Less about Models; More about Architecture

0:00Welcome to the Practical AI Podcast, where we break down the real-world applications of artificial intelligence and how it's shaping the way we live, work, and create. Our goal is to help make AI technology, practical, productive, and accessible to everyone. Whether you're a developer, business leader, or just curious about the tech behind the buzz, you're in the right place. Be sure to connect with us on LinkedIn, X, or Blue Sky to stay up to date with episode drops, behind the scenes content, and AI insights. You can learn more at practicalai.fm. Now, onto the show. Welcome to another episode of the Practical AI Podcast. This is Daniel Whiteneck. I am CEO at Prediction Guard, and I'm joined as always by my co-host, Chris Benson, who is a principal AI and autonomy research engineer at Bucket Martin. How you doing, Chris? How you doing, great to be here. Looking forward to a conversation here.

1:01Yeah, excited to chat about all sorts of things, both in terms of background and current work with Chaitan Gupta today, who is chief AI officer at RAC space. Welcome, Chaitan. How are you doing? I'm doing good. Thanks, Daniel and Chris, for having me. Quite excited about the conversation. Let's start. Yeah. Like I say, we had bonded even yesterday when we were chatting about our backgrounds and in physics and mathematics. I know you started out in mathematics and spent a bunch of time at Hattachi. 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? Sure, sure. I know. So my background is actually quite diverse. It's sort of a typical of people in my role. I started off with the PhD in mathematics, then I joined Hilded Packard Labs with the research scientists, working on data mining, machine learning. Then I joined Hattachi as a principal researcher, I remember correctly.

2:02And then I sort of, you know, through the management ladder was when I left Hattachi in 2026, I was leading all of AI research at Hattachi globally. And this is a very strong team. And that's what I did. And we were focused a lot. We started with focus on industrial AI as we defined it at that time. And as the sort of industry matured, we sort of side looking at a broader spectrum of things. And that's what I was doing. And then I joined Rackspace and I've been here for four months. So it's been sort of a very, very fascinating journey. And in some sense, if I try to this podcast, I was thinking about my career. And I think you don't think about these things. And I realized that if you, in some sense, I have followed the trajectory of the whole community at a broad level. I'm going to try to remember, if you remember guys, you were all doing machine learning, making small models, solving specific problems.

3:04And there was a community in Bay Area that was focused primarily on cell accommodation systems. And as Chris, you would know, in companies like Lockheed Martin or Hattachi, we focused industrial problems. And so that's what I was doing for the first half of my career. And then more than the half. And then obviously deep learning became important beside looking at vision models, language models that have been blossomed in total, as language models. And in some sense, a lot of the traditional machine learning problems today are much more soluble with automated tools, with sort of wipe coding and so on and so forth. And the new challenge now is, yes, you can do all this in machine learning and AI. But how do you make it accessible to more people? How do it make it actionable? How do you make it much more sort of safe? And that's where sort of rack space comes in. So in some sense, if you look at the trajectory of problems, that's how I have traveled in some sense, like looking for trouble. Sort of speaking. I love that. So I've got it. I want to go back for a second.

4:05I'm going to drag you back as you kind of went through your timeline for a second, because as you were talking about kind of coming up through the ranks, the managerial ranks at Hitachi, you know, 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. 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. And yet you kind of came through the ranks at Hitachi. And I'm wondering if you have any thoughts about like what, 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 and they are aspiring to move up through the ranks themselves. 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 just from a kind of a growth learning standpoint if you, if you would

5:09mind sharing. Because that's a hybrid and I was more prepared to answer the idea of integration. Sorry. You know, but I think that's actually a very, very interesting question. I think one thing that I think helped in my career at Hitachi was we decided to take a bit on industrial AI. This was 2016, 2017 and not very many people were taking about it. So we decided a small research lab in North America. And I said, given where Hitachi is, it's a giant industrial concrete, it makes a lot of sense, but there wasn't a lot of background around it. And it was a risky bet we were in Bay Area. We were competing with like the likes of Facebook, Google for talk talent. But we took that risk. We were sort of forward looking in the sense that we figured that industrial AI, physically AI will become important. And we tried to succeed and we did succeed to a large extent. I think that sort of paved the way in some sense for management, for our leadership,

