Hitachi before you were able to come over to RAC space?
那在你来到 RAC space 之前，是靠了哪些东西让你能够在 Hitachi 一路升上去、承担起领导角色的？
05:04
Just from a kind of a growth learning standpoint, if you, if you would mind sharing.
就是从一个成长学习的角度来说，如果你愿意分享的话。
05:09
Because that's a hybrid and I was more prepared to answer the question.
因为那是一个 hybrid 的问题，而我本来更准备去回答的是另一个问题。
05:13
Sorry.
抱歉。
05:15
But I think that's actually a very, very interesting question.
但我觉得那其实是一个非常、非常有意思的问题。
05:18
I think one thing that I think helped in my career at Hitachi was we decided to take a bit on industrial AI.
我觉得在我 Hitachi 的职业生涯里，有一点帮助很大，就是我们决定在 industrial AI 上押一点注。
05:26
This was 2016, 2017 and not very many people were taking talking about it.
那是 2016、2017 年，当时没有多少人在谈论这个。
05:31
So we started a small research lab in North America.
于是我们在北美成立了一个小型研究实验室。
05:34
And I said, given where Hitachi is, it's a giant industrial concrete.
然后我说，考虑到日立的定位，它就是一个庞大的工业集团。
05:38
It makes a lot of sense, but there wasn't a lot of background around it.
这很说得通，但当时相关的背景积累并不多。
05:41
And it was a risky bet.
而且这是个冒险的赌注。
05:43
We were in Bay Area.
我们当时就在湾区。
05:44
We were competing with like the likes of Facebook, Google for talk talent.
我们是在和像Facebook、Google这样的巨头抢顶尖人才。
05:49
But we took that risk.
但我们还是赌了这一把。
05:51
We were sort of forward looking in the sense that we figured that industrial AI physically I will become important.
我们也算是有前瞻性的，因为我们当时判断，工业AI会在物理世界变得非常重要。
05:57
And we tried to succeed and we did succeed to a large extent.
我们努力去成功，也确实在很大程度上成功了。
06:03
And I think that sort of paved the way in some sense for management for our leadership
我觉得那在某种意义上为我们的管理层铺平了路，从CEO往下，让大家有信心，相信内部人才可以把日立带到下一个高度。
06:10
from the CEO downwards to have confidence and internal sort of talent to take Hitachi to the next level.
而如果回头看，我会说，要向前看，看看转角处有什么。
06:18
And I think if I look back, I would say look forward, see what's sort of around the corner.
这很难，但我觉得如果你花时间去思考，不只是去读，
06:25
It's difficult, but I think if you spend time thinking about it, not just reading about it,
而是真正去思考，去观察你周围行业正在发生什么。
06:31
but thinking about it as well, just stating what's happening in the industry around you.
然后去冒险，去押注一些不同的、新的东西。
06:35
And take a risk, take a bit on something different, something new.
并且冒一次险，在某个不一样、新鲜的东西上押一点注。
06:39
That is aligned with your company's direction as well, obviously.
这显然也和你们公司的方向是一致的。
06:43
Then I think that sort of shows your leadership in terms of looking ahead in ability to take risks and then to execute.
那我觉得这体现了你在前瞻性、承担风险和执行方面的领导力。
06:52
I think that's maybe that's what worked for me.
我觉得这也许就是我成功的原因吧。
06:55
I would say.
我会这么说。
06:56
Yeah, I love that answer because even Chris, I don't know if you remember, but the beginning of this year
是的，我很喜欢这个回答，因为克里斯，我不知道你还记不记得，但今年年初
07:02
I think we said, Chase and we do a kind of forward looking episode usually at the beginning of each year.
我记得我们说过，蔡斯和我们通常会在每年年初做一期展望未来的节目。
07:08
One of the things we talked about about kind of coming into its own was this idea of physical AI, I think, if I'm remembering right.
我们当时聊到的一个正在崭露头角的话题就是实体AI这个概念，如果我没记错的话。
07:16
And so you were very much ahead of that as you mentioned, you know, you took maybe a gamble or a risk looking towards that direction.
而你显然走在了那前面，正如你所说，你知道，你当时朝着那个方向下了一个赌注或者说冒了一次险。
07:23
I'm wondering for the listeners who might not be as familiar with industrial AI, physical AI, these sorts of terms.
我想问一下，对于那些可能不太熟悉工业AI、物理AI这类术语的听众来说。
07:31
If you could just help us understand kind of maybe what that meant when you started getting into it and what that means now.
你能帮我们理解一下，当初你刚开始接触时它意味着什么，现在又意味着什么吗？
07:39
If those things are different, you know, in any sort of way.
如果两者有什么不同的话——你知道，任何方面的不同。
07:43
That's a very good question, Daniel.
这是个很好的问题，Daniel。
07:45
And I think it certainly doesn't mean the same thing.
我觉得它肯定已经不再是同一个意思了。
07:48
Nothing means the same thing anymore.
现在所有东西都不再是原来的意义了。
07:50
But when we first started out, we were trying to solve industrial problems.
但当我们刚开始的时候，我们想解决的是工业问题。
07:56
So if you look at the industrial value chain from design all the way to manufacturing.
所以你看工业价值链，从设计一直到制造。
08:01
And this is not just manufacturing, but you think about power plants.
而且这不光是制造业，还有发电厂。
08:04
You think about rail systems. You think about any industrial system.
你想想轨道交通系统，再想想任何工业系统。
08:08
You could sort of organize the challenges in say maintenance design production.
你可以大致把这些挑战分成这么几类：维护、设计、生产。
08:16
You look at these verticals.
你去看这些垂直领域。
08:17
And if you took a step back, you realize that although every vertical is different, the data itself is different.
如果退一步看，你会发现，虽然每个垂直领域不一样，数据本身也不一样。
08:23
But there are a lot of commonalities in terms of the problems that you would solve.
但你要解决的问题之间有很多共性。
08:27
And in terms of the techniques that you could bring to bear on there.
而你能用上的技术手段也有共性。
08:31
So initially it was much more around prediction, like you failure predict.
所以一开始主要是各种 prediction，比如说 failure prediction。
08:34
That was a classical predicament in this problem.
那正是这类问题里面一个典型的困境。
08:37
And then can you recommend the right repair to a technician?
然后你能不能给技师推荐合适的维修方案？
08:41
Whether that technician is an automotive technician or sort of in some other domain doesn't really matter.
无论那个技师是汽车技师，还是其他领域的技师，这都没关系。
08:46
So but the math behind it was somewhat similar.
但背后的数学原理其实是有些相似的。
08:49
And that's how it started out.
我们就是这么起步的。
08:51
And those are the kinds of problems we were we were solving around maintenance around quality,
我们当时在维护、质量这些方面解决的就是此类问题——
08:59
how do you sort of predict if the part was going to fail and so forth.
比如怎么去预测一个部件会不会出故障，诸如此类。
09:02
And then as the industry evolved both the physical industries,
后来随着行业的发展，实体行业——
09:09
meaning physical companies like say Hitachi decided to collect more data.
也就是像日立这样的实体公司，决定去收集更多的数据。
09:13
And sort of the AI industry also was evolving towards deep learning.
当时AI行业也在朝着深度学习的方向演进。
09:17
And as the technology was becoming more mature, then computer vision related problems became much more important.
随着技术越来越成熟，计算机视觉相关的问题就变得重要多了。
09:24
And can you detect defects on the surface?
你能不能检测出表面的缺陷？
09:27
Can you count number of cars in a parking lot kind of a problem?
你能不能数出停车场里有多少辆车，这种类型的问题。
09:31
And then as we got into a large language models,
然后当我们进入大语言模型时代，
09:35
then there was further expansion of what you could do in the industrial world itself.
工业领域本身能做的事情又进一步扩展了。
09:40
So one big move obviously is what we what people call physical AI now,
所以一个很大的转变，显然就是现在大家所说的物理AI。
09:44
which is sort of contested definition, but the idea that you could do robotics at scale.
这算是一个有争议的定义，但关键在于你能够大规模地应用机器人技术。
09:49
So part of the work that we've done on automation with traditionally reinforcement learning now could sort of be
因此，我们在自动化方面所做的工作，传统上使用强化学习，现在某种程度上可以
09:56
then much more formal larger basis with sort of around robotics.
然后以更正式、更宏大的基础，围绕机器人技术展开。
10:00
And you make multiple robots work together or even a robot can work in this environment or not.
