 Welcome back everyone to theCUBE's live coverage here in Las Vegas where SAS innovate 2024. I'm John Furrier, your host of theCUBE with my co-host Dave Vellante, head of CUBE Research. Our next guest is John Boyd, Vice President Solutions Product Management at SAS. He's got the keys to the kingdom on the product mission and he's got the solutions. Great to meet you. Thanks for coming on theCUBE. Well thanks for having me. Great to meet you. So you get the product management, you get the solutions. So right now we're in a mode where people are experimenting with AI, rolling out production workloads here at the keynote, Brian Harris and the team showed a customer with a production workload with an AI stack that only had really Amazon as an external company and then a lot of SAS and then customer data and stuff and Genive AI and outcome, not a lot of other things. So it's a real enterprise workload, big time. Correct, correct. And that's what we're seeing like the Viya platform is great for that sort of model governance, right? To get the design time of that model, what you want to do to get the business outcome. And then from a solutions perspective, we're taking a lot of that investment and trying to figure out how do we create that runtime environment to adopt that model within our solutions and whether it's coming from what Udo's organization is going to do with his announcement or it's something a customer is going to build on their own et cetera and build a sort of business workflow around that to run the people's businesses. So you are involved in the billion dollar investment. You're not you, your team and SAS, you work for SAS. You guys announced a billion dollars to be invested into the SAS teams to build AI. Can you share some of the results? What are we seeing? I mean, obviously we saw, we were at Explore last year only six months ago. Now we're seeing fruit on the tree there. Can you share information, data and color commentary? Sure, I mean, there's a lot that's gone in with that. So the first thing was getting the solutions on Viya 4 as you heard from Brian this morning. So that was a big investment to get in there. And now you're seeing more and more innovation coming out. Last year we released SAS Health and we released audiences for CI 360. In the upcoming months we're going to have energy forecasting we released in SAS Clinical Acceleration Repository. And then towards the back half of the year we're also looking at SAS Marketing Decisioning. So the intelligent decision that you would have saw at the opening session today and how that's going to evolve for marketers. And then obviously all the GNI work that's happening along with UDOS work as well. Can you talk about the domain specific IP that you have there specifically? Yes. Let's unpack that and understand where SAS has unique IP that is contributing to these solutions. Yes, well it's an interesting thing, right? If you sort of look at the market out there, you see a lot of people that are in the sort of data and analytics space are people that are vendors in specific industries that are leveraging data and AI. And I think that's where our solutions make us unique in SAS because we bring both to the table. And why that's sort of an important is because when customers out there looking to simplify their ecosystem and reduce vendor costs, et cetera, we're one of the few options that can provide that both play for a build versus buy for them. And how we've done is have that domain specific IP that's gone deep within that industry so we're solving for that industry problem. But as we've done that, customers have trusted us, say, hey, what about this other business problem that isn't industry that's more horizontal related? So it's sort of created what I call woven product architecture between the vertical industry solutions we're doing and the horizontal business processes we're also getting into as well. What are the salient aspects of that industry solution? There's data, where's the data come from? There's the industry specific IP. What other components are there to make that package consumable? Yeah, sure. So you have the data and one of the things I'm pushing for in our solutions is really reducing our time to value. So having that data in a format for our product but looking at industry standards out there to hydrate that so it makes it an easier lift for the customer and not have that ETL burden that a lot of people have. And then also once you have that information, all the power of buy analytics come into play. All the things you see from the platform, the capabilities, being able to deploy these models into the environment and in conjunction with the business workflow is what solutions is about. And you talked about the horizontal business process. How do you think about or how are your customers thinking about changing their processes as a result of AI and the sort of gen AI awakening? Are you seeing like a lot of movement there yet or expecting to see movement and how would that affect your horizontal strategy? Yeah, so I think there's a lot of interest. I think people are still trying to figure out how to get value, right? I mean, we can talk in theory about all the things you can do in gen or AI but when you're running a business, they want to know what's the bottom line and how