6:11from the CEO downwards to have confidence and internal sort of talent to take Hitachi to the next level. And I think, so if I look back, I would say look forward, see what's sort of around the corner. It's difficult, but I think if you spend time thinking about it, not just reading about it, but thinking about it as well, just stating what's happening in the industry around you. And take a risk, take a bet on something different, something new, that is aligned with your company's direction as well, obviously. Then I think that sort of shows your leadership in terms of looking ahead in ability to take risks and then to execute. I think that's maybe that's what worked for me. Yeah, I love that answer because even Chris, I don't know if you remember, but the beginning of this year, I think we said, Chatea and we do a kind of forward looking episode usually at the beginning of each year. And one of the things we talked about about kind of coming into its own was this idea

7:12of physical AI, I think, if I'm remembering right. And so you were very much ahead of that, as you mentioned, you took maybe a gamble or a risk looking towards that direction. I'm wondering for the listeners who might not be as familiar with industrial AI, physical AI, these sorts of terms, if you could just help us understand kind of maybe what that meant when you started getting into it and what that means now, if those things are different, you know, in any sort of way. No, that's a very good question, Daniel. And I think the, it certainly doesn't mean the same thing. Nothing means the same thing anymore. Right. But when we first started out, we were trying to solve industrial problems, right? So if you look at the industrial value chain from design all the way to manufacturing, and this is not just manufacturing, but you think about power plants, you think about rail systems, you think about any industrial system, you could sort of organize the challenges in say maintenance, design, production.

8:16You could look at these verticals. And if you took a step back, you realize that although every vertical is different, the data itself is different, but there are a lot of commonalities in terms of the problems that you would solve. And in terms of the techniques that you could bring to bear on there. So initially it was much more around prediction, like you failure predict that was a classical prediment in this problem. And then can you recommend the right repair to a technician, whether that technician is an automotive technician or sort of in some other domain doesn't really matter. But the math behind it was somewhat similar. And that's how it started out. And those were the kinds of problems we were solving around maintenance, around quality, how do you sort of predict if the part was going to fail and so forth. And then as the industry evolved both the physical industries, meaning physical companies like say Hitachi, they started collecting more data. And sort of the AI industry also was evolving towards deep learning.

9:17So and as the technology was becoming more mature, then you then computer vision related problems became much more important. And can you detect defects on a surface? Can you count number of cars in a parking lot kind of a problem? And then as we got into large language models, then there was further expansion of what you could do in the industrial world itself. So one big move obviously is what we what people call physical AI now, which is sort of contested definition, but the idea that you could do robotics at scale. So part of the work that we've done on automation with traditionally reinforcement learning now could sort of be then a much more formal larger basis with sort of around robotics, right? And you make multiple robots work together or even a robot can work in the environment or not. So the traditional sort of industrially I kind of problems also became on a larger scale. So people started expecting much better answers, a way for humans to communicate. Metaverse sort of becomes becomes important where you could construct these sort of virtual

10:22worlds for training, for problem resolution, right? So in some sense, the the field has moved along with the maturity of the AI technology as well. And then this is a really exciting moment to be in that space as well, I would say, because a lot of a notion will come there. Yeah, I know in, you know, that you have you've run labs both in North America and Japan. And as we're talking about physical AI at this point, it occurs to me as I'm listening to you that, you know, we have a global audience here and people from different parts of the world have different experiences in terms of this move into physical AI and some of these processes. And I would imagine, I don't want to put words in your mouth, but I would imagine your time in Japan. Yeah, there's a lot more robotics out there. And as, as, as we're all sitting here in the United States, there's a little bit less exposure to most people out there in terms of physically on robotics.