而且你让多个机器人协同工作，或者甚至一个机器人能否在这种环境中工作。
10:05
And also the traditional sort of industrially I kind of problems also became on a larger scale.
同时，传统上那种工业类的问题也变得更加大规模。
10:11
So people started expecting much better answers a way for humans to communicate.
于是人们开始期待更好的答案，以及一种人类交流的方式。
10:16
Metaverse sort of becomes becomes important where you could construct these sort of virtual worlds for training,
元宇宙某种程度上变得越来越重要，你可以在其中构建这类虚拟世界用于训练，
10:24
for problem resolution, right?
用于解决问题，对吧？
10:26
So in some sense the the field has moved along with the maturity of the AI technology as well.
所以在某种意义上，这个领域也是随着AI技术的成熟度一起往前走的。
10:34
So that's it. And then this is a really exciting moment to be in that space as well, I would say,
就这样。而且，我想说，身处这个领域也是一个非常激动人心的时刻。
10:38
because a lot of a notion will come there.
因为那里会出现很多新的想法。
10:40
Yeah, I know in you know that you have you've run labs both in North America and Japan.
对，我知道，你在北美和日本都运营过实验室。
10:46
And as we're talking about physical AI at this point, it occurs to me as I'm listening to you,
而且，既然我们现在在聊 physical AI，我听你说的时候突然想到，
10:53
that you know we have a global audience here and people from different parts of the world have different experiences
你知道，我们这里有全球观众，来自世界各地的人有着不同的经历，
11:01
in terms of this move into physical AI and in some of these processes.
在迈向 physical AI 以及其中一些过程方面。
11:06
And I would imagine, I don't want to put words in your mouth, but I would imagine your time in Japan,
而且我猜——我不想替你把话说出来——但我觉得你在日本的时候，
11:12
there's a lot more robotics out there.
那里的 robotics 要多得多。
11:14
And as we're all sitting here in the United States, there's a little bit less exposure
而且，我们现在都坐在美国这边，大多数普通人对于 physical 和 robotics 的接触确实会少一些。
11:20
to most people out there in terms of physically and robotics.
我觉得，就我来看，我们在这方面是有点落后于世界其他地区的。
11:24
I think we're kind of in my view we're kind of lagging other parts of the world in that capacity.
我挺好奇，当你在看这些事情的时候，你对全球不同地区的经历是什么样的？
11:30
I'm kind of curious as you as you're looking at this kind of what your experiences were about different parts of the globe
以及这些经历是如何推动 physical AI 这个概念在你说的这些不同场景中向前发展的？
11:38
and how that's moving the notion of physical AI forward in these various contexts you're talking about.
还有，你知道，随着 embodied intelligence 越来越多地和 robotics、edge computing 这些结合起来发挥作用，你对此有什么想法？毕竟你有这些经历。
11:46
And maybe you know with the advent of kind of embodied intelligence coming more and more into play
这挺有意思的，让我想到这一切其实有个文化维度。
11:51
with robotics and edge computing and stuff, what your thoughts are around that, you know with your experiences.
加上机器人、边缘计算这些，你对此有什么想法，结合你自己的经验来说。
11:57
That's a interesting sort of reminds me of cultural dimension to all of this.
这挺有意思的，让我想到这里面其实有文化层面的维度。
12:02
So what I relearned and I should not make a cultural generalization and often people,
所以我重新学到一点——而且我不应该对文化一概而论——很多时候人们，
12:07
but I learned that at least say folks in Japan are much more open to robots than say folks in North America.
但我发现至少可以说，日本人比北美人更接受 robots。
12:16
I think that is something cultural, I can't explain it to why.
我觉得这是文化上的东西，我也说不出为什么。
12:18
And that's why even like very many years ago we were thinking about robots for elderly support.
这就是为什么即使在很多年前，我们就在考虑用 robots 来帮助老年人。
12:24
Even if they could not move, at least they could talk and understand and be empathetic.
即使它们不能动，至少能说话、能理解、能表达同理心。
12:28
So getting the empathy right was a very hard problem.
所以把同理心做到位是一个非常难的问题。
12:31
And just the way you know a machine would talk to a human.
就是你知道的，机器跟人说话的那种方式。
12:36
It's how do you introduce the right sort of the language, the empathy, the warmth that typically humans have for each other.
也就是如何引入合适的语言、同理心和温暖——就是人与人之间通常有的那种东西。
12:43
And so in that sense, so for the use cases like elderly care, I would say yes.
所以从这个意义上说，对于像老年护理这样的使用场景，我会说是的。
12:50
So the United States was sort of behind other sort of Asian countries,
所以 United States 在某种程度上落后于其他一些亚洲国家，
12:55
not all the Asian countries, but like the leading countries in Japan, China and so on and so forth.
不是所有亚洲国家，而是像 Japan、China 这样的领先国家，等等。
13:00
And then when it comes to industrial robotics as well, you're right, the center of gravity is not in North America.
而且说到 industrial robotics，你说得对，重心不在 North America。
13:06
Like the center, if you think about large language models, the center of gravity is North America.
比如说，如果你想到 large language models，重心就在 North America。
13:10
But in sort of in robots for the industrial world, the center of gravity is sort of in China today.
但在工业世界的机器人领域，今天的重心某种程度上在中国。
13:17
You've seen sort of the robots to multiple things.
你也看到了，机器人被用于多种事情。
13:20
And but I think we are catching up.
不过我觉得我们正在迎头赶上。
13:23
We have some excellent startups that are sort of trying very many new ways for this.
我们有一些很棒的初创公司，正在为这个尝试很多新方法。
13:29
The robots to sort of work in the physical world, right, whether they are for industrial use or whether they are for sort of more commercial use.
这些机器人是要在物理世界里工作的，对吧，不管是工业用途还是更偏向商业用途。
13:38
And I would say the commercial use robots would be the first one to have an impact,
而且我觉得商业用途的机器人会最先产生影响，
13:42
because a lot of industrial processor also already in some sense robotized,
因为很多工业流程在某种程度上也已经机器人化了，
13:47
because there's task specific, they don't need to be that general.
因为任务很特定，它们不需要那么通用。
13:51
Like more general robotics are needed, more for sort of human environment,
而更通用的机器人技术，更多是用于人类环境这种场景，
13:55
whereas a lot of factories are quite automated already, right.
但很多工厂其实已经很自动化了，对吧。
13:58
And I think that's what that's where the next sort of battle lines are.
我觉得那大概才是下一轮战线所在。
14:03
And I think in that there is, I mean, it's, it's anyone's came right now, I would say.
而且我觉得在那方面，我的意思是，现在真的谁也说不准，我会这么说。
14:08
Because some of the underlying math is where, you know, we are better than anyone else,
因为有些底层数学，你知道，我们比任何人都强，
14:13
some of the underlying mechanics, maybe other other countries are better than us.
有些底层机制，也许别的国家比我们做得更好。
14:17
But yes, but culturally you are spot on Christ, right.
但是没错，在文化方面你说得非常准，Christ，对吧。
14:20
So there was a broader acceptance.
所以就出现了更广泛的接受度。
14:23
And maybe that is still there of robots for day-to-day interaction in Japan,
而且也许在日本，那种让机器人参与日常互动的接受度现在依然存在。
14:31
does that answer question or did I sort of take it down sort of?
这算回答到你的问题了吗，还是我有点跑偏了？
14:34
That was a great answer, I appreciate it.
回答得很好，谢谢你。
14:36
I hope that you're finding this episode practical and helpful.
我希望你觉得这期节目很实用、很有帮助。
14:42
As the name suggests, we want this podcast to be practical, not just hype.
就像名字说的那样，我们希望这个 podcast 是实用的，而不只是炒作。
14:47
And that's one of the reasons why we partnered with the Midwest AI Summit,
这也是我们和 Midwest AI Summit 合作的原因之一，
14:52
which is happening October 15th in Indianapolis.
这场峰会将于 10 月 15 日在 Indianapolis 举行。
14:56
This summit is more than just a bunch of hyped talks.
这次峰会不只是一堆炒作性质的演讲。
15:00
There's really practical things there, like an AI engineering lounge,
那里确实有很多实用的东西，比如 AI engineering lounge，
15:04
where you can sit down with experienced AI practitioners, like myself and others,
你可以和像我及其他同行这样有经验的 AI practitioners 坐下来，
15:09
and work through your architecture, your tool questions, your roadmap, your strategy, etc.
一起解决你在 architecture、tool、roadmap、strategy 等方面的问题。
15:15
And that way you can actually leave with a huge value from the event.