much money is this going to save me or how much money is going to make me. So from that perspective, we are looking at gen or TVI from that lens, not of just having gen or TVI for that purpose to where we can put it in the most critical aspects of the business process to help them to be more profitable. So right now it's very tactical, right? I mean, while we can opine about how things are going to change down in the future, today it's like show me the money. Yes, correct. I mean, in my philosophy of a gen or TVI right now, it's like you're seeing a lot of use cases in terms of querying data. That's your first step, right? And then the API, calling APIs based on that API orchestration to execute a business process. And I'm excited to see Brian Harris's hot take and my hot take for sort of gen or TVI is everyone's talking about prompts, but I think we're going to get to a point where it's promptless gen or TVI where it's working in the background and thinking about looking at the data and saying, hey, this is what you should be doing. Here are my recommendations. Which of course of actions do you want to take? I'm trying to, from a product perspective, get ahead of the puck because I think that's where we're going to go, but we're setting up all the things to do that. I'm really glad you brought that up because I was saying to Brian, even though everyone's going to Gaga, including myself over the models, I said the saved prompts is a really indicator of where this is going. I mean, that is essentially the beginning of where you're going where, okay, when you get some of that work, you save it and you use it. And then there'll be more of that. And what's going to happen is AI will figure out quickly what the real prompt is trying to ask, based on the market basket of prompts it went through, almost like Dr. Goodwright is talking about compiler versus interpreter. You're kind of compiling all the LLM prompts and saying, hey, this is the ones that are going to be runtime ready. So LLM or foundation model ready AI is going to come. That's like a compiler conversation or an interpreter. I mean, it makes things better. What's your reaction to that? Yeah, yeah, I mean, really everything about this is just learning on top of each other. And that's what we're all about with this, right? Just building on top of it. And if you can have that where it's sort of a promptless AI and it's sitting in your business workflow where it's clearly telling you when something's successful and when something's not, you can learn from that and prove the recommendations going forward and thus improve your business operations exponentially as you move forward. Talk about the couple variables we want to kick your thoughts on. So all the buzz has been, oh, training and inference. Okay, you got to train models. Sometimes you don't because it's already trained. It's raw data. It's pure model. Inference gives the impression it's something's happening there. We're getting some logic reasoning and they say that going on. Then you got the prompt response. So assume promptless. I agree with that. Assume that happens. If that happens, then you're going to have a lot of reasoning going on and then a lot of reinforced learning happening on the sides. Okay, great. Now, question, okay. If I'm going to have all this reasoning, how do I change my application knowing that it's scoped end to end to manage the best reasoning, best workload to performance, hardware? How do I match what I need at any given time to be elastic like? So, okay, I'm going to do a prompt that's promptless. That's a lot of IO in my mind. Then you say, well, I'm going to need to think about something. I might shift that workload prompt to a different cluster or a different system. This is going to require orchestrations and require scheduling. It sounds like an operating system to me. What's your reaction to that? I mean, it's an interesting thought because I mean it is because you're basically trying to mimic human behavior and what we're thinking. So it is like an operation. It's basically a human operating system for mimicking one, right? That's what we're getting to. Now, that's what excited about me with the solutions and the possibilities here because one of the things that we have is solutions across these horizontals across multiple disciplines. So to get the best answer, you're going to want something that connects something on like, for example, looking at your sales and how does that impact my cashflow downstream? And having an interconnected horizontal suite of solutions on like that is going to give our customers the best decision making advantage possible because they're going to know the bottom line and how it ends. And you're going to need data and you need performance. If you don't have the data, there's a lot of blind spots. Dave calls it Swiss cheese. There's holes in the, what could be looked at. And that's really where the hallucinations come in that everyone's seeing on the mainstream. So the question is, how do you make all that data available? I mean, one, you got to just, it's just naturally available. It's like, hey, I'm addressable. How do you guys see that with SaaS? I mean, how do you guys think about, do you make it available? Is it in a self-contained system? Yeah, I mean, from a data perspective, again, going