Accelerate AI-Powered Retail Experiences & Enterprise Workflows with IntelliX Prime™ Pre-Built AI Accelerators on IBM Power S1112 & S1122 | IBM Channel Partner Webinar - ILOG SA
Hello, everyone. Thank you so much for joining IBM webinar. Before we get started, I'd like to go over some housekeeping items. At the bottom of your screen, you should see a Q&A box where you can start submitting your questions for our presenters today. Later today, you will receive a link to the on-demand recording and a link to the slide deck. So without further ado, I will pass it over to David. So what Fiducia does is it expands Rob, you might want to. Can you guys hear us? Now, yes. We can hear you, Rob, not David, though. Can you people hear me now? Yes. Yes, go ahead. Okay. All right. Not sure what's happened there, but I want to welcome Fiducia to the IBM Power ecosystem. We're excited to have them as a partner. IBM Power has been known historically as an enterprise solution with high reliability, availability, and serviceability. But without an ecosystem, we don't have a play in the market. So what Fiducia and other ISVs are bringing to the table are giving us the ability to scale beyond the traditional database enterprises. Fiducia brings open-weight models to power, allowing us to run in a secure and sovereign environment. So you can run with Fiduci in the cloud, or you can runs on power, bringing AI next to the data. It's a very strong solution, and we're looking forward to taking this to the marketplace. And I'll turn that over to you, Rob. Thanks, David. So what makes power different and why would power be great for this solution is what I'm going to try to cover here. Before I do, I always like to answer two questions that I normally get on power, and that is, you know, is power got a roadmap? Is power going forward? The answer is yes. Right now we're on Power 11. We continue to strive towards the future with our chiplet architecture and continue to drive our chip forward. So I always like to show this because people always ask and wonder, you know, what's next? What's next. The other thing that's unique, especially in accelerated workloads that I like to point out that's different about Power is our virtualization. So most virtualizations like VMware, Parallels, or Microsoft's virtualization is what we call full virtualization. Power's virtualisation is actually para-virtualization. So we have less than a 3% overhead, and it's closer to the hardware. This is important in these types of workloads primarily because it's all about getting as much out of the hardware as you possibly can. When you invest in these equipment and you're spending as much money as we're talking about in GPUs, it's all about getting the hardware as much out of the hardware you can. I left these slides in here. I won't talk to them, but it's for anyone who's interested because people always want to know, like, okay, how does this compare to that from PowerVM versus VMware? So I left These slides in Here for that. The other big difference that we have that I always like bringing up because it's a huge difference is our OMI memory. This is absolutely perfect for accelerated workloads like fiduciary. It's basically what we did is we created a new memory standard where we took the DDR chip and put it onto the memory. This is why in Power 10, you were able to go from DDR4 to DDR5. Everyone else in the industry, the DDR protocol is ran on the actual CPU. By separating this out, this also allowed us to create a new protocol so that them can talk to the CPU with a new protocol, reducing error rates and allowing us to get actually more throughput. In the systems that we're going to be presenting, I'm a picture guy, so everybody always likes to see it. this is actually the DIMM that's in there. One other unique thing about this memory DIM is not only does the DDR live, the DDR controller live in this GPF ASIC here, but we're also mirroring the memory modules on the Dimm itself. So if you lose a memory module, it still continues to operate. And we also have a spare module on there as well. So going into power with the Fiducia AI, The two servers that we're kind of looking at for this solution right now, there's always room to grow, but the ones that we'll kind of focus on is the S1112 and the S1122. Obviously, we have other servers in the brand. I always like to show this so everyone knows where we're at and what's possible, right? One of the nice things about the S11012 is its packaging. So it is a 2U half-width rack module. You can get two in each sled, or you can get a desktop model. Kind of breaking this out, showing the insides here. We have four NVMe drives in the front. You can see our memory, and then we have four PCI slots in the back. this is the desktop layout and then this is the diagram of how the board is laid out this is a nice packaging for this if you're looking to do accelerated workloads on site or you have a workload on site where you want to run something and you don't have to have a whole rack it could be a