Beth Williams, the global portfolio lead for artificial intelligence, applications, and data at Dell Technologies, joins host Dave Vellante, co-founder and co-CEO of SiliconANGLE Media, at the iconic New York Stock Exchange. In this engaging episode of "AI Factories: Data Centers of the Future," Williams discusses Dell's transformative approach to enterprise artificial intelligence and the evolving landscape of AI applications in data centers.
In this insightful discussion, Williams shares expertise in helping organizations transition to AI-ready environments. Known primarily for its hardware, Dell's consulting prowess in areas such as data strategy and technical feasibility emerges as a well-kept secret. The conversation highlights how Dell assists enterprises in selecting the right AI use cases, ensuring data readiness, and proving value early. This is supported by the insights from theCUBE Research and their video hosts.
Key takeaways from the discussion include the challenges and opportunities in AI adoption, particularly the importance of beginning with data readiness assessments before tackling large-scale projects. Williams emphasizes the role of agentic AI in improving process workflows, as organizations strive to harness data's full potential. This is in line with both their insights and those shared by analysts.
Forgot Password
Almost there!
We just sent you a verification email. Please verify your account to gain access to
theCUBE + NYSE Wired: AI Factories - Data Centers of the Future. If you don’t think you received an email check your
spam folder.
Sign in to AI Factories - Data Centers of the Future.
In order to sign in, enter the email address you used to registered for the event. Once completed, you will receive an email with a verification link. Open the link to automatically sign into the site.
Register for AI Factories - Data Centers of the Future
Please fill out the information below. You will receive an email with a verification link confirming your registration. Click the link to automatically sign into the site.
You’re almost there!
We just sent you a verification email. Please click the verification button in the email. Once your email address is verified, you will have full access to all event content for AI Factories - Data Centers of the Future.
I want my badge and interests to be visible to all attendees.
Checking this box will display your presense on the attendees list, view your profile and allow other attendees to contact you via 1-1 chat. Read the Privacy Policy. At any time, you can choose to disable this preference.
Select your Interests!
add
Upload your photo
Uploading..
OR
Connect via Twitter
Connect via Linkedin
EDIT PASSWORD
Share
Forgot Password
Almost there!
We just sent you a verification email. Please verify your account to gain access to
theCUBE + NYSE Wired: AI Factories - Data Centers of the Future. If you don’t think you received an email check your
spam folder.
Sign in to AI Factories - Data Centers of the Future.
In order to sign in, enter the email address you used to registered for the event. Once completed, you will receive an email with a verification link. Open the link to automatically sign into the site.
Sign in to gain access to theCUBE + NYSE Wired: AI Factories - Data Centers of the Future
Please sign in with LinkedIn to continue to theCUBE + NYSE Wired: AI Factories - Data Centers of the Future. Signing in with LinkedIn ensures a professional environment.
Are you sure you want to remove access rights for this user?
Details
Manage Access
email address
Community Invitation
Beth Williams, Dell Technologies
Beth Williams, the global portfolio lead for artificial intelligence, applications, and data at Dell Technologies, joins host Dave Vellante, co-founder and co-CEO of SiliconANGLE Media, at the iconic New York Stock Exchange. In this engaging episode of "AI Factories: Data Centers of the Future," Williams discusses Dell's transformative approach to enterprise artificial intelligence and the evolving landscape of AI applications in data centers.
In this insightful discussion, Williams shares expertise in helping organizations transition to AI-ready environments. Known primarily for its hardware, Dell's consulting prowess in areas such as data strategy and technical feasibility emerges as a well-kept secret. The conversation highlights how Dell assists enterprises in selecting the right AI use cases, ensuring data readiness, and proving value early. This is supported by the insights from theCUBE Research and their video hosts.
Key takeaways from the discussion include the challenges and opportunities in AI adoption, particularly the importance of beginning with data readiness assessments before tackling large-scale projects. Williams emphasizes the role of agentic AI in improving process workflows, as organizations strive to harness data's full potential. This is in line with both their insights and those shared by analysts.