11:24I think we're kind of, in my view, we're kind of lagging other parts of the world in that capacity. 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 and how that's moving the notion of physical AI forward in these various contexts you're talking about. And maybe, you know, with the advent of kind of embodied intelligence coming more and more into play with robotics and edge computing and stuff. What your thoughts are around that, you know, with your experiences. I think that's a interesting sort of reminds me of cultural dimension to all of this, right? So what I relearned and I should not make a cultural generalization and often people, but I learned that at least, say, folks in Japan are much more open to sort of robots than say, folks in North America. I think that is something cultural. I can't explain it to why. And that's why even like very many years ago, we were thinking about robots for elderly

12:24support, right? Even if they could not move, at least they could talk and understand and be empathetic, right? So getting the empathy right was a very hard problem. And just the way, you know, a machine would talk to a human, right? So how do you introduce the right sort of the language, the empathy, the warmth that typically humans have for each other? And so in that sense, so for the use cases like elderly care, I would say yes. So the United States was sort of behind other sort of Asian countries, not all the Asian countries, but like the leading countries in Japan, China and so forth. And then when it comes to industrial robotics as well, you're right, the center of gravity is not in North America. Like the center, if you think about large language models, the center of gravity is North America. But in sort of, in sort of robots for the industrial world, the center of gravity is sort of in China today, you've seen sort of the robots do multiple things. And but I think we are catching up. We have some excellent startups that are sort of trying very many new ways for this, the

13:30robots 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. And I would say the commercial use robots would be the first one to have an impact because a lot of industrial processor also already in some sense, robotized because there's task specific, they don't need to be that general. Like more general robotics are needed, more force of human environment, whereas a lot of factories are quite automated already, right? And I think that's what that's where the next sort of battle lines are. And I think in that there is, I mean, it's, it's anyone's came right now, I would say. And so, yeah, some of the underlying math is where, you know, we are better than anyone else, some of the underlying mechanics, maybe other countries are better than us. But yes, but culturally, you are spot on, Chris, right? So there was a broader acceptance and maybe that is still there of robots for day to day interaction in Japan, compared to St. Mathemory.

14:32Does that answer your question? I sort of take it down sort of. It goes great answer, I appreciate it. I hope that you're finding this episode practical and helpful. As the name suggests, we want this podcast to be practical, not just hype. And that's one of the reasons why we partnered with the Midwest AI Summit, which is happening October 15th in Indianapolis. This summit is more than just a bunch of hyped talks. There's really practical things there, like an AI engineering lounge, where you can sit down with experienced AI practitioners, like myself and others, and work through your architecture, your tool questions, your roadmap, your strategy, etc. And that way you can actually leave with a huge value from the event. Again, the event is happening October 15th in Indianapolis, some amazing speakers, some practical advice. Don't miss this event. Make sure you're there October 15th in Indianapolis.

15:33You can check us out at MidwestAISummit.com, and you can use practical AI20 to get 20% off registration. Again, Midwest AI Summit, and you can use practical AI20 to get 20% off registration. So Chitin, I want to circle back to a comment that you made as you were talking about the arc of your career and landing at RAC space. You mentioned something to the effect of going with the technology, but now moving to a place where you could make it more accessible and safe, etc. and operationalize it if I'm understanding you right. Could you help us understand that dynamic a little bit more around the technology? Now it almost seems like with AI everything is feasible, but not everything is easy, right? Or not everything is, you can't operationalize everything, you can't make it accessible.

16:34Could you help us understand your thought process around maybe that transition then from that world of industrial AI to now where when you're thinking a lot about these types of issues around operationalizing, scaling, making accessible these technologies in an actual useful and safe way? So, when you're working with industrial partners, both internal and external, we were playing a team of highly trained researchers, engineers, PhDs and all of that. And so we will take up a customer problem, we'll understand the data, we'll build the model and then we'll work with the customer deployed, right? So and that deployment just not meant sort of throwing the model over the wall, but you sort of work with the customer, you talk to them, you understand exact business pain point, you understand the workflows and then you integrate your model in the workflow. And typically that model was small enough that you could have it in their own environment,