desktop model and even if you wanted to go on a rack the the packaging just makes it perfect for this also makes it perfect for growth if you have more than uh a need for more than one machine um like i said on this on the um sled we actually allow two two per sled here if if we need to take and that doesn't fit the bill and you need something bigger for growth we're also looking at the L1122. This allows us to have more memory. It doesn't use our DCM modules, so we can get more cores in here. We also have more IO options, but the bigger thing here is this will also allow us to take advantage of Spire as well. I left the layout here for those who are interested in it we also have internal storage capabilities there as well this is an example config that we left for the the 1122 for those who are interested in what this would look like we also had a example config for the 1112 as well we're targeting and starting with the 1112 first because of the packaging, and the way that everything fits. But like I said, the 1122 is also an option. So we left two example configs for anyone who would be interested in what this would look like. With that, I'll hand it over. Awesome. Rob, thank you very much. Can you all hear me there? Can you confirm, please? Awesome. Rob, thanks for the amazing overview, talking through the internals in terms of how the system is designed. Today, we'll actually go through what Sovereign AI is, how it's deployed on S1112, and with a specific use case, a retail use case. What we're discussing right now is applicable across industries. We'll do some examples. I'll be sharing my screen shortly. We'll be doing demos. And folks who are interested, have your phones ready. You can actually scan QR codes and then see the activations live on your phone. And I'll walk you through step by step. But before we get started, I am Ganesh Harinath, and it's a privilege and pleasure to be here. David and Rob, thank you very much for a warm welcome to the Power Group. We've been working with David, Rob, and the power team for the last few weeks, and the demo is a reflection of what we have live working on S112, and it's magical. So with that, and I have Jonathan Taylor, who is the chief business officer who's with me. Maybe, JT, you might want to say a word or two, and then I'll kick it off. We can't hear you. Okay. I'll continue. JT will come back, Jonathan Taylor will come back towards the end to share how easily we can actually get to a point, working with our customer to get this live all the way from technology to procurement perspective. Let me go in full slide screen mode, much easier for me to switch between slides and demos. But do not hesitate. Feel free to actually show the questions. Sorry, write the questions and then we'll be very happy to take the questions a team stand by. They're going to read out or they're going to text me and then I'll be happy to reply to those questions as we continue to talk. Now, so the story that I wanted to share today, the whole world is actually talking about AI And it's very natural for people to start to think about open AI and then other frontier models where we start to experiment. Experimentation is great. But today, in the next 30 minutes, we'll actually see how real it is. We can actually make all these things happen on the power boxes. We'll start with the smallest. We'll be quick. We'll show results working with the customers and customer use cases and then so on. What's the state of affairs today? You see, enterprises have control on data, and the data as context is passed to frontier models and the prompts, and this process intelligence is as well passed to the frontier models. There's a lot of risk, and costs are going up as well. So how do we address that, and then how are enterprises starting to take a look at it? Many of them, they're in their very, very early stages, and it's a perfect time for us to actually start to socialize Sovereign and start to socialize sovereign on IBM Powerbox, and most importantly, in days. What we're discussing can fairly be deployed and accomplished in days, giving enterprise an accelerated path to get their AI workflows. It's not AI. AI workflows have high value, primarily because these workflows get integrated to their existing ecosystem, which we will actually reflect in a few minutes. So the slide that you're looking at is a reflection of Fiducia's journey at this point. Our technology has been deployed on IBM Cloud, and we have used Watson X. We have made a lot of progress. Then PowerVS, it's deployed in PowerVs. It supports x86, but most important story that we would like to share today is how we can actually bring not just AI, but AI workflows onto power. And I'm saying S1112, the smallest box. So advantage of being sovereign is having complete control on intelligence, complete control of data. Be more compliant, right? And then have complete insight into our AI models and then so on. So one of the primary drivers is open-weight models. And we'll see some of that working on the Power 11 box