Global Portfolio Lead, AI, Apps and DataDell Technologies
In this theCUBE + NYSE Wired segment from AI Factories – Data Centers of the Future, theCUBE’s Dave Vellante sits down with Beth Williams, Global Portfolio Lead for AI, Apps and Data at Dell Technologies, to unpack how enterprises move from pilot experiments to production-scale outcomes. Williams explains Dell’s often-under-the-radar professional services (spanning consulting through managed services) and why success starts with choosing the right first use cases, testing with real data and mapping a clear path to ROI and post-pilot placement (on-prem AI fact...Read more
exploreKeep Exploring
What is the role of the organization mentioned in the text in relation to enterprise AI and customer readiness for AI?add
What factors contribute to the failures of AI pilot projects, and how can companies improve their chances of success?add
What approach is taken to encourage organizations to begin their data transformation journey?add
What were the main use cases being considered for AI implementation?add
>> Hi, everybody. Welcome back to The New York Stock Exchange. We're here above the Options Exchange and the Buttonwood podium, the iconic NYSE. You're watching AI Factories: Data Centers of the Future. My name is Dave Vellante, and we're a super excited to have Beth Williams here. She's the Global Portfolio Lead for AI, Apps and Data, three of our favorite topics, from Dell Technologies. Beth, great to see you again. Thanks for coming in.
Beth Williams
>> Great to see you, too.
Dave Vellante
>> You are in the organization that's helping customers get ready for AI, and we're specifically in this series, Beth, we're talking about enterprise AI. We know there's a lot of action in the cloud. We see all the green. We know all these last weeks and months here at the New York Stock Exchange with all the CapEx that's going in and the clouds and the neoclouds. I know you guys are supplying a lot of those, but we're really interested in focusing on enterprise AI, bringing AI to the data, how organizations are thinking about that. I want to start with, first of all, share your role because I think a lot of people don't really understand that Dell has this capability and works very closely with customers. They think of you as mostly a hardware company, which you are, but you've also got this consulting capability, so explain your role, if you would.
Beth Williams
>> Yeah, I think we're probably the best kept secret for enterprise, even within Dell. As you say, we're often known for our storage and our laptops, but one of our biggest capabilities is around services, professional services. That can cover anything, say, from consulting services all the way through to managed services. My role is to make sure that we've got the right services for the customer so we can support them all the way through from the initial kind of day zero strategy all the way through to that day to manage services.
Dave Vellante
>> Okay, so customers obviously excited about AI. Boards are saying, "Hey, what's your AI strategy?", but what are you seeing? What are you hearing from customers? What are the big challenges and concerns? Obviously, they're concerned about the legal and compliance, but what are you hearing in the field?
Beth Williams
>> I mean, what we're seeing as well with some of the kind of recent studies like the MRT studies, we're seeing a lot of AI pilot failures, and I think there's a number of reasons for that. Some of it, I think, is maybe the choice of the pilots to begin with. We learned the hard way within Dell. We've got to choose those initial use cases really well. One of the things that we try and do with our customers is learn from our mistakes, if you like, and make sure that you choose the right use cases to start with. They're looking at things like business feasibility and technical feasibility, but at the same time looking at the data that's involved in the pilots. We have a kind of three-pronged approach. We look at the high value, we also look at how technical it is, but we look at the data readiness as well. Then, what we're seeing then, if we get all those things combined, if we can do a sort of proof of value with customers really early, make sure the pilots are tested on real data, what can then show is a real good path to production and also a really good path to ROI. Those initial choices around use cases, I think, are really important, but then also where are those use cases going to live ongoing? Are they going to be on-prem? In which case our AI factory's fantastic. Are they going to be in the cloud? In which case is data privacy important? Looking at that production next step as well. As I say, in the industry, what we're seeing is a lot of AI pilots stalling, but then also not necessarily know where those AI pilots should run after they've gone through that pilot phase on-prem, cloud, where is that placement for both the AI and the data?
Dave Vellante
>> There's also a lot of obviously POCs going on in the cloud, low risk, don't have to spend the CapEx. You could spin up, spin down. That's great, and maybe you can even use synthetic data to try to just get a groove swing going. Then, we're seeing a lot of customers saying, "Okay, now we want to go into production, we're going to do that on-prem." We're going to bring the AI to the data, we like to say, but they don't necessarily have their full data act together for that use case. They've, like I say, maybe they've used synthetic data or maybe a subset of their data. How are you helping customers get their data act together? What are you seeing? What are you recommending? How do you start? I mean, I'm nervous about customers trying to like a snake swallowing a basketball, so how are you addressing that? How are customers addressing it?