这样你就能真正从这次活动中收获巨大价值。
15:20
Again, the event is happening October 15th in Indianapolis, some amazing speakers,
再说一次，活动是10月15日在Indianapolis举办，有一些很棒的演讲者，
15:26
some practical advice. Don't, don't miss this event.
还有一些实用建议。千万别，别错过这个活动。
15:29
Make sure you're there October 15th in Indianapolis.
确保你10月15日到场，在Indianapolis。
15:33
You can check us out at MidwestAISummit.com
你可以在MidwestAISummit.com上看我们的信息。
15:37
and you can use practical AI-20 to get 20% off registration.
而且你可以用practical AI-20在注册时享受8折优惠。
15:42
Again, Midwest AI Summit and you can use practical AI-20 to get 20% off registration.
再说一遍，Midwest AI Summit，使用practical AI-20注册可享8折优惠。
15:50
So Chitanya, I want to circle back to a comment that you made
那么Chitanya，我想回到你之前说的那个观点。
15:54
as you were talking about the arc of your career and landing at RAC space.
当你谈到自己职业发展的轨迹，最后落脚在 RAC space 的时候，
15:59
You mentioned something to the effect of going with the technology,
你提到过一种说法，大意是“跟着技术走”，
16:03
but now moving to a place where you could make it more accessible and safe,
但现在却转到了一个能让技术更普及、更安全的位置，
16:10
and etc. and operationalize it if I'm understanding you, right?
以及，等等这些，如果我没理解错的话，还要把它落地运营，对吧？
16:15
Could you help us understand that dynamic a little bit more around kind of the technology?
能不能帮我们再多理解一下这种动态，就是围绕技术本身的这种动态？
16:21
Now it almost seems like with AI everything is feasible,
现在看起来，有了 AI 之后，似乎一切都是可行的，
16:26
but not everything is easy, right?
但并不是每件事都那么容易，对吧？
16:28
Or not everything is, you can't operationalize everything, you can't make it accessible.
或者说，不是每件事都——你没办法把所有事情都落地，也没办法让它们都普及。
16:33
Could you help us understand your thought process around maybe that transition
你能不能帮我们理解一下你当时的思考过程，就是那个转变——
16:38
then from that world of industrial AI to now when you're thinking a lot
从工业AI那个世界，到现在你花很多时间思考
16:43
about these types of issues around operationalizing, scaling,
这些关于落地运营、规模化、
16:47
making accessible these technologies in an actual useful and safe way?
让这些技术以真正有用且安全的方式变得可及的问题？
16:55
So when you're working with sort of our industrial partners,
所以当你和我们的工业合作伙伴合作的时候——
17:00
both internal and external, actually, we were playing a team of highly trained
无论是内部还是外部，其实我们当时是在跟一支训练有素的团队较量，
17:04
like researchers, engineers, PhDs and all of that.
里面有研究员、工程师、博士什么的。
17:08
And so we will take up a customer problem, we'll understand the data,
所以我们会接手一个客户的问题，去理解数据，
17:12
we'll build the model, and then we'll work with the customer deploy it.
我们会先build model，然后跟客户一起deploy。
17:16
So and that deployment just not meant sort of throwing the model over the wall,
所以这种deployment并不是说把model直接扔过墙就算了，
17:20
but you sort of work with the customer, you talk to them,
而是你会跟客户一起协作，跟他们聊，
17:23
you understand exact business pain point, you understand the workflows,
去理解他们具体的业务痛点，理解他们的workflow，
17:26
and then you integrate your model in the workflow.
然后把你的model integrate到workflow里面。
17:29
And typically that model was small enough that you could have it in their own environment.
而且通常那个model都足够小，可以放在他们自己的environment里面。
17:34
So and this is what we did again and again.
所以我们反复就是这样做的。
17:37
But if you now look and if you look back maybe even five years,
但如果你现在回头看，甚至可能回看五年之前，
17:42
most of us were not using AI directly in our day-to-day world, right?
我们大多数人当时并没有直接在日常生活中使用AI，对吧？
17:46
So AI was always mediated through maybe a solution that was developed,
所以AI总是通过某个解决方案来间接发挥作用，那个方案可能是由某个模型开发、部署和运营的，对吧？
17:54
deployed and operated by some by some model, right?
所以是他们在承担那个责任。
17:57
So they took that responsibility.
现在有了生成式AI，AI已经变得大众化了，对吧？
17:59
Now with generative AI, AI has become democratized, right?
如今有了生成式AI，AI已经变得普及化了，对吧？
18:03
Everyone has access to AI, every enterprise wants to use AI.
每个人都能使用AI，每家企业都想用AI。
18:06
And you don't have this team of PhDs and researchers everywhere
但并不是哪里都有这样一批博士和研究员，
18:10
who can sort of build a model and deploy it.
能帮你把模型建好并部署上线。
18:13
First of all, it's cost prohibitive, right?
首先，成本就高得离谱，对吧？
18:15
These large language models are typically very difficult to build,
这些大语言模型通常非常难构建，
18:18
they're very expensive to build, right?
训练成本极其高昂，对吧？
18:20
So the whole model of how you would bring AI machine learning
所以，把AI和机器学习带给企业的那套传统模式，
18:25
to an enterprise changes with generative AI.
会随着生成式AI而发生根本改变。
18:28
So that's the challenge, right?
所以这就是挑战所在，对吧？
18:30
So at a high level, but what does that mean in practical terms, right?
从宏观层面看是这样，但落到实际操作上又意味着什么呢，对吧？
18:33
The number of things, right?
涉及的事情数量，对吧？
18:34
So you ask a question, right?
所以你问一个问题，对吧？
18:36
I use, for example, some language language model.
比如我用某个 language model。
18:39
Now I could simply ask, what's the capital of said all the way?
那我就可以直接问，沙特阿拉伯的首都是什么？
18:42
So for that, I don't need to go to say,
所以对于这个，我不需要去，比如说，
18:44
GPD, waste my tokens are expensive.
GPT，浪费我的 tokens，tokens 很贵。
18:46
If I have a local model, I should ask it that question.
如果我有一个 local model，我就应该问它那个问题。
18:50
It will be much more cheaper.
那会便宜得多。
18:52
And I, and, right? So that's number one.
然后，对吧？所以这是第一点。
18:54
Number two is this is an issue raised by Satinadeela
第二点是Satinadeela提出的一个问题，
18:57
and even I think Jensen,
我觉得Jensen也提过，
19:00
whenever you, as an enterprise, say,
就是当你作为一个企业，说，
19:03
now you want to say, I want to operationalize AI,
现在你想说，我要把AI投入实际运营，
19:05
AI is accessible, I want my staff, and everyone to use AI.
AI是触手可及的，我希望我的员工，所有人都能用上AI。
19:09
Now, every time you ask a question, say,
现在每次你问一个问题，比如说，问一个大语言模型，从某种意义上说，你是在把你的数据发给它们。最新的模型纠错阶段网站，对吧？所以现在，如果你要部署自己的模型，假设你说我不会在企业内部开发自己的模型，或者用我自己的数据去部署，那你怎么保证它的行为是合规的？
19:12
to a large language model,
对于一个大型语言模型来说，
19:14
in some sense, you are sending your data over to them.
从某种意义上说，你是把数据发送给了他们。
19:17
So your data sovereignty is not guaranteed, right?
所以你的 data sovereignty 是得不到保障的，对吧？
19:21
So the way Satinadeela sort of said it,
所以 Satinadeela 那种说法，
19:24
you're losing your alpha, right?
你就是在失去你的 alpha，对吧？
19:26
That's your IP.
那是你的 IP。
19:27
So that's another problem.
所以这是另一个问题。
19:29
Then an AI, in some sense, today is jagged.
然后 AI 在某种意义上，现在是参差不齐的。
19:33
So meaning that there are some tasks for which AI is really good at.
也就是说，有些任务 AI 确实非常擅长。
19:38
So I can really write some very good sort of software program using AI,
所以我可以真的用 AI 写出一些非常好的软件程序，
19:42
right? I can code with AI.
对吧？我能用 AI 写代码。
19:44
But if you try to write an email with AI,
但是如果你试着用 AI 写一封邮件，
19:46
you realize that it's not all that great, right?
你会发现其实没那么好，对吧？
19:48
It's quite, right?
还是挺一般的，对吧？
19:49
So I get AI emails and I sometimes annoy me
所以我收到 AI 写的邮件，有时候它们挺让我烦的，
19:52
because they are sort of very verbose
因为它们有点过于冗长，
19:54
and they're very cliché in some sense, right?