back to reducing the time to value and hydrating with the big systems out there, where all this data is, where are we running business on? We're running on ERPs, right? So connecting to those ERPs and getting that business information into our solutions, hydrating it. So we're taking their transactional information and putting into an enterprise analytic platform and gaining insights across our entire business. What are you seeing in terms of adoption? And this is GNAI specific data, but in the recent survey, about 1,800 IT decision makers with our partner ETR. It was astounding, about 18% of the respondents said, they're not doing GNAI. And so I went out and poking around a little bit and trying to find out why. And people said, well, first of all, it's moving too fast. It's really hard to predict and we don't trust it. We're in a regulated industry, for example. And so we're going to sit back, like the bubblegum shrimp. We're going to let the hurricane take, you know, wash away and then we're going to come in and dump it down on our beds. Forrest Gump reference. Or what? It's a Forrest Gump reference, nice. Yeah, yeah, yeah, exactly. So, right? You know what I'm talking about? That scene where they just sort of wait for everybody to get washed out. They end up monopolizing the shrimp business and all the boats were parked. Let everybody else make the mistakes. Okay, so does that, does that, those stats hold water? The other thing we see is that over 40% of the customers say they're stealing from other budgets to fund AI. It's not like the macro is just, everybody's writing checks. Right, there's a balance there with, you know, the two-year T-bill at 5%. Okay, are you seeing that in, does that resonate with you or align with what you're seeing? Are there specific industries that are more aggressive or less aggressive because they're maybe not as regulated? What are you seeing? Yeah, I think it does resonate. I think it's probably people going through the learning curve at Gen AI. So, everyone got excited, first of all, right when it sets us out and they saw all the hallucinations up front, right? And we've heard this from SAS before. It's like really, it's a conversational interface at this point in time. You still need the other AI and business process to give you what the real answer is behind it, right? So, when they realize that level of work is still needed to it, then they're starting to shy away a little bit, right? Especially in regulated industries and things like that. I think that's where SAS solutions come into play because we can do that for them, alleviate the burden and that lift for them so that they can be turnkey solutions that come in and get that information and the results that they want without having to invest all the intellectual manpower internally within their own company to go develop and build it themselves. John, talk about the strategy of SAS, vis-a-vis other alternatives. The customers that you have, they deal with everyone else too. You got a lot of ecosystem partners, I saw the big logos up on the stage on the screen. Some people want to have a horizontal play and some people have a vertical play depending on what you have. Some have built-in open data platforms, some want to have industry-specific solutions. So you have a real diverse approach to industries. And then complicate that with the third dimension, which is the domain-specific models and intellectual properties. So okay, I have all this data very specific to my domain, but I want horizontally scalable everything else. I want to integrate APIs with a competitor or potentially a partner. All this is going on. So the customers have to deal with multi-vendor, multi-environment, now multimodal models. What makes SAS differentiated in that market because that's the current situation? Yeah, correct. And I think the differentiation is the optionality that we provide to say, you can buy whatever you want, you can build whatever you want, and it doesn't have to be ours, right? We are not afraid to say, we're boxing people out. We're going to stand on the value of our solutions as they stand. We're going to create an open ecosystems where other people can integrate and they can choose where they want to plug in things, where they want to use our things, et cetera, and we will win them on the value that we want. So you're saying to customers, you have optionality, we're going to give you optionality, be a choice. You can, we think we've got a good system. And where do you say your winning strategy is when you say, but that being said, we're better here. Fill in the blank. We're better because of blank. Yeah, we'll give you optionality, but I'll tell you, having an enterprise analytic solution that goes across your business workflow is going to give you insights that none of these niche players can give you because they're only looking at their very specific parts of the business process. And two, we're going to come in a much more value because we're going to give you one vendor, one price point to work with for all these solutions going across, versus having to negotiate and manage all those ecosystems, different infrastructure and internal IT staff that you have to manage all that. So your key is across multiple workflows, across multiple environments, and potentially