17:35right? So, and this is what we did again and again. But if you now look and if you look back like maybe even five years, most of us were not using AI directly in our day to day world, right? So AI was always mediated through maybe a solution that was developed, deployed and operated by some by some one right? So they took that responsibility. Now with generative AI, AI has become demogatized, right? Everyone has access to AI, every enterprise wants to use AI. And you don't have this team of PhDs and researchers everywhere who can sort of build a model and deploy it. First of all, it's cost prohibitive, right? It's large language models are typically very difficult to build, they're very expensive to build right? So the whole model of how you would bring AI machine learning to an enterprise changes with generative AI. So that's the challenge, right? So at a high level. But what does that mean in practical terms, right? The number of things, right? So you ask a question, right?

18:36I use, for example, some language language model. Now, I could simply ask what's the capital of said always? So for that, I don't need to go to say, chat GPT based my tokens are expensive. If I have a local model, I should ask that question, it will be much more cheaper. And I, and right, so that's number one. Number two is this is a issue raised by Satyadeela and even I think, Jensen, like whenever you as an enterprise say, now you want to say, I want to operationalize AI, AI is accessible, I want my staff and everyone to use AI. Now every time you ask a question, say to a large language model, in some sense, you are sending your data over to them. So your data sovereignty is not guaranteed, right? So the way Satyadeela sort of said, you're losing your alpha, right? That's your IP. So that's another problem. Then an AI in some sense today is jagged. So meaning that there are some tasks for which AI is really good at, right?

19:38So I can really write some very good sort of software program using AI, right? I can code with AI. But if you try to write an email with AI, you realize that it's not all that great, right? It's quite, right? So I get AI emails and I sometimes annoy me because they are sort of very verbose and you know, they're very cliched in some sense, right? So although we thought that writing is the strength of AI, but it is turning out that it's good at it, but not that great at it. Once again, still do better at writing emails, maybe then AI can. So that means in terms of capabilities of where AI is good or bad is jagged, right? So so so from an enterprise and they need to know, right? So this is where I should use AI. And then to my first point earlier, whether I should use a local model because that's more expensive should I use a larger model? And then how do I preserve my alpha, right? But there's only three questions that are today difficult to answer. Now the jaggedness of AI also is in terms of how do you guarantee that it behaves in a

20:42responsible manner, right? So we all heard that anthropic sort of the latest model, fact, bugging phase website, right? So now if you're going to deploy your own model, suppose you say I won't develop my own model within the enterprise on my own data and you deploy it, how do you guarantee that behaves in a regulated in a way that is safe? And also if you're using someone else's AI, like you're using a large LLM, how do you ensure that it behaves within Godrains, right? It behaves in a way that is suitable for your enterprise, right? So the problem of governance and assurance becomes very important. The government, the problem of how do you maximize the value of your AI through economics becomes important. The value of preserving your IP becomes important and the problem having the right architecture to so that you can cater to the multiple needs within the enterprise from programming to writing emails to summarizing to research can be done in sort of a safe, guaranteed

21:47manner so that you can sort of interchange the models if needed, you can set it right. So because the model technology is changing very rapidly. So all of that requires a very systematic way of thinking about your AI architecture. And I think this is the next frontier, this is the next challenge, right? So now how do we go from simply sort of chatting with an AI through an interface to working in the enterprise where there are people, processes in all of the complications? There's a very long-winded answer, but I hope sort of it addressed the question. You're not only was it good, I'd like to actually get you to extend it a little bit by throwing a couple of extra logs on the fire. One of the challenges that we see in industry right now, and it's been evolving over the past, especially over the past year, is where open weights or open source models are available from. And there's so many external considerations that get brought into bear as, you know, while

22:50to your point earlier, that the center of gravity for model development, you know, for L&M development may be still in the US. A lot of those are closed. The number of available open weights models kind of shrunk a little bit with, you know, Meta's kind of going away from that. And Nvidia started stepping up a little bit more because the rest of the commercial industry in the US was reducing. And we're seeing an explosion of capability in terms of new models that are kind of rapidly catching up with very close followers or potentially equal from China. And overlaying all this, you have all these countries have their various exports, concerns, import concerns, what you're allowed to use. And that makes it quite complicated as an ecosystem for companies in various businesses to try to figure out what makes sense for me. You talked about governance. You talked about data sovereignty.