而且在某种程度上很陈词滥调，对吧？
19:57
So although we thought that writing is the strength of AI,
所以尽管我们以为写作是 AI 的强项，
20:01
but it is turning out that it's good at it,
但结果发现它很擅长这个，
20:03
but not that great at it.
但也不是那么擅长。
20:04
Humans can still do better at writing emails, maybe.
也许人类在写电子邮件方面还是能做得更好。
20:07
Then AI can.
那AI也能做到。
20:08
So that means in terms of capabilities of where AI is good or bad,
所以这意味着，就AI能力擅长和不擅长的方面来说，
20:13
it's jagged, right?
它是参差不齐的，对吧？
20:15
So from an enterprise, they need to know, right?
所以对于企业来说，他们需要知道，对吧？
20:21
So this is where I should use AI.
所以这就是应该使用AI的地方。
20:23
And then to my first point earlier,
然后回到我之前的第一点，
20:25
whether I should use a local model
我是否应该用 local model，
20:27
because that's more expensive, should I use a larger model.
因为那个更贵，我是不是应该用更大的 model？
20:30
And then how do I preserve my alpha?
然后我该怎么保住我的 alpha？
20:33
So there's only three questions that are today difficult to answer.
所以今天只有三个问题是很难回答的。
20:36
Now the jaggedness of AI also is in terms of how do you guarantee
而 AI 的参差不齐也体现在，你怎么保证
20:41
that it behaves in a responsible manner, right?
它会负责任地行事，对吧？
20:43
So we all heard that anthropic, sort of,
所以我们都听说过 Anthropic，怎么说呢，
20:46
the latest model fact bugging phase website, right?
最新的模型事实纠错阶段网站，对吧？
20:52
So now, if you're going to deploy your own model,
所以现在，如果你打算部署自己的模型，
20:55
suppose you say I won't develop my own model within the enterprise
假设你说，在企业内部我不会开发自己的模型
20:59
or my own data and you deploy it,
或者用我自己的数据，然后你部署它，
21:01
how do you guarantee that behaves in a regulated,
你如何确保它在受监管的环境中表现得当？
21:03
in a way that is safe?
以一种安全的方式？
21:04
And also if you're using someone else's AI,
而且如果你使用的是别人的AI，
21:07
like you're using a large LLM,
比如你使用一个大型LLM，
21:09
how do you ensure that it behaves within Godrains, right?
你如何确保它在护栏内运行，对吧？
21:12
It behaves in a way that is suitable for your enterprise, right?
它以适合你企业的方式运行，对吧？
21:17
So the problem of governance and assurance becomes very important.
所以治理和保障的问题变得非常重要。
21:21
The government, the problem of how do you maximize the value of your AI,
治理，也就是如何最大化你的AI价值的问题，
21:25
the tokenomics becomes important.
代币经济变得重要。
21:27
The value of preserving your IP becomes important
保护你的知识产权的价值变得很重要
21:31
and the problem having the right architecture
以及拥有正确架构的问题
21:34
so that you can cater to the multiple needs within the enterprise
这样你就能满足企业内部的多种需求
21:38
from ship programming to writing emails to summarizing to research,
从船舶编程到写邮件到总结到研究
21:42
can be done in sort of a safe, guaranteed manner
都能以某种安全、有保障的方式进行
21:46
so that you can sort of interchange the models if needed,
这样你在需要的时候就可以更换模型
21:50
you can sort of, right?
你可以这样做，对吧？
21:51
So because the model technology is changing very rapidly,
因为模型技术变化非常迅速
21:54
so all of that requires a very systematic way of thinking about your AI architectures.
所以这一切都需要一种非常系统化的方式来思考你的AI架构。
21:59
And I think this is the next frontier.
而我觉得这就是下一个前沿。
22:01
This is the next challenge, right?
这就是下一个挑战，对吧？
22:02
So now, how do we go from simply sort of chatting with an AI through an interface
那么现在，我们怎么从仅仅通过一个界面和 AI 聊天，
22:10
to working in the enterprise where there are people, processes in all of the complications?
到真正在企业里落地，面对各种人员、流程，还有那些复杂的状况？
22:14
There's a very long-winded answer, but I hope sort of it's addressed.
这个问题答案会很长，但我希望能把它说清楚。
22:18
Great question.
好问题。
22:19
Yeah.
对。
22:20
You're not only was it good, I'd like to actually get you to extend it a little bit
不只是说得好，我还想让你再展开一点，
22:23
by throwing a couple of extra logs on the fire.
再往火里添几根柴。
22:28
One of the challenges that we see in industry right now,
我们现在在行业里看到的一个挑战是，
22:32
and it's been evolving over the past, especially over the past year,
而且它一直在演变，尤其是在过去这一年里，
22:36
is where open weights or open source models are available from.
就是 open weights 或 open source 模型的来源。
22:41
And there's so many external considerations that get brought into bear
而且有太多外部考量会被牵扯进来，
22:47
as, you know, while, you know, to your point earlier,
就像，你知道，不过，嗯，你之前说的那个点，
22:51
that the center of gravity for model development, you know,
就是模型开发的重心，你知道，
22:54
for L&M development, maybe still in the U.S.,
也就是 L&M 的开发，可能仍然是在 U.S.，
22:58
a lot of those are closed.
其中很多都是 closed 的。
22:59
The number of available open weights models kind of shrunk a little bit with,
可用的 open weights 模型数量有点缩水了，随着……
23:04
you know, meta is kind of going away from that.
你知道，Meta有点在偏离那个方向。
23:06
And Nvidia started stepping up a little bit more because the rest of the commercial industry
而Nvidia开始更多地站出来，因为美国其他商业行业
23:13
in the U.S. was reducing.
在缩减。
23:14
And we're seeing an explosion of capability in terms of new models that are kind of rapidly catching up
而我们看到能力在爆发式增长，新模型正在迅速追赶，
23:21
with very close followers, or potentially equal from China.
跟得很紧，或者可能从中国那边达到持平。
23:26
And overlaying all this, you have all these countries have their various exports,
而且在这之上，所有国家都有自己的各种出口管制、进口顾虑，什么能用什么不能用。
23:34
concerns, import concerns, what you're allowed to use.
这让整个生态系统对各行各业的企业来说都变得相当复杂。
23:38
And that makes it quite complicated as an ecosystem for companies in various businesses
如果你还没开始，那就挑一两个影响比较大的问题，从那儿入手。
23:45
to try to figure out what makes sense for me.
去试着搞清楚什么对我来说才是合理的。
23:47
You talked about governance, you talked about data sovereignty,
你谈到了 governance，也谈到了 data sovereignty，
23:51
how do you navigate, you know, with some of these big issues,
你知道，面对这些比较大的问题，你会怎么去应对？
23:55
they go beyond the technology of AI and the implementation of AI
它们已经超出了 AI 技术和 AI 实施本身的范畴，
23:59
and can affect, you know, management concerns all the way up to the CEO and the board of directors.
而且会影响，你知道，从管理层关切一直到 CEO 和董事会。
24:05
How do you start, if you're a company now, it's late 2026,
如果你现在是一家公司，时间已经是 2026 年底，
24:09
and all this is rapidly developed, how do you look at all these
这一切又发展得这么快，你怎么看这些，
24:14
and make decisions for strategic interests in your organization going forward
然后为你的组织接下来的战略利益做决定呢？
24:19
because it's quite the quagmire at this point.
因为现在这已经是个泥潭了。
24:22
Yeah, no, and it's quagmire and the changes that are dizzing pace as well, right?
对，没错，而且不仅是泥潭，变化速度也快得让人头晕，对吧？
24:27
And so there is no easy answer to that.
所以这个问题没有简单的答案。
24:30
I don't want to address the question of open model stove.
不过，我不想讨论 open model 这类事。
24:33
And I know you said that we are somewhat behind as in terms of open model.
而且我知道你说过，我们在 open model 方面有点落后。
24:39
But I think the beauty of United States is that the right incentive we really step up.
但我认为 United States 的好处是，只要有正确的激励，我们真的会加把劲，对吧？
24:43
Right? So now we realize, may the AI commit realises as well that look with open source models,
所以现在我们意识到——也许 AI community 也意识到了——你看，有了 open source models，
24:51
maybe they're from China today, obviously Nvidia is doing a great, great job in it, right?
也许它们今天来自中国，显然 Nvidia 在这方面做得非常非常好，对吧？
24:55
There is a lot of merit in that.