shortly. So as I was mentioning, we can't walk up to an enterprise and then say, okay, we have the smart intelligence system, and then you have to replace all your existing systems or make big changes. The way we all can actually bring in AI into the ecosystem is through AI workflows. With Fiducia and with IntelliX Prime, we empower the enterprises to very quickly create these AI workflows and integrate it with the existing system. That's what is very important. While we have control on intelligence, we help integrate these AI workflows into their existing system, and we'll see how. And the benefits, we all are very, very well aware of the benefits of having conversation with your customers, with all these different interactions and then so on. But at a very, very high level, intelligence is amazing, but intelligence has to go hand in hand with data. That is the customer's data. And we make that part very simple as well as part of this journey. And at the end of the day, you actually see workflows being created, and it's inevitable. Every enterprise, small or big, they have to start to adopt AI workflows. And what we do as a team is to make sure that this is operationalized seamlessly and in days and in dates. Now, the undercurrent, not many people actually pay attention to. uh intelligence translates to accelerators gpus and then so on but what we are talking about is ai workflows it's a combination of bringing in data cleansing the data making api calls and then having the rag systems up and working and then feed it to the intelligent system and then get some results out and the composition of the compute that we need to run such systems is fascinating. 80% is CPU, which is not very well talked about. And the number is just a rough ballpark. This is just to give you an idea. But majority of these services require CPU. And just let's reflect on IBM Power. It's been exceptionally good in enterprise ecosystem, fault tolerant, right? It performs very, very well. Large applications have been deployed. We continue to take that, and then we can couple that with accelerators. Rob mentioned Spire, and there are other GPU capabilities that will be operationalized very, very soon, which will be socialized with IBM, all on its way. But I'm saying we don't have to wait. We don't need to wait to really have our enterprise customers go and then start to activate their AI workflows. And today, we can do that on IBM Power 11 S1112. do, and we'll see that in a demo shortly. Now, as I was mentioning, intelligence is a portion of it. IntelliX is an intelligence orchestration layer. KnowledgeX is accelerator, which actually brings in the enterprise data and gives better context to intelligence. And SpeedShotX is nothing but a very easily configurable UI-driven application or accelerator that you can actually use to create AI agents, workflows, and then so on. At a very high level, you have a very, using IntelliX Prime, you have very solid intelligence layer, and you have vertical accelerator coupled together to help enterprises accelerate the AI workflow adoption. Not intelligence adoption, not just an API call, not the GPT call, but workflow adoption within enterprise. And just to make it simple, this is our website, fiducia.ai.com. When you go there, IntelliX Prime is foundational and VideoX, KnowledgeX, SpeedShotX are accelerators, bringing all the way from infrastructure to all these capabilities. It's nothing but acceleration, jump-starting enterprises to start to see AI workflows within their organizations. And this is a very, very high-level view, how the systems, the platform and accelerators would blend to yield transformative AI workflows within the organization. With this, I'll actually switch to demo. I'll pause, and then I'll want to check if there are questions, JT, Kathy, or others. Get to the demo. Okay. I just want to be sure you're able to see my screen. You're seeing IntelliX orchestration layer or the foundation layer. I'm logging in, but just want it to be very, very clear. This is directly going to the power box on IBM Network Lab. Sorry, I have to stop sharing for a minute. I have just copied the passwords, please. Sorry about that. Sorry about that. J.T., if there are any questions, maybe you might want to, while I get the passwords for this, please. J.D., maybe you may want to say a few words while I take this. Yes, Ganesh loads up the capabilities of IntelliX Prime. Does anybody have any questions at this time or use cases they would like to explore or address at this point? okay I think I am I am back I'm logged in let me let me go back to the stream yard screen and then start sharing okay perfect so I had to stop sharing just because I had to enter username and password to log into the system, I wanted to give a little overview in terms of what you're seeing right now. The URL points back to the S1112 configuration. What it means is 10 cores, half a terabyte RAM, and 10 terabytes of storage. Is that a requirement? it's a good starting point to actually see very large language