companies. Yes. Go ahead, please. No, I mean, giving all the optionality and flexibility, meeting our customers where they're at and where they need us. How do I consume these industry specific models? I can subscribe to them. Is it a consumption model? Is it, how do you price that? How do I get started and what is it going to cost me? Yeah. I mean, even there, I think we're going to have optionality. Some people might want those models just by itself and not even use them with the solutions. Go for it, right? But all those models coming out, I'm going to build a way to integrate those into the solutions so that you can pick and choose. So think about having a solution that let's take a credit rating, right? It's taking a risk of default and putting a model on there and having something in our solution that says here, here's a graph that looks like if you had subscribed this model, here's the information and outcomes you would get out of that. But for you to have this, you need to subscribe to this model and purchase it over here. So again, even within the solution itself, giving the optionality to our customers of whether they want that specific model, whether they want to build their own model and plug it in and get that functionality and solution with their own development or getting it somewhere else. John, it'd be great to have you on theCUBE because you're touching product management and solutions which the cloud and AI is very Lego block like. But now you got N10 put that all together. A lot of opportunities. I want to ask you about what's next to future of SaaS because you talk about time to value, okay? But there's also the, you got the design time versus run time and then you got the general AI piece behind it. But what is the between design time and run time and what's the distinction between those two factors? Yeah, sure. So the design time is really getting the models trained up in terms of getting them functional and getting the outcome that you want. That takes a lot of horsepower. That's perfect for buying what that provides there. But once you have that, you don't need that same horsepower to execute it on a transactional basis to get the outcomes that you want. And that's what I'm talking about run time. Solutions are really run time in nature. They are delivering business outcomes once the models are already produced and you have the AI capabilities that you need. So embedding in with that workflow, let's reduce that cost in terms of the infrastructure that you need, et cetera. So it lets us to go downstream in the market and also pass down some of that savings to our customers. Not to bring up run time again, but when you think about generative AI, it's generating. It's not static world that's pre-programmed. It's generating at run time. Again, Dave, back to the operating system. That run time, things got to happen. They got to load. That's going to require new software. That's the key to success, right? Yeah, correct. And there's a lot of pieces that are happening that you can see and we're looking at all of it with SAS, right? We're trying to figure out design runtime, time to value, generative AI, taking models, embedding in the solutions. And that's my role here, putting all these pieces together and delivering. Final question real quick. What's the vibe like inside SAS right now? We've got a billion dollars, saying, hey, that's a North Star. I'm sure the founders like, okay, we're all in. That's like saying, we're burning, we're going to transform right now. We heard that last event we were at. What's the vibe like inside the folks in SAS right now in Congress? I mean, I think they're excited. I mean, look around here, right? And all the things that we're cranking out. And I mean, I thought the opening session was amazing. You see the passion and people up there demoing workbench and things like that, et cetera. And the capabilities that they're able to do in such a short time, right? We were only here, I think, back in October, right? A few months ago. All the things we're announcing now that weren't announced even a few months ago, here we are. It's nice to see fruit on the tree, right? Coming off that harvest. Not yet harvest, it's still blooming. Yes, yes. Too many metaphors. Scissor and Stake, fruit on the tree. Next year, when we're here talking next, what are we going to be talking about? What do you anticipate is going to be the top story and what you'll be talking about? Yeah, I think you're going to talk, you're going to be seeing a lot more solutions, again, looking at the design time versus run time. So us going further down market with that factory, you're going to see us getting more traction with the models that we're going to release. And having those embedded within our solutions. And then also keep on reducing that time to value, looking at all those industry standards out there, so that we can hydrate that and get our customers up and running as fast as possible. I mean, it's a great opportunity to accelerate the application process, the value, utility, government, citizens, enterprise customers. Yes. Thanks for coming on. John Boyd, Vice President Solutions, Product Management, SAS. I'm John Furrier with theCUBE. But Dave Vellante, my co-host, will be right back with more coverage. theCUBE, the leader in high tech coverage.