23:52How do you navigate, you know, with some of these big issues that go beyond the technology of AI and the implementation of AI and can affect, you know, management concerns all the way up to the CEO and the board of directors. How do you start, if you're a company now, it's late 2026 and all this is rapidly developed. How do you look at all these and make decisions for strategic interests in your organization going forward because it's quite the quagmire at this point. Yeah, no, and it's quagmire and the changes that are disenged piece as well, right? And so there is no easy answer to that. I just, I don't want to address the question of open model stuff. And I use, and I know you said that, you know, we are somewhat behind as in terms of open model, but I think the beauty of United States is that the right incentive we really step up, right? So now we realize, may the AI commit realises as well that look with open source models, maybe they're from China today.

24:53Obviously Nvidia is doing a great, great job in it, right? There is a lot of merit in that. And from what I know of people I talk to and friends I talk to, I think it's a matter of time before our open source models will be sort of everyone else. That's number one, right? On, on, on number two, like how do enterprises get started? I think they have to sort of fix few things, I would say, right? Fix few things, meaning they need to sort of understand how much of their workload is sensitive today, meaning if you are saying doing an HR query, maybe it's okay to go to NLM. So one thing you sort of figure out what data is something you really want to work around. And there ideally you should sort of think about local models, open weight models in your own environment that you control. So I would say that's one not start you try to try to fix. Don't marry into any model family because they will swap in and out both for commercial reasons, geopolitical reasons, right? So all of that will sort of evolve, change.

25:54So don't marry into any sort of, my marry into an architecture will be of thinking, right? Meaning that these are the workloads that I can push to the X2 to an LLM outside LLM. This is these are workloads that I need to have on prem or in a growing environment. And what are my governs and assurance layer that I want to have? So what are the properties that are important to my enterprise, right? And that I really need to enforce in the way I work with AI. So I think if you have some of those principles pinned down, it becomes an easier way to get started. And I think most, and then I, a year or so ago, I would have said, I pick one problem up and then do it all the way and then pick the second one up, then do it all the way. Because at least six months to a year ago, there was so many pilots and like not as much impact on the bottom line or top line of a corporation. But I think people are learning that lessons are sick, but yeah, but that's the other thing,

26:54I pick one or two sort of problems that are meaningful in terms of impact if you're not started yet and start with the, but for most enterprises, which are already somewhere in the journey that have started, I would say stop thinking models and start thinking architectures, right? Enterprise architectures for AI, right? So that would be sort of my one line. If you would for how to go about it. If that makes sense. I love how Chaitan and this episode is emphasizing the need for operational sovereignty as we move into agentic AI. 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. 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 least agency for the agents operating in your environment and limit the blast radius of those agents. This runs in your own infrastructure.

27:54You have complete sovereign control, whether that's on-prem, air-gapped, or in your cloud VPC and it's available on the AWS and Azure marketplaces. I would encourage you to check us out at predictionguard.com slash practical AI. Again, check us out at predictionguard.com slash practical AI. So Chaitan, you brought us to the point of talking through kind of thinking about architecture and 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. 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, I think maybe there's listeners out there that are thinking, oh, I work for an industry company. I'm not a nation state. I don't need, I don't have a sovereign cloud. What is sovereignty have to do with me?

28:55Could you help clarify that because some people might be thinking or have different views of what that means and kind of bring it down to maybe the commercial or the industry setting. What does sovereignty mean in that sense both in terms of maybe privacy and control? Yeah. So that's an excellent question Daniel. And I think that's exactly the evolution that has happened, right? Folks typically associated sovereignty with sort of a nation state. A nation state is sovereign and it needs to have their own AI stack. And that's how sort of the conversation started. But I think we 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. And that is an IP and given how powerful these AI tools can be.