这确实很有道理。
24:57
And from what I know of people I talk to and friends I talk to,
而且据我所知，从我和朋友、业内人士聊的情况来看，
25:00
I think it's a matter of time before our open source models will be taught of everyone else.
我觉得迟早有一天，我们的开源模型会领先于其他人。
25:06
That's number one, right?
这是第一点，对吧？
25:08
On number two, like how do enterprises get started?
第二点，企业该怎么起步呢？
25:13
I think they have to sort of fix few things, I would say, right?
我觉得他们得先解决一些问题，是吧，可以这么说。
25:17
Fix few things, meaning they need to sort of understand how much of their workload is sensitive today,
解决一些问题，意思就是他们需要先搞清楚，自己现在有多少工作负载是敏感的，
25:22
meaning if you are saying doing an HR query, maybe it's okay to go to NLM.
也就是说，如果只是做一个HR查询，可能交给LLM也没问题。
25:26
So one thing you sort of figure out what data is something you really want to work around.
所以有一点你会慢慢搞清楚，什么样的数据才是你真正想围绕它来工作的。
25:31
There ideally you should sort of think about local models, open weight models in your own environment that you control.
理想情况下，你应该想想用 local models、open weight models，跑在你自己掌控的环境里。
25:39
So I would say that's one North Star you try to try to fix.
所以我觉得那就是一个你努力去锚定的 North Star。
25:42
Don't marry into any model family because they will swap in and out both for commercial reasons, geopolitical reasons.
不要跟任何 model family 绑死，因为它们会因为商业原因、地缘政治原因被换来换去。
25:50
So all of that will sort of evolve, change.
所以这些东西都会不断演化、改变。
25:53
So don't marry into any sort of, my marry into an architecture will be of thinking.
所以不要陷入任何架构上的执念，或者说是对 architecture 的固化思维。
25:58
Meaning that these are the workloads that I can push to an LLM outside LLM.
也就是说，要区分哪些 workload 是我可以交给外部 LLM 的。
26:05
These are workloads that I need to have on-prem or in a growing environment.
哪些 workload 需要放在 on-prem 或者一个不断扩展的环境里。
26:10
And what are my governs and assurance layer that I want to have?
那我要的 governance 和 assurance layer 又是什么呢？
26:16
So what are the properties that are important to my enterprise?
所以对我的企业来说，哪些特性是重要的？
26:19
That I really need to enforce in the way I work with AI.
那些是我在使用 AI 的工作方式中真正必须落实的。
26:23
So I think if you have some of those principles pinned down, it becomes an easier way to get started.
所以我觉得如果你把这些原则定下来，起步就会更容易。
26:30
And I think most, and then a year or so, I would have said, pick one problem up and then do it all the way.
而且我觉得，大概一年前的话，我会说：先挑一个问题，然后把它做到底。
26:37
And then pick the second one up, then do it all the way.
然后再挑第二个，也做到底。
26:40
Because at least six months to a year ago, there were so many pilots and not as much impact on the bottom line or top line of a corporation.
因为至少在六个月到一年前，有太多的 pilot 项目，但对公司的利润或营收没有产生太大影响。
26:48
But I think people are learning that lessons, but that's the other thing.
但我认为大家正在吸取这些教训——不过那是另一回事。
26:53
Pick one or two sort of problems that are meaningful in terms of impact if you're not started yet and start with there.
但对大多数已经在路上、已经起步的企业来说，我会说别再想模型了，开始想架构——面向AI的企业架构。
26:59
But for most enterprises, which are already somewhere in the journey that have started, I would say stop thinking models and start thinking architectures, enterprise architectures for AI.
这大概就是我的“一句话建议”，如果你要的话，就这么去推进。
27:10
So that would be sort of my one liner, if you would, for how to go about it.
我特别喜欢Chetan在这期节目里强调的，随着我们进入agentic AI时代，运营主权有多重要。
27:17
I love how Chetan in this episode is emphasizing the need for operational sovereignty as we move into agentic AI.
当agent在你的基础设施里拥有自主性，并且触及你的关键系统时，保持控制、限制爆炸半径，绝对是至关重要的。
27:25
As agents have autonomy in your own infrastructure and touch your critical systems, it's absolutely crucial that you maintain control and limit the blast radius.
当agent在你的基础设施中拥有自主权并触及你的关键系统时，保持控制并限制爆炸半径绝对是至关重要的。
27:36
That's why I'm so privileged to be leading a company called Prediction Guard, which has a self-hosted control plane that allows you to maintain
这就是为什么我非常荣幸能领导一家叫Prediction Guard的公司，我们有一个self-hosted的控制平面，让你能够为你环境里运行的agents保持最低限度的agency，并限制这些agents的爆炸半径。
27:45
least agency for the agents operating in your environment and limit the blast radius of those agents.
它运行在你自己的基础设施里，你有完全的自主控制权，不管是在本地、air-gapped环境，还是在你云端的VPC里。
27:52
This runs in your own infrastructure, you have complete sovereign control, whether that's on-prem, air-gapped, or in your cloud VPC.
而且它在AWS和Azure的marketplaces上都有上架。
28:01
And it's available on the AWS and Azure marketplaces.
我鼓励大家去PredictionGuard.com斜杠practical AI看看我们。
28:05
I would encourage you to check us out at PredictionGuard.com slash practical AI.
再说一次，去PredictionGuard.com斜杠practical AI看看我们。
28:11
Again, check us out at PredictionGuard.com slash practical AI.
所以Chetan，你带我们聊到了架构思考这个点上。
28:16
So Chetan, you brought us to the point of talking through kind of thinking about architecture.
我想，我确实想回过头来聊聊，你是怎么思考这个问题的，以及怎么在RAC space里把它落地实现的。
28:22
And I want, I do want to come back to that here in a second in terms of how you're thinking about that and enabling that at RAC space.
我想，我确实想稍后回到这个话题，谈谈你在这方面是怎么考虑的，以及如何在RAC space里实现这一点。
28:29
But before I do that, I wanted maybe to get your perspective on a word that you used a little bit ago, which was sovereignty, which has to do.
但在那之前，我想听听你对一个你刚才用到的词的看法，就是“主权”这个词，它其实挺有深意的。
28:39
I think maybe there's listeners out there that are thinking, oh, I work for an industry company, I'm not a nation state.
我觉得可能有些听众会想，哦，我在行业公司工作，我又不是国家政府。
28:48
I don't need, I don't have a sovereign cloud.
我不需要，我没有主权云。
28:50
What is sovereignty have to do with me? Could you help clarify that?
主权跟我有什么关系？你能帮我澄清一下吗？
28:55
Because some people might be thinking or have different views of what that means.
因为有些人可能会想，或者对这个词有不同的理解。
29:01
And kind of bring it down to maybe the commercial or the industry setting.
然后把它落到商业或行业场景里。
29:06
What does sovereignty mean in that sense, both in terms of maybe privacy and control?
在这种意义上，主权意味着什么？比如隐私和控制方面？
29:11
So that's an excellent question Daniel.
Daniel，这是个非常好的问题。
29:14
And I think that's exactly the evolution that has happened.
我觉得这正是已经发生的演变。
29:17
Folks typically associate sovereignty with sort of a nation state.
人们通常会把主权和民族国家这类概念联系在一起。
29:22
That nation said is sovereign and it needs to have their own AI stack.
这个国家是主权的，它需要有属于自己的 AI stack。
29:26
And that's how sort of the conversation started.
差不多就是这样开始的吧。
29:28
But I think a few months ago, the conversation shifted because people realized that whenever you are interacting with a large language model, that is not sort of in your own environment, you are sharing your data, your context, your processes, your information.
但我觉得，几个月前话题就变了，因为人们意识到，每当你和一个 LLM 交互——而这个 LLM 又不在你自己的环境里运行——你就是在分享你的数据、上下文、流程和信息。
29:48
And that is an IP and given how powerful these AI tools can be.
这其实是个 IP 问题，而且想想这些 AI 工具现在有多强大。
29:54
So that is an IP in terms of your data, your knowledge that you're giving to someone else.
所以，你交给别人的数据和知识，这些都属于 IP。
30:03
And not only that it can be acted upon using AI to build solutions that might impact you as a corporation.
而且不仅如此，别人还可以用 AI 拿这些信息去构建解决方案，而这些方案可能会影响到你作为一家公司。
30:11
So the notion of the sovereignty sort of sort of comes down to not just a nation, but also an entity, enterprise entity.
