 In the context of this new product, we're really talking about people who are both training and fine-tuning models, and these are things which are really impactful for customers, because just to take an example of one of our early adopters for the new product release, it's a content company, a Cora equivalent. For example, if you look at them being able to train a large language model, which is driving their end-user business metrics, such as user engagement, which is the most important thing for them, just being able to use their custom data and leverage it to improve their key business metrics and get these initiatives to market as quickly as possible with as few people as possible. Hi, this is your host, Subnal Bhartiya, and welcome to Let's Talk About AI. Today, we have with us once again, Adit Madan, Director of Product Management at Alaxio. Adit, it's great to have you on the show. Hi, Subnal, nice to be here. The theme of today's discussion is the new announcement that you folks are making today, which is new data platform for AI. But before we go deeper into this announcement, let's just refresh memories of our viewers. What is Alaxio all about? Subnal, Alaxio is a software-defined solution in the data infrastructure stack, uniquely positioned at the intersection of compute and storage. And what we do at this layer is we bring in efficiency and ease of access to data. Our existing customers are using Alaxio to scale out not just the capacity of the data platform, but also adopt a hybrid and multi-cloud strategy. In today's modern world, we kind of live in like AI driven world, right? Whatever we do, even though it's a small ring watch, it has machine learning capability. My cameras is checking my face, as you can see there. What does it mean in today's world? How far we have come. And then I would also love to talk a bit about generative AI, but let's talk about model learning, machine learning, all these things from the perspective of the announcement that you are making today. So generative AI shares a lot of common underpinnings with what we were calling deep learning before. So generative AI is actually based on a specific kind of deep learning technique. And there's a new model, kind of model, which is called the transformer architecture, which is different from when we were previously talking about deep learning. We typically talk about things which are called convolutional neural networks, RNNs very common on the computer vision side of things. So the building blocks and the characteristics of the models, whether it's the deep learning kind of CNN, whether it's the deep learning kind of transformer networks which power the generative AI today, both kinds of models have similar demands from a system like Luxio. And then today I'm excited to share that our early adopters of the new product are actually training large language models on top of Luxio, which is based on the new transformer architecture and also the different kinds of CNNs which are common for computer vision and in the autonomous driving industry. There was a time when we used to talk about, hey, every company should have a cloud strategy, every company should have software strategy, otherwise you will not survive. Do you think it is apt to say that today every company should have some form of AI strategy as well? Or you feel that no AI, and when we talk about AI, of course AI is a broad setting. It's hard to actually define what you mean by AI. Or you feel that no, there are only certain industry that need, or you feel like no, just the way we all have to move from software to cloud, AI should also be the core of your IT strategy. No, I do agree with that statement that AI does need to be at the core of the strategy for everyone at this point, given the impact that it's having, and especially with the advancements in AI recently, the impact to productivity all the way from engineers who are coding to your support and sales team who's trying to debug what's happening with prospects. The impact, and it's so profound, it's such a big change in productivity that AI in general needs to be part of the core strategy for every company out there right now. I do believe that. What is specific niche, especially for enterprise customers that you are addressing, which you felt was lacking before this announcement? Yeah, so in the context of this new product, we're really talking about people who are both training and fine-tuning models, and these are things which are really impactful for customers because just to take an example of one of our early adopters for the new product release, it's a content company, a Cora equivalent. And for example, if you look at them, them being able to train a large language model which is driving their end-user business metrics such as user engagement, which is the most important thing for them, just being able to use their custom data and leverage it to improve their key business metrics and get these initiatives to market as quickly as possible with as few people as possible. These are all of the things which are kind of driving the need for Luxia and at the same time are also very impactful for many, many companies in this space these days. You and I have talked about the cultural aspect earlier as well. When it's come to this announcement, what kind of teams are you targeting within organizations? Yeah, so when we've spoken before about Luxia and Luxia Enterprise, we've been talking about the data and data infra and platform folks within the large enterprises. So these would be people who serve the