29:55So that is an IP in terms of your data, your knowledge that you're giving to someone else. And not only that it can be acted upon using AI to build solutions that might impact you as a corporation, right? So the notion of then sovereignty sort of sort of comes down to not just a nation but also an entity, right? Enterprise entity for example. And I'll sort of give me an interesting extension to it. And then that means that how do I protect my own IP, how do I protect my own data, how do I ensure that my model behaves in the way that I want it to behave, my AI behaves in a way, not model sorry, 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, right? It is not someone else's constitution that I have to use but my own sort of constitution as from a sovereign standpoint. And I think that idea will extend further as we go ahead, maybe to an individual as

31:00well, right? So we are quite sort of now, you see the idea of sharing our private information with the band and sort of speak, right? But I think this idea of sovereignty I think will eventually extend to humans as well, to us as well, where we'll say look, how do I protect my own data? Well, when I'm interacting with these language models because people are sharing a lot of private information now, right? They are sort of using them as therapists, as guides, as friends. So this solution of sovereignty will come all the way down and the idea is I am this entity, I am my own interest that are distinct from someone else's interests and I need to protect that's great. And I love, well, I love that definition. I think it gets people thinking in the right direction, but I also loved how you actually kind of made a distinction there when you talked about AI versus model and you brought us to the point of thinking about architecture before. And I'm wondering if you can help us now that you're kind of, you're at RAC space, you're

32:03helping RAC space think through the architecture that needs to be enabled for different enterprises. 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. 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. It's almost like you can look at that AI stack and it can be very overwhelming to understand how to put all the pieces together. What is the right architecture? Could you help us understand maybe how you're helping RAC space enable 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

33:04that's supporting the deployment of AI. So in some sense RAC space today goes from, as you say, from chip to outcome, right? Because we have an appartement with AMD, we have our own data centers. So R stack goes all the way, right? And for our customers, we provide the whole sort 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. 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 RAC space. And I know I completely agree that this can be quite overwhelming, but I think it's not all that bad. Just you 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, right? So if you think about, 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 LLMs, but also

34:09your own model, right? There's a model library. And on top of it is an inference layer. So inference layer is how you get some intelligence out of a model, right? So that is the inference layer. And on top of it, 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. So harness is what in my mind ties 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. It is the coding harness, right, around the underlying machine learning model that enables you to do something very useful. 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. And harness simply thinks, think through of harness as sort of the machinery that sort

35:13of specifies the logic for the underlying AI model to use and setting a set of toolsets that it could use, right, through MCP or whatever. And then in any enterprise, they will be multiple harnesses, right? So you will have a coding harness. You might have a harness for your agents for HR. You might have a harness for agents for blah, blah, blah. And so ideally, you should think of building an orchestration layer to manage multiple harnesses. And on top of it, you can then think of the consumption layer. And around the whole thing should be wrapped, I would say, with the governance and assurance planes. So if you think of this, this is not so different from how we thought about other sort of stacks in the past. It's 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. And everything can then be done through sort of the way we have done that in the past,

36:13through APIs, the specifications, who says what to pull them, how do they interact. And that's how these multiple layers can interact. 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. No, I think that's quite an elegant way of describing how architectural components fit together. I really found myself gravitating to the way that you were explaining it. And in my head, as you were doing that, I had a question which I think you started, you've already started to answer, but I wanted to extend it a little bit. And that is from a customer standpoint, as they are looking at the products and services that, whatever their business is, that they're offering their own customers. And there's some sense of stability to achieve that value. They need the product or service to be reliable to their own customers over time.

37:19And yet on the back end, your customer who is providing that service to them 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. And I was in my question, which maybe I have a glimpse of, was how do you manage the tumultuousness take to end 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. I'm guessing that that's somehow being managed through the orchestration layer in terms of how you're doing that. And you need to build evils for your own workloads. I think one thing people often gravitate towards is, these are benchmark, this model is doing something else, it's better than the other in benchmark. But those benchmarks are only guidelines in some sense.