models operational so what you are seeing is an admin view admin view of intel x prime where if they have to do some checks the user can come and then actually start to do checks so what we have deployed on the S112 box, you have access to QEN 3.5 on this box. Granite will do some tests, very quick tests with Granite. And also, if enterprises want to use this as a gateway to frontier models, they always can. We have configured 5.4 here. It is not a requirement at all, unless and until it's a use case and requirement. So what we are discussing right now, we always think in terms of air gap, meaning it does not need any frontier model support to build your AI workflow. If there is a need, if this enterprise has a need. We don't want to not let them configure. They can come here, configure, make it available to the enterprise. But be able to actually see all the activity that is happening across all these different models. We'll start off with one of the complex scenarios. Image processing is complex. I chose Quen and then very soon I'll pick Granite and then we'll just ask, we'll start off with a very simple question and say, oh, I'm just trying to, okay, which model are LLM that's actually working on S112, and the response time is 2.5 seconds, and the total token. Look at the parameters that you're actually starting to see. Why this is important? This is tokenomics. This is where companies are starting to pay a lot of money to make calls to Frontier while risking the data going out, the intelligence going out. And what you have seen is real. Enterprises can actually host, control, and experience what you're seeing on CPUs. We have not even touched the accelerators with CPUs, and this was a very simple text prompt. We'll just go switch to Granite and ask the exact same question. to 2.5 seconds. And this is a very powerful frontier model with a lot of compute behind it, 2.1 seconds, right? Now we'll do something very, very interesting. Interestingly, as we speak, I'm traveling in the Camp Roberts area where a lot different use cases are actually being discussed. and image processing is super important this is a synthetically generated data and what I'm saying is give me nine lines a very simple prompt what nine lines is the most important things that that you can actually see on the image from the image process it and send it back to me I'm a soldier I need real-time information and look at look at how quickly we were able to actually start to see a response from response from uh gp sorry this is gpt 5.4 let me switch back to 3.5 i'm doing the exact same exercise with 3.4 but let's pay attention to the time and tokens too about two to three seconds This is image processing. This is a frontier model, and we do the exact same thing with QAN 3.5 at this point in time. We can onboard multiple open-weight models for folks who are starting to understand what open-wait models are. Open-wait modules are available to download, install it, and deploy, which is deployed here on our platform. it'll take about eight to nine seconds on cpus to actually see the very similar results image processing image being sent to the power 11 box intelliX platform look at the depth of information that we are actually getting back for the same prompt on on the open-weight models. To me, I've been doing AI for 15 years. Just for context, I was the vice president and chief technology officer at Verizon, which we all know, a large corporation. I was running the AI portfolio for nine plus years. As part of the job was primarily to design systems, massive systems to deploy it in other telcos. And that was the external facing work that our team was primarily doing at Verizon. And now, with some of those experiences and the partnerships that we have with IBM, what I'm seeing is it's fascinating. 9.5 seconds, the first token. Look at the depth of information that you actually have from an open-weight model. Fascinating. And that was a simple defense kind of a prompt. We'll actually go talk about a few corrosion detection. Oh, so this is an image. This is again synthetically generated image I just uploaded just for this particular demo. The prompt would be... and it's super important to actually start to see know and believe three four five seconds you start to the recommendation but one thing that i wanted to call out as we as it continues continues to write. We have not fed the enterprise data yet. This is purely an intelligence back from the LLM. So very quickly we'll actually switch to KnowledgeX. I'll give a high-level overview in terms of what KnowledgeX is and how KnowledgeX, which is very important, be coupled to intelligence so that it actually gives the right messaging and then so on. It goes on and And you can always fine-tune the prompt to get the kind of information that you're looking for and then so on. And the open-weight models ecosystem is actually starting to flourish very, very well. The results from 3.5 is what you're already witnessing, these results. We have tested 30 billion model, MUSE model, fascinating. It's