data analysts within the organization. Hundreds and even in some of our cases, thousands of data analysts who are using the platform itself. Now with this new product, Luxia Enterprise AI, we are targeting two segments and two personas in a way. One is within the large enterprises which are familiar customers, we are expanding to the sister AI platform teams within those organizations. So these are the folks who are serving the many data scientists within those organizations. And at the same time, we are also entering a new market, a new market which tends to be a lot smaller in terms of the sizes of the enterprises and also the size of the real data real estate that they have. And then this is typical, for example, I was mentioning autonomous driving before. So within those organizations, there tend to be specialized AI and training teams. So that's kind of the two personas, but platform folks who are using a do-it-yourself platform are the buyers of Luxia. There is a kind of gap in supply and demand of these folks. Market is also changing. Talk a bit about how under stress organizations are to also be able to get on board with all these AI activities happening and how LXU in a way make it easier, simpler for it so that organizations can deal with this either with a smaller team or free time of their teams to focus on other things that add a lot of business value to their organization. I feel like right now we all know that given the importance as we talked about of AI and an AI strategy within the enterprise, that the knowledge and there's a lot of learning aspect to it on how to appropriately make use of the technology and the infrastructure that is available right now. So, and that's why there's a lot of AI consultancy companies flourishing as we speak. Now, the way that we are helping with this journey is that as part of this learning journey for companies, there is a phase of experimentation, which is needed, experimentation and fast iteration in trying out things like trying out, training new models and seeing what the impact is going to be without really a deep understanding or without the certainty that people might have had in a more mature market. So what we help in this case is by bringing efficiency to the infrastructure which is empowering these applications and by allowing people to do more with less, they are able to really to achieve this fast iteration which is of the utmost importance right now. Now, can we talk about the architectural or technology underpinning of Eluxio Enterprise AI if you can just walk us through some of the cool components? What is this comprised of? In Eluxio Enterprise AI, we have a brand new system architecture. We are calling it a distributed object store repository architecture. It is something which is very different from what we had before. It's uniquely designed, purposefully designed for the kinds of workloads that we were talking about, serving the needs of specialized compute hardware. This architecture not just scales out pretty much to the limits of our data center to provide, to solve the needs of the specialized compute hardware, but it also does not require a dedicated hardware to run on. So it's available, something available on the specialized compute hardware itself and also exposes interfaces like a POSIX API and a REST API which are required by different Python and PyTorch applications, commonly used compute engines in this space. Talk a bit about the state of market where when you, of course, you're coming with enterprise solutions which does mean that there are a lot of takers, but... Is it where we are? Are we at a phase where you feel there's a lot of education needed? To show the value of these? Or you're like, no, this is more or less like... We have moved to a phase where companies are demanding solutions. We don't have to invest any time in education or awareness about, hey, this can help you. Where are we? We are definitely at a stage in which we don't need to educate people about the need for the technology in the space that we are talking about today and alongside enterprise AI. Then the need for data access is actually a much more well-established and well-understood problem. So as we started, as we were incubating this and we were making this proposition to early adopters of the solution, we were very pleasantly encouraged by the response that we got and how easily understood this problem was. Before we wrap this up, can you also talk about how does today's announcement fit into all the offerings that you folks have? And if you can also talk a bit about the pricing model for this enterprise AI solution. So before today, we had a single enterprise commercial offering. We were calling it a Luxio Enterprise. And going forward, we are forking that off into what was called a Luxio Enterprise is now called a Luxio Enterprise data. And we are introducing a brand new product which is based on the new architecture that I mentioned and it's specifically purposefully designed for the AIM and workflows that we just spoke about. This platform is available in a subscription model. It's based on the amount of storage and concurrency that is demanded from the Luxio platform itself. Adit, thank you so much for taking time out today and of course, discuss to this announcement also. I really appreciate your insight, your take on the larger AI, generative AI space. Thank you for all those insights. Thanks for the pragmatic approach there. And I would love to chat with you again. Thank you. Thank you so much. I enjoyed our conversation as always.