38:19They don't represent your workloads. And as we said earlier, the boundary of AR is jagged, right? So although a certain model might do very well on a benchmark, it doesn't necessarily translate into that it will do very well for your workloads as well. So I would say having your own evils is very important. And once you make that eval, then in some sense you can have very consistent experience for your customers. So you can swap in models in an out based on your evaluation results, right? So you should say, OK, I have the same evils from my previous model to the new model, the cheaper, lighter, whatever, or it's safe, I will use this. So I think you need to build an eval as part of the orchestration framework. And that can then help you decide which sort of harness to go to, sorry, sorry, daddy. No, I think that I was just going to say, I think that ties into what you were 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, et cetera, is a key piece of that.

39:24And 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 piece together. And also thinking about the, you mentioned the term outcome, you know, what outcome are you after? And are you really testing for that outcome? I think that's a key piece of it. I wonder. Yeah, just to add to that, right? If you remember, when we were doing sort of these machine learning models for industrial use cases, you had this golden data set. So before you could deploy, the customer will say, prove your model works on this golden data set, right? So I think it's the same, like now we call it eval, it 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 and so on. So I'll be sure to interrupt you, but I'm sorry. That's great. That's great. Yeah, I love that. And I guess kind of as we get closer to the end here, I want to give you a chance to,

40:28you know, you're sitting in this chief AI officer role, you've kind of navigated this career, are coming to back space. 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 and the things that you're encouraging RAC space to think about as we're going into the rest of this year and next year. What's on your mind as that chief AI officer for RAC space? What's kind of, yeah, what sorts of challenges are at kind of the top of your mind as you're laying down to sleep at night or coming to the end of the day? What's at the top of your mind that needs to be addressed within RAC 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?

41:29I would say a couple of things. And I know 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 we partner with for what layer? What are the different types of customers? Like what makes sense? No, there's no one's answer that fits all. 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. The other thing I think about and I think in this case, in the sense RAC 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 org. Meaning what we're saying now is if you can build it internally and use it internally, then you can go to the next turn. So if I have an AI solution that I have proven on my own workloads, then I can have confidence and go to my customer and say, you could use this.

42:32So what I'm trying to do is from an internal standpoint, build a discipline around it. So it's not that the central project is less valuable and it's like a throwaway project. 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 underserved today and especially in a sovereign sort of environment. So how do I bring solveninity, self-solveninity 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 them. 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 those are some things that sort of stuff. Yeah, I think that's great. I love that perspective, of course, from where you're sitting.

43:32You have a broad view of what challenges you're trying to address and what's important. So always eager to get insight there. This has been a great conversation, Chetan. I would very much encourage our listeners to check out what RAC space is doing. We'll include some links in our show notes. Is there anywhere in particular, Chetan, that obviously people can go to the website, see what you're doing with AI? Did anything to highlight in terms of what RAC space is doing, kind of a jumping off point for people or anything to highlight as we close out here in terms of what's currently available from RAC space and what people can explore? So I would say that the questions that RAC is trying to answer is where the industry is going towards. So you should think about those as well. How do you maintain, how do you build AI that is sovereign, that is safe, that meets your outcomes and all of that, right? So those are things that everyone should think about. And those are the kind of solutions that RAC space is bringing to market.

44:35So please visit us, send me an email or reach out to me on LinkedIn if you have specific sort of questions as to what RAC space is doing. But I would encourage everyone, if this is RAC space or not, to think systematically about these problems, because that's how you can ensure that AI succeed in your own environment. 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 RAC space. Thank you so much for joining. Thank you, Chris and Daniel, fantastic conversation. Thank you for hosting. Love it. All right, that's our show for this week. If you haven't checked out our website, head to practicalai.fm and be sure to connect with us on LinkedIn, X or Blue Sky. You'll see us posting insights related to the latest AI developments, and we would love for you to join the conversation.

45:36Thanks to our partner Prediction Guard for providing operational support for the show. Check them out at PredictionGuard.com. Also, thanks to Breakmaster Syllinder for the Beats and you for listening. That's all for now. But you'll hear from us again next week.