所以“主权”这个概念，某种程度上不只跟国家有关，也跟实体——企业实体有关。
30:20
For example, and I'll sort of give me an interesting extension to it.
比如说，我再给这个概念做一个有趣的延伸吧。
30:24
And then that means that how do I protect my own IP?
那这样一来，问题就是：我该怎么保护我自己的 IP？
30:28
How do I protect my own data?
我怎么保护我自己的数据？
30:29
How do I ensure that my model behaves in the way that I want it to behave?
我怎么确保我的 model 按我想要的方式表现？
30:34
My AI behaves in a way, not a model, sorry.
我的 AI 是以一种方式表现的——不是 model，抱歉。
30:36
My AI behaves in a way that I want it to behave with sort of the governance and assurances that I think are appropriate in my environment, not someone else dictating to me what my governance should be.
我希望我的 AI 按我想要的方式表现，同时带有我认为适合自己环境的 governance 和保障，而不是由别人来规定我的 governance 应该是什么。
30:47
It is not someone else's constitution that I have to use, but my own sort of constitution as from a sovereign standpoint.
这不是我要被迫采用别人的 constitution，而是我自己的、从主权立场出发的 constitution。
30:53
And I think that idea will extend further as we go ahead, maybe to an individual as well.
而且我觉得这个想法会继续延伸下去，也许最终会到个人层面。
31:00
So we are quite sort of now, you see the idea of sharing our private information with the band and so to speak.
所以我们现在已经相当习惯这种想法了，你看——把私人信息分享给品牌（brand）之类的，可以这么说。
31:08
But I think this idea of sovereignty I think will eventually extend to humans as well, to us as well.
但我觉得 sovereignty 这种理念最终也会延伸到人类身上，也就是我们自己。
31:16
Maybe we'll say, look, how do I protect my own data when I'm interacting with these language models?
也许我们会说，看，当我和这些 language models 交互时，怎么保护我自己的数据？
31:21
Because people are sharing a lot of private information now.
因为现在很多人在分享大量隐私信息。
31:24
They are sort of using them as therapists, as guides, as friends.
他们有点把那些模型当成 therapist、guide、朋友在用。
31:28
So this notion of sovereignty will come all the way down.
所以这种 sovereignty 的概念会一路往下延伸。
31:31
And the idea is I am this entity.
意思是，我是一个独立的实体。
31:33
I am my own interest that are distinct from someone else's interest and I need to protect it.
我有我自己的利益，这些利益跟别人的不一样，我需要去保护它。
31:38
That's great. And I love, while I love that definition, I think it gets people thinking in the right direction.
那很棒。我喜欢这个定义，我觉得它能让人往正确的方向去思考。
31:44
But I also loved how you actually kind of made a distinction there when you talked about AI versus model.
但我也很喜欢你刚才在讲 AI 和 model 的时候，确实做了一种区分。
31:53
And you brought us to the point of thinking about architecture before.
而你之前已经把我们引向了思考架构的方向。
31:58
And I'm wondering if you can help us now that you're kind of you're at rack space, you're helping rack space think through the architecture that needs to be enabled for different enterprises.
我想问的是，既然你现在在Rackspace，你在帮Rackspace梳理不同企业需要什么样的架构支持，你能不能帮我们理一理这个。
32:09
Actually, that in itself becomes a little bit complicated because like you said, models are not the same as, quote, AI that you're deploying in the sense of, oh, maybe there's an agent that uses multiple models.
其实，这件事本身就有点复杂，因为就像你说的，模型并不等同于你部署的所谓“AI”，意思是，哦，也许有个agent会调用多个模型。
32:22
It has a harness. It connects to MCP servers. There's a governance element to it. There's an observability element to it, et cetera, et cetera.
它有个harness，它连接到MCP服务器，里面有治理的部分，也有可观测性的部分，等等等等。
32:32
It's almost like you can look at that AI stack and it can be very overwhelming to understand, you know, how to put all the pieces together.
这几乎就像你去看那个AI技术栈，理解怎么把所有东西拼在一起，会让人特别头大。
32:41
What is the right architecture? Could you help help us understand maybe how you're, I think, helping rack space enable.
正确的架构是什么？你能帮我们理解一下，我觉得，你是在怎么帮Rackspace实现这个的？
32:50
The architecture, the architectures that are important for people now and how you're encouraging your customers, your partners to think about that architecture, not just as a model, but as a whole architecture that's supporting the deployment of AI.
现在对人们来说重要的架构是什么，以及你如何鼓励你的客户、你的合作伙伴去思考这个架构——不只是把它看成一个模型，而是看成支撑AI部署的整个架构。
33:06
So in some sense, rack space today goes from, as you say, from chip to outcome, right, because we have an app partnership with AMD. We have our own data centers. So our stack goes all the way, right.
所以从某种意义上说，今天的Rackspace，就像你说的，从芯片到结果，对吧，因为我们和AMD有应用层面的合作。我们有自己的数据中心。所以我们的技术栈是贯穿到底的，对吧。
33:19
And for our customers, we provide the whole set of private AI kind of environment, private out kind of environment where they can safely run their AI in their own environment, completely controlled by them.
对我们的客户来说，我们提供一整套私有的 AI 环境，一种私有化的环境，让他们可以在自己的环境里安全地运行 AI，完全由他们自己掌控。
33:30
And this is what we are trying to do. I'm trying to build this sort of stack out for our customers partners and also internally at rack space.
这就是我们想做的事。我正在努力为我们的客户、合作伙伴，还有Rackspace内部，搭建这样一个完整的技术栈。
33:40
And I know I completely agree that this can be quite overwhelming, but I think it's not all that bad.
而这就是我们正在努力去做的。我正在为我们的客户、合作伙伴，以及 Rackspace 内部，构建这样一套 stack。
33:46
You should think, if you take a step back and think systematically about it, it's sort of following the same paradigms of architecture and design that have come before.
你应该退一步，系统性地去看这个问题，它其实遵循的是之前已经出现过的那些架构和设计范式。
33:55
So if you think about like you start with, say, compute layer, then you have your data. It's fine. And you have a model layer now by the model layer, it could mean not just not just sort of the large elements, but also your own model, right. There's a model library.
而且我知道，我完全同意，这确实可能会让人应接不暇，但我觉得其实也没那么糟。
34:10
And on top of it is an inference layer. So inference layer is how you get some intelligence out of a model. So that is the inference layer.
而在它上面是一个inference层。所以inference层就是你怎么从模型里获得某种智能。那就是inference层。
34:20
And on top of it, it's, I would say the harness and I think you bred the word harness and I think it's very important to think of harness as a sort of key construct in how you deploy your AI.
你应该这么想，如果你退一步看，系统性地思考一下，它其实是在遵循之前已有的那些架构和设计范式。
34:31
So harness is what in my mind ties a machine learning model to an outcome that you can use. So when, for example, people use a cloud code, it's not that you are directly working with the model itself.
所以harness，在我看来，是把一个机器学习模型和你实际能用的结果连接起来的东西。比如说，当人们用云代码的时候，并不是你直接在和模型本身打交道。
34:45
It is the coding harness right around the underlying machine learning model that enables you to do something very useful.
正是那个围绕底层机器学习模型的 coding harness，让你能做非常有用的事情。
34:54
So you take the same model and you build two different harnesses and you will get two different outcomes. So it's very important to think through what sort of a harness looks like.
所以同一个模型，你搭两个不同的 harness，就会得到两种不同的结果。所以非常重要的是，要想清楚 harness 应该是什么样的。
35:03
And harness simply thinks, think through of harness as sort of the machinery that sort of specifies the logic for for the underlying underlying AI model to use and and setting a set of tools sets that it could use right through MCP or whatever.
而 harness 呢，简单想就是一套机制，它规定了底层 AI 模型要用的逻辑，也给它设定了一组它能用的工具集，通过 MCP 或者别的东西来用。
35:24
And then in any enterprise, there will be multiple harnesses right so you will have a coding harness you might have a harness for your agents for HR might have a harness for agents for blah blah blah.
然后在任何企业里，都会有多个 harness，对吧？你会有 coding harness，也可能有给 HR 的 agents 用的 harness，还可能给其他什么什么的 agents 用的 harness。
35:35
And so ideally you should think of building an orchestration layer to manage multiple harnesses right and on top of it, you can then think of the consumption and around the whole thing should be wrapped I would say with the governance and assurance planes.