very comparable with the frontier models. And this ecosystem is evolving. And the story today is for enterprise workflows, we have compute capabilities that we can deploy using power series. We can deploy intelligence and then switching to, and I'll park this here, the intelligence story. We can go, there are more labs and sessions on its way. The last slide will actually share the schedule. We can bring a customer use case and then actually really try to solve, stitch all these things together live as part of the conversation. That's going to happen. We'll be talking more about it towards the end of the talk. But before we switch to KnowledgeX, what I wanted to share, summarize what you saw. To me, it's fascinating. Being an AI for so long, but to me, what I believe is from experiences in the past and then what we are seeing right now, We don't need accelerators to get AI going for enterprise, and this is just the beginning. We tried Qend, Granite, and we tried to compare that with GPT-5.4, a frontier model. Ganesh, we have two questions for you. Go ahead, JT. Go ahead. First question is, if a customer has a SPRE card, how can IntelliX add value? If a customer has? A SPRE Card. A SPIRE Card. Oh, yeah. Oh that's an amazing question. What you are seeing has we have not used SPIRE Accelerator at all. We can use SPIRE accelerator to address two very important things. One, the first token response time will drastically improve. And we can work together to actually demonstrate that. And the second one, which is even more important is, and this is where I want to be very mindful when I actually when I say things, it's a good starting point, but say if 100 people want to actually make a call to the LLM, then you really need more compute. And that's where scale comes into play. And with Spire, you can actually increase that concurrent request coming into you. So there are two things, very important things that we can address. One is the first token time improvement. The second one is a number of concurrent user requests can go up on the same footprint. And Spire at this point in time is specifically focused on certain machine learning models. I spoke to the Aspire team. I don't want to make any statements on behalf of IBM, but one thing that I can actually say is it's going to be more generic going into 2027, supporting a broader open weight models. And I exactly don't know which models. We can take this offline and then rob us here and then we'll get the right answers around Aspired. But that's a great question. That's a good question. Yeah, I'll add, Ganesh. I mean, this question was around someone who already had Spire, but where Spire plays here is if they're on a system and the workload grows, it's a growth option for us, like Ganesh said. So a customer could deploy on a smaller footprint and then grow, and versus buying a bigger system, it is a growth. Is that fair to say, Ganesha? Oh, definitely, yes. There are two ways to grow. So you can think about growing with larger GPUs, right? You can grow with Spire. So the way how I see compute from compute perspective, inferencing, I just want to be very, very clear too, from inferencing compute perspective. CPUs are a great place to start. Accelerators like Spire, amazing growth. And massive, you can always think about GPUs and other things as well. But we want to explore CPU path as best before we actually start to explore GPUs. And I'll have David and Rob address any GPU-related questions. But separately, the platform has been tested across hybrid environments, x86, GPUs, CPUs, IBM Cloud, right? All these permutations and combinations, and we have PowerVS, we have tested it on PowerVs on the power on cloud as well. So there's various tests that have been done because these components are different. So the first one that you saw is IntelliX Prime that can be hosted in very, very high compute. And then we can actually look at other low compute areas for some of the other modules that we'll be discussing. And I think if I could add, Ganesh, yeah, I was going to add too. I think what's incredible about this partnership and what we're trying to do here is, you know, everyone hears about GPUs and the shortages and everything else. What we're solving here is going after accelerated workloads, more cost effective, right? Ganesh and his team has written an efficient way of doing business without having to pull in GPUs. Now, we could use GPUs, but that's not the point. The point is the power core is good enough as it is. And as you need more cores, the way that we're limiting the systems and keeping things cost-effective is adding Spire as a growth option. So we believe there's a huge business here where customers want to do accelerated workloads, but they're not going to spend the capex on massive GPUs. Is that fair, Ganesh? It's perfectly perfect. I can't say that any better. Thank you. And I'll go ahead and also mention for those that, you know, everybody always asks about IBM and the GPUs. We are working on an offering in the case where