所以理想情况下，你应该考虑建一个 orchestration layer 来管理多个 harness；在它之上，再考虑 consumption。整个外面还应该包上 governance 和 assurance planes。
35:49
So if you think of this, this is not so different from how we thought about other sort of stacks in the past is a matter of just abstracting out that if the use cases and commonality and think from that point of view and once you do that sort of this falls out naturally at least to me.
所以如果你这么看，这和我们过去想其他各种 stack 的方式没什么不同。无非就是把 use cases 和它们的共性抽象出来，从这个角度去想。一旦你这么做，整个结构自然而然就出来了——至少对我来说是这样。
36:07
And and everything can then be done through sort of the way we have done that in the past through API's the specifications who says what to do how do they interact and that's how these multiple years contract.
而且所有东西都可以通过我们过去那种方式来做——通过 API 和 specification，规定谁做什么、怎么交互，多个层就是这样形成 contract 的。
36:19
So so yes, I mean, it looks overwhelming, but I think the design principles are the same like in the past and it should not be all that intimidating.
所以，是的，这看起来可能有点吓人，但我觉得设计原则和过去是一样的，不应该那么让人畏惧。
36:30
No, I think that was I think that's quite an elegant way of of describing how architectural components fit together. I really I really found myself gravitating to to the way that you were explaining it.
不，我觉得那是一种非常优雅的描述方式，来解释这些架构组件是怎么拼在一起的。我真的发现自己很被你那种解释方式吸引。
36:44
And in my head as you were doing that, I had a question which I think you started you've already started to answer about I wanted to extend it a little bit and that is from from a customer standpoint as they are looking at the products and services that you know whatever their business is that they're offering their own customers.
在我听你讲的时候，我脑子里冒出一个问题，我觉得你已经开始回答了，但我想稍微延伸一下——就是从客户的角度来看，当他们看着那些产品和服务，你知道的，不管他们自己的业务是做什么的，他们要提供给自己的客户。
37:05
And there's some sense of stability to achieve that value you know they need the they need the product or service to be reliable to their own customers over time.
然后他们需要某种稳定性来实现那个价值，你知道的，他们需要那个产品或服务对他们自己的客户来说是长期可靠的。
37:17
And yet on the back end your customer who is providing that service to them is is trying to navigate those decisions that you just were describing in terms of what harnesses for what kinds of jobs I want to get done.
但另一方面，你的客户——也就是给他们提供那个服务的人——正在努力应对你刚才描述的那些决策，就是用什么“harness”来处理什么类型的任务，他们想完成什么样的工作。
37:33
And and I was in my question, which maybe I have a glimpse of was how do you manage the tumultuousness ticked and of the evolving set of models, the never ending set of new harnesses and stuff that are always coming out while keeping that customer experience downstream steady and level based on the value that you're trying to provide.
然后我的问题——我可能已经隐约看到答案了——就是你怎么去管理那种动荡不安，那种不断演变的模型集合，还有那些层出不穷的新harness之类的东西，同时还能让下游的客户体验保持平稳和一致，基于你想要提供的那个价值。
37:57
I'm guessing that that's somehow being managed through the orchestration layer in terms of how you're doing that right and you need to sort of build a balance for your own workloads.
我猜这某种程度上是通过orchestration层来管理的，对吧，你需要为自己的workloads建立一种平衡。
38:08
I think one thing people often gravitate towards is like oh these are benchmark this model is doing something else is better than the other in benchmark but those benchmarks are only guidelines in some sense they don't represent your workloads.
我觉得人们常常会倾向于那种想法，就是“哦，这个模型在benchmark上表现好，那个模型在另一个benchmark上更强”，但那些benchmarks在某种意义上只是参考，它们并不代表你的实际workloads。
38:21
And as we said earlier.
而且就像我们之前说的，错误的边界是参差不齐的，对吧，所以虽然某个模型可能在benchmark上表现很好，但这并不容易转化为它在你的任务上也会表现出色。所以我会说，拥有自己的评估体系非常重要，一旦你做了那个评估，某种程度上你就能为你的客户提供非常一致的体验。
38:24
The the boundary of air is jagged right so although a certain model might do very well on a benchmark doesn't easily translate into that it will do very well for your works as well so I would say having your own evolves is very important and once you make that evaluate then in some sense you can have very consistent experience for your customers.
空气的边界是参差不齐的，对吧，所以虽然某个模型在某个基准测试上可能表现很好，但这并不容易直接转化为它在你的工作场景里也会表现很好。所以我觉得，拥有自己的评估体系非常重要，一旦你做了那个评估，从某种意义上说，你就能给客户提供非常一致的体验。
38:44
So you can swap in models in an out based on your evaluation starts right so so you should say okay I have the same evolves from my previous model to model new model is cheaper lighter whatever or you know it's safe I will use this right so I think so you need to build an evaluate as part of the orchestration framework and that can then help you decide which sort of harness to go to sorry sorry daddy.
所以你可以根据你的评估结果随时换模型，对吧？所以你会说，好，我有同样的效果，从我的旧模型到新模型，新模型更便宜、更轻量，或者你知道它是安全的，那我就用这个。所以我觉得你需要把评估作为编排框架的一部分来构建，然后它可以帮助你决定该走哪个流程。抱歉抱歉，打断一下。
39:08
No I think that I was just going to say I think that ties into what you're saying about some of the intuition that we've had from building software over the years and architecture over the years certainly testing into and testing etc is a key piece of that and maybe the kinds of tests are slightly different or there's different ways of testing in this case but I love how you tied that that piece together and also thinking about the you mentioned the term outcome you know what outcome are you after.
不，我觉得我刚才想说的是，这跟你说的那些直觉是连在一起的，就是我们这些年做软件、做架构积累下来的直觉，当然测试啊、集成测试啊这些是关键的一部分，可能在这种情况下测试的种类稍微不同，或者有不一样的测试方式，但我很喜欢你把那部分串起来，还有你提到的那个词，结果，你知道你追求的是什么结果。
39:37
And are you really testing for for that outcome. I think that's a key piece of it.
而且你真的在测试那个结果吗？我觉得那是关键的一部分。
39:44
Just to add to that and if you remember when we were doing sort of these machine learning models for industrial use cases got this golden data set so before you could deploy the customer will say prove your model works on this golden data.
再补充一点，如果你还记得我们以前做那些工业场景的机器学习模型的时候，会有一个黄金数据集，所以在部署之前，客户会说，证明你的模型在这个黄金数据上有效。
39:59
So I think it's the same like now we call it a well that is at a more comprehensive but the basic idea is the same right you got to prove that your model works on data that is relevant to me before you deploy it for my customers in some sense.
所以我觉得现在也是一样的，现在我们叫得更全面一些，但基本思路是一样的，对吧？你必须在某种程度上证明你的模型在我相关的数据上有效，然后才能部署给我的客户。
40:14
So sorry sorry interrupt you by the way that's great make that connection.
所以抱歉抱歉打断你，不过那真的很棒，把那个联系起来了。
40:18
Yeah yeah I love that and I guess kind of as we as we get closer to the to the end here I want to give you a chance to you know you're sitting in this.
对对，我喜欢这个，而且我觉得随着我们接近尾声，我想给你一个机会，你知道你坐在这个首席AI官的位置上，你一路走来经历了这样的职业生涯，回到了Backspace。
40:28
Chief AI officer role you've kind of navigated the this career are coming to back space.
首席AI官这个角色，你算是走过了这条职业路径，现在又回到了这个领域。
40:36
Obviously I know you can't share you know anything that's not public but I wonder if you could give us a sense of what are the types of challenges.
显然我知道你不能分享任何非公开的信息，但我想知道你能不能让我们大概了解一下，你们面临的是哪些类型的挑战。还有你作为Rackspace的首席AI官，在进入今年剩余时间和明年的时候，你在鼓励Rackspace团队思考哪些事情，你脑子里在想什么？嗯，对，就是当你晚上躺下准备睡觉或者一天快结束的时候，哪些挑战是你最挂心的，需要在Rackspace内部，也许更广泛地在行业里解决，才能确保我们往前推进，产出我们想要的那种AI成果、可及性和安全性？
40:46
And the and the things that you're encouraging rack space to think about as we're going into the rest of this year and next year what's what's on your mind as that chief AI officer for rack space what's what's kind of.