that may be needed. You know, maybe we have a customer out there that's so large that we need a GPU. We are looking on an offer that will be coming soon. So we'll have an answer for that as well. But the main focus right now, as Ganesh said, is the cost-effective. The people, I mean, the retail example, I think, is great. I mean retailers are not looking to spend millions and millions on GPUs. They sell, you know, in your example here, they sell clothes. That's their business. Totally. Awesome. Thank you, Rob. Thank you. And very quickly, so what I did while Rob was speaking is actually I said list all the power models, right? So what is happening in KnowledgeX? B, you get to upload and index all your data into KnowledgeX and organize data that will be needed and can be coupled with different prompts that go to start to build those workflows. And what I was trying to do is I was starting to validate. There's a common term called chunks that are actually used. The data is vectorized and it's already organized. But this is the important message that I would like to share. This is deployed on as the same box, the same footprint, S1112. Look at the response time, one, two, three seconds. You actually get all the relevant information from a very large file, which then will be coupled with the AI that we actually have seen to create AI workflows. I'll show a few retail workflows, and I just want to be sure that we save five minutes towards the end. In that way, we share a very clear plan in terms of how we are partnering with IBM and also the clinics that are on its way where we can discuss the customer use cases and scenarios, not just discuss, but get a few things done. Because what we believe is making AI real in days. Let's see. So I quickly wanted to do, this is a retail demo. I'm just switching to retail. But I also want to be very mindful when I actually say this. So this demo is not deployed on the S11. So we use this demo to showcase this to customers. And these are a few things that can be easily deployed on power. There are certain capabilities for which you definitely need Frontier Model, which is nothing but a try-on, very deep learning, intensive, a lot of blending, and then so on. And that's one of the reasons why I actually mentioned if there are use cases like retail, you're still sitting on a small power box, not letting all your confidential information go out from the enterprise. Oh, but we wanted a frontier model to do something that's possible. So a quick example is bag. Look at the magic. I used voice. The page refreshed and the bag showed up. Blue shirt. That's future of retail. We're not going to click, click and click. We're going to start to use voice. We can use any voice and we can use natural language too to start to see the response and then the page being reassembled to what you're actually looking for. But we're not going to just leave it there. We're going to hit the try on button. And let's see, I'm going to go back a little bit. So when that comes up, so what exactly is happening here is retail, pretty cash-strapped. They want things to happen very quickly. We figured out a way where if there is a small large lot, that's about it. That's me from black to blue shirt. That's magic of workflows. I wanted to show this primarily because what you saw is a workflow. You saw the entire web page being reconfigured using Wwise. you're seeing a try on live on the website and this website is accessible too maybe JT might want to share the link I don't know if there's a way to share you can always go to the website and then try it on but I also want to be mindful and then be very transparent and this is a pointer to the cloud deployment at this point in time but that's just one example but I wanted to run through some slides very quickly so that we all know and believe that this is real and what i'm sharing is a real story a very high-end luxury brand kate barton um sancho is there on the call and kathy from ibm instrumental in actually bridging the gaps working with them and working with uh with watson x working with the the brand luxury brand owner kate martin herself and getting this activation. And the story is as follows. New York fashion show. And they wanted to actually have a try-on live where people can come and then try Kate's dress. Five days, that's about it. And you can scan the QR code. I think it'll go straight to Kate's site. You can actually look at it. That's all women's dress You can take a look at the QR codes. but it is actually integrated to the Shopify website as well. But the live demo was the live activation. Within five days, we took all Kate's address and then converted it into an activation. And that was a blast, including it was magical to see Miss New York when I actually asked how she liked the activation. She said, oh, what I'm wearing is awesome. this is recorded by the way what I'm wearing is with her permission what I am wearing is awesome but Kate's dress is beautiful too this technology is pretty cool is what she mentioned but the activation in today's world retail you have video there's a lot of social media there's numerous things and when I say accelerator SpeedShot X is an accelerator you can create all these things without code very quickly because the life of the content is so short the life of the activation is so short you can't spend six months for two days you spend one hour for two days or 30 minutes for two years and that's the power of accelerators where you use these accelerators to create workflows we really specialized very well with retail we are working industrial we're working in defense and other areas as well and this was the show five days and Kate got very excited she wanted it to be integrated with Shopify website two days that's about it Two days, it's integrated with the website. The QR code points you back to Shopify website. But there's a story to tell. New York Fashion Show partnership, IBM partnership was so crucial for us. Without that, I don't think we would have had the Kate Barton story because we had to work together as a team to make this happen. Thanks to Sancha. Thanks to Kathy. And now for the power, David, Rob, and the large power team working together. But going back to Kate, featured in TechCrunch. this is real, WWD, fashion, CFDA, I call it Oscars for Fashion, primarily in the New York area. We got featured there, and a lot of blogs we did. But also, I was touched when I started to realize this is being taught at Rutgers University. This has become a very fascinating story. And now this is not intelligence. This is not frontier models. This is workflow and value for those workflows. But how do you deal with large enterprises? Kate had 100 products. There's a lot of manual configurations and so on. We started to work with a very large brand who had 100,000 plus products. How do you solve that problem? And these things help us create. And that's where the partnership with IBM is so valuable for Fiducia, where, one, we get to innovate. We get to innovate very fast, primarily because that was an IBM opportunity where we innovated. And now any retailer for the try-on activation, less than a minute, they go add a try-ons button with a URL, and that's about it. They're done. And it's completely integrated with the platform, IntelliX Prime platform. Future, it is changing. So try on a lot of these things that we have actually seen. We humans, we go, we touch, click, and then so on. There's one big change that is coming for retail industry. The barcodes are going to become QR codes. We are working with GS1 Innovation, who issues these barcods to see how we can actually come up with these activations that can socialize to the broader ecosystem. And these are massive, massive changes that are on its way. And we are in the middle of a lot of these changes. And then we are the trial blazers where we are actually starting to socialize, not on slides, but real, on activation, on IBM Cloud, on PowerBS, and now S1.1, 1.2. Now, the one very important thing that I want to reflect on, agents would start to act or make decisions on our behalf or help our lives easy. What does it mean? More data, more AI, and this is inevitable. That's where the world is headed. In the morning, I would actually wake up and say, two days from now, I'm attending a dinner party. I would like to be dressed in white and black. Can you get those options out to me? You'll see the option, you go, you approve, and then it gets shipped. That's the world that we would actually start to live in. Whether it is six months, one year, or two years, time will tell. But what I can say is it is inevitable. Every enterprise, small or large, even the smallest brands, they need AI. And cost is important. And it's a myth that we need accelerators to run AI workflows, which you have already seen. Let's start small with your customers and then start to grow. This slide is nothing but a reflection of different capabilities that can be pulled together for activation. And I'll spend like one minute and then say a few things about the offering, how you can get this, and then pass the ball to JT and Kathy to close the call. Maybe me, JT, or Kathy will actually close the called. What you're seeing is all there. But what exactly it means? The end product is AI workflows. How is it possible? You can deploy it on Power 11, Box, the smallest S112, depending upon the use cases, and graduate to others. You can start immediately on VS or on IBM Cloud. IntelliX prime layer can be deployed on any of these things and accelerators the accelerators that I mentioned even VARX for example defense accelerators are there I'm at Camp Roberts right here nearby and my day is to actually talk to talk through a lot of defense use cases on CPU power configurations because everybody's cost conscious. And top layer is a reflection of high-level verticals. Doesn't matter which vertical your customer or potential customer falls into. We can actually start AI workflows and make it real for them, cost-effectively, but production-grade, enterprise-grade using IBM