有几件事，我脑子里大概有两三件吧。第一件显然是架构，对吧，就是你怎么让它更好，你怎么定义它，在哪个层面跟谁合作。不同类型的客户是什么样的，什么才说得通——没有一刀切的答案。所以围绕架构的那些总体设计问题，以及底层的那些组件，对我来说是既感兴趣又很重要的，我们也在思考这些。
41:01
Yeah what sorts of challenges are at kind of the top of your mind as you're as you're laying down to sleep at night or or coming to the end of the day what's at the top of your mind that that needs to be addressed you know within rack space and maybe the industry a little bit more broadly to make sure that we move forward to produce the types of AI outcomes and the accessibility and the safety that we're after.
另一件我在想的事，我觉得在这个意义上Rackspace挺独特的，但其他公司可能也能从我们这儿学到东西——我想建立一个，我不知道该用什么词，像一面镜子或者一个意义——我们现在说的是，如果你能在内部把它建起来并在内部用起来，那你就能出去把它卖给外部客户，对吧。所以如果我有一个AI解决方案，我已经在我自己的工作负载上验证过了，那我就能有信心去跟客户说，你可以用这个。所以我试图做的是，从内部的角度，围绕这个建立一套纪律，对吧。不是说中心项目价值更低，像个一次性项目那样。意图应该是，如果你做得好，那我们就把它轮换出去给我们的客户。所以我觉得这是第二点。第三点是关于治理、保障和编排。我觉得这三层是……
41:29
There was a couple of things and there's a one sort of the couple of things I think about right so one is obviously the architecture right so like how do you make it better how do you define it would be partner with for what layer.
对对对，我觉得这很棒。我很喜欢这个视角，当然从你坐的位置来看，你对要解决哪些挑战、什么最重要有很广的视野，所以总是很期待听到你的见解。
41:41
What are the different types of customers like what makes sense no there's no ones answer that so the general design questions around architecture and all then sort of underlying components of that of interest and of importance to me and we think about that.
不同类型的客户是怎样的？什么才合理？不，没有统一的答案。所以围绕架构的通用设计问题，以及底层的那些组件，对我来说既有趣又重要，我们也会思考这些。
41:57
The other thing I think about and I think in this case in the sense rack space is unique but like could be like other companies could learn from us as well I want to build I don't know what the right word of for it is like a mirror or meaning what we're saying now is if you can build it internally and use it internally then you can go sell it externally right so if I have an AI solution that I have proven on my own workloads.
另一件事我在想，我觉得在这方面，机架空间确实很独特，但其他公司其实也可以从我们这里学到东西。我想建立一个——我不知道该用什么词来形容——就像一面镜子，或者说一种意义。我们现在说的是，如果你能在内部把它建起来并自己用，那你就可以把它拿出去对外卖，对吧？所以如果我有一个AI解决方案，已经在我自己的工作负载上验证过了。
42:25
Then I can have confidence and go to my customer and say you could use this and so what I'm trying to do is from an internal standpoint build a discipline around it right so it's not that the central project is less valuable and it's like a throwaway project right the intent should be that if you do it well then we rotate it out to our customers so that's I would say that's number two and number three is around sort of the governance assurance and orchestration I think these three layers are.
这样我就能有底气去跟客户说，你可以用这个。所以我在内部想做的，其实是围绕这件事建立一套纪律，对吧。不是说中央项目价值更低，或者就是个随便做做的项目，不是的。意图应该是，如果你做得好，我们就把它轮换出去给客户用。所以我觉得这是第二点。第三点呢，是关于治理、保障和编排这一块。我觉得这三层就是。
42:54
Under serve today and especially in a sovereign sort of environment right so how do I bring sovereignty self sovereignty to our enterprises to our customers and in a way that it is cost effective it is safe it is reliable and I can orchestrate multiple workloads for them because any enterprise would require multiple kinds of workloads yeah I work loads for that and I think that is sort of the broader design question I think about I'm not sure if I answered your question correctly but sort of those are some things that I don't know.
在当今的环境下，尤其是在主权化的环境里，对吧？所以我怎么把主权、自我主权带给我们的企业和客户？而且要成本高效、安全、可靠，同时我还能为他们 orchestrate 多个 workload？因为任何企业都需要多种不同的 workload。对，我在这方面做了很多工作。我觉得这就是我在思考的宏观设计问题。我不确定有没有准确回答你的问题，但大致就是这些，有些我确实也还不太知道。
43:24
Yeah, yeah, I think that's great I love that perspective of course from from where you're sitting you have a broad view of what challenges you're trying to address and what's important so always eager to to get insight there.
对，对，我觉得这很棒，我很喜欢这个视角。当然，从你所在的位置来看，你对要应对的挑战和什么最重要有很全面的看法，所以我一直很期待能从中获得一些洞见。
43:40
This has been a great conversation Chetan I would very much encourage our listeners to check out what rack space is doing will include some links in our in our show notes is there any where in particular Chetan that obviously people can go to the website see what you're doing with AI but anything to highlight in terms of what rack space is doing kind of a jumping off point for people or or anything to highlight as we close out here in terms of what's what's currently available from rack space and what people can explore.
对对，我觉得这很棒。我很喜欢这个视角。当然，从你所在的位置来看，你对你所要解决的挑战和什么更重要有着宏观的视野，所以我也总是特别期待听到你的洞察。
44:10
So I would say that the questions that rack is trying to answer is where the industry is going towards so you should think about those as well right how do you maintain how you build AI that is sovereign that is safe that meets your outcomes and all of that right so so those are things that everyone should think about and those are the kind of solutions that rack space is bringing to market so please visit us send me an email or reach out to me on LinkedIn if you're specific sort of
所以我想说，Rack试图回答的问题是行业未来的发展方向，所以你也应该思考这些，对吧？你如何维持，如何构建一个自主的、安全的、符合你业务目标的AI，等等，对吧？所以这些都是每个人都应该思考的事情，而Rackspace正在将这类解决方案推向市场。所以请访问我们，给我发邮件或在LinkedIn上联系我，如果你有具体的……
44:39
questions as to what sort of rack space is doing and but I but I would encourage everyone is this rack space or not to think systematically about these problems because that's how you can ensure that AI succeed in your own environment.
Chetan，这次对话非常精彩。我非常希望听众们去了解一下 Rackspace 在做的事情，我们会在 show notes 里放一些链接。Chetan，有没有什么特别的地方，大家可以到网站上看看你在 AI 方面做了什么？有没有什么值得特别强调的，作为大家了解 Rackspace 的切入点？或者在我们结束前，有什么想重点说的？关于 Rackspace 目前已经有什么，以及大家可以去探索的内容？
44:54
That's that's great well thank you so much for joining us Chetan look forward to having you back on the show to give us some updates as you continue to advance with rack space. Thank you much so much for joining.
太好了，非常感谢你加入我们，Chetan。期待你再次来到节目，随着你在Rackspace的持续推进，给我们带来一些最新进展。非常感谢你的参与。
45:06
Thank you Chris and Daniel fantastic conversation. Thank you for hosting.
我想说，Rackspace 试图回答的问题是行业正在朝哪个方向发展，所以你也应该思考这些问题，对吧？你如何维护、如何构建一个主权化、安全、并且符合你业务成果的 AI？所有这些。这些是每个人都应该考虑的事情，也正是 Rackspace 正在推向市场的解决方案。所以欢迎访问我们，给我发邮件，或者在 LinkedIn 上联系我，如果你对 Rackspace 具体在做什么有疑问的话。但我也要鼓励每个人，不管是不是 Rackspace 的客户，都去系统性地思考这些问题，因为这样你才能确保 AI 在你自己的环境里取得成功。
45:20
All right that's our show for this week. If you haven't checked out our website head to practical AI dot FM and be sure to connect with us on LinkedIn X or Blue Sky.
好了，这就是我们本周的节目。如果你还没有访问过我们的网站，请前往Practical AI dot FM，并务必在LinkedIn、X或Blue Sky上与我们联系。
45:30
You'll see us posting insights related to the latest AI developments and we would love for you to join the conversation thanks to our partner prediction guard for providing operational support for the show.
你会看到我们发布关于最新AI发展的见解，很欢迎你加入对话。感谢我们的合作伙伴 prediction guard 为节目提供运营支持，去 predictionguard.com 看看吧。也感谢 break master cylinder 提供的节拍，还有你的收听。今天就到这儿，我们下周再见。
45:40
Check them out at prediction guard dot com also thanks to break master cylinder for the beats and you for listening that's all for now but you'll hear from us again next week.
请在Prediction Guard dot com查看它们。也感谢Breakmaster Cylinder为我们提供节拍，也感谢你的收听。今天就到这里，我们下周再见。