Power 11. With that, clinics are coming up. I think slides will be shared. You might want to take a snapshot and then reach out to Kathy, David, JT, or I. But this is just the beginning of our conversation. We really want to work with you, bring in your customer use cases. All your questions we'll answer, or together we'll actually build a small activation on Power 11 and be sure that it actually works and engage your customers and then have them use more power, Power 11, Power VS, IBM Cloud for their AI workflows. And Watson X is a big part of our journey. I just wanted to reflect. We include Watson X on the cloud and also we can have calls made to Watson X from any any endpoint as well so with that uh i'll stop sharing but there are let's see how i go there okay i'll start sharing so a few things that i wanted to say that way we have a clear call to action this is not the end clinics are on its way at IBM and Fiducia, we are accessible to actually set up those sessions to discuss use cases. But how do we get started with the customer? We can start off with the smallest box. We don't have to wait for anything. We can actually get the boxes ordered. And there's a 25% promotion till the end of the year going on for IntelliX Prime offering. 75K is the list price for the Intelli X Prime. and the accelerators that can be deployed, and see some of the magic that you have seen. That covers 10 cores. Anything beyond 10 cores, you have 11 or more than 11 and other combinations with Spire accelerators, then that is 150k, but the 25% off is till end of the year. But this can be activated as we speak on IBM Cloud, not in the Power Ecosystem, but on IBM Cloud. As a trial, zero cost, 30 days, click of a button. Sorry, I couldn't get to that demo. You go, order, give your email, activation is up, and you can actually start to create some from AI agents and then so on, and then just work with just to get a feel. We'll be happy to socialize that. PowerVS is a great place which can be deployed. We can wake up some of the dark course. But we want to see this being deployed on S1112. Good value for dollars. And S1122. With that, I pass the ball back to JT. JT, I'll have you say a few words. Sorry, I took the whole time. Real quick before we pass it there, Ganesh, I just wanted to make a comment. Guys, in the beginning of the slides, if you remember, we covered two example configs that could be ordered today, which is what Ganesh was referring to. And we're working with Ganesha team to actually enhance the 1112 to be more of an appliance. And Ganesa, there's a question about, is this an application that sits on top of Red Hat with power? Yes. Good question. Yes. We have tested it for Ubuntu and Red Hat. Yes. It's not a requirement, but we have both options depending upon the customer's preference and need. We, part of IBM, we want a Red Hat, and it's tested on Red Hat power combination as well. OK, JT, off you go. Does anybody want to post any additional questions for this session? As Ganesh has mentioned, that we have workshops that would be happening every Friday between now and towards mid-December. Feel free to email Kathy or David at IBM to schedule some time to join those workshops. And we'll be posting those workshop calendar dates on LinkedIn as a follow-up as well. And then gonna actually talk about our special promotion that all activation between now and end of the year will be a 25% discount. And this also will lock in the prices. So what we're saying here that your price will be locked in with a 25% discount for all future deployments in 27. Perfect. Awesome. Thank you. I'd like to add that we do have plenty of seller resources available on Seismic, so you guys can go enjoy yourself, as well as the demos that Ganesh was sharing with you today are available on seismic. All you have to do is pull your phone up and hit the appropriate QR codes or click on the browser link, and you can do the exact same demos that he shared with you. If you're looking for the configuration and pricing, here is the e-config as well as the software that you can purchase. So you don't have to wait. You can get started today. And as you can see, the on-prem software will be available on the 14th of November. So start having those conversations today so that we can help you move this along and hopefully turn this into 4Q revenue. I'll turn it back to you, Ganesh. Okay, awesome. Again, thank you all very much. It's a privilege and pleasure to talk to each one of you. We are here. Let's make this real for our customers. and don't miss out on the 25% discount which will expire end of this year but try it on we are here to help and also we are very grateful for the partnership with IBM very rare it happens for somebody like me who's been an engineer for life to discuss all the way from silicon all theway to taking applications live and we are making great progress we would like you to be part of this journey thank you all very much and enjoy the rest of the day wherever you are bye
- Company
- ILOG SA
- Keywords
- ILOG SA