Beth Williams
>> We try not to scare people into saying you need to do like a two-year transformation project before you do anything on-prem. That's the first thing we try to make sure we get right to start with, which is we ourselves have been on that kind of data transformation journey for quite a while. We've got some great experience with our own Dell IT about how we've built a kind of framework which allows us to have data products as a service, but we started small and we didn't wait until we had all those pieces of the jigsaw in place before we started doing use cases. What we try and do with customers is, first of all, work out where they are. We have this data readiness assessment that we work through with customers, which has the six dimensions that we look at, and those dimensions are what we had to learn ourselves we need to get in place before we did large-scale use cases and data readiness. We look at those six dimensions, which are things like strategy and governance, whether or not you've actually got the skills in place to do this kind of new data product landscape, which is kind of a different way of thinking about data. What are your abilities at the moment around pipelines to find the data and just the data? Do you have data catalog? Are you virtualizing data? Silos of data are a real big problem for customers right now. They've got lots of data in a lot of different places, structured and unstructured. Are they using virtualization? Do they have things like observability and quality in place? Then, finally, where we obviously have quite a few great products is that the actual platform technology itself. We look at those six dimensions to start with with customers, and this might sound like it's a long process, but actually we can do this really quickly. We have this accelerator workshop, which we can do in about half a day to a day. It gives us a good sort of heat map as to where customers are and where we might need to help them in terms of following our journey to that full data product as a service. Then, what we do is instead of saying, "Right, okay, everybody stop for a year whilst we build this amazing data platform," we say, "Okay, let's start small. Let's take your important use cases. Let's make sure we get the capability in place for those use cases, and let's incrementally build on those capabilities all the way from, say, a lake house to virtualization, the governance, all the way up to data products." As I say, it's an iterative journey that we do, and at the same time, we're giving value in terms of implementing real use cases with real data whilst we're building that data platform.
Dave Vellante
>> I have kind of a strange question for you, but I hope I can make it make sense. You know how we talk about AGI all the time, artificial general intelligence, and let's say we hear with whatever state AI is today, GPT-5, and AGI is somewhere over here and there's this gap. We're not really quite exactly sure how we're going to get there. We know with scaling laws and the increased performance we're going to get there eventually. We're not sure when, we're not sure how. Is there a similar dynamic with your customers in that there's this North Star, agentic AI, I've got millions of agents? I've got these AI just spinning workflows and my organization has 10X productivity and this is this amazing North Star. We're not really sure how we're going to get there. Do you even bother laying out that North Star and trying to figure out a roadmap as to how to get there, how to harmonize the data, how to have an agent control framework, how to secure it, et cetera? Or do you say, "Look, let's just get some quick wins and get a groove swing going, muscle memory, all these sports analogies, and then sort of figure out what that reasonable North Star is?" How would you recommend approaching that?
Beth Williams
>> Yeah, I think that's a great question. The one thing around AI at the moment is I'm not entirely sure we know what that North Star is. I think we might know what it is, but there's so much change and innovation happening at the moment that if you think too far ahead right now, I think we might be slightly off the mark. Whilst I like having a North Star, I'd like to think maybe a year or two in advance. Right now, I mean, you mentioned it. Right now, agentic is where it is. We started off the journey with large language models and then we realized quite quickly that we needed to think about what in context those models with, say, RAG. Okay, so we weren't that's something you go through all this training to get our own models. We're still going to take pre-trained models, add our context using RAG. What we're seeing as the next iteration of that, to your point, is very small kind of steps towards that North Star is agentic RAG. One of the issues with using large language models, even if you are using retrieval-augmented generation is that you can still get hallucinations. You can still get incorrect answers. I think one of the big problems people have seen so far with large language models is they don't necessarily trust the output. Until you can trust the output, you can't trust your agents. One of the things that we've seen really expand is this ability to use agents within that RAG workflow so that you can improve the data quality. You can improve the output. Less hallucinations, more relevance. That was the first kind of step up from RAG to agentic RAG. Then, we started looking at, "Okay, so now where can we use agents for other workflows? Where can we start to say, 'Let's build an agent to help a sales assistant', or, 'Let's build an agent to help a support assistant?'" We're starting to see the processes that are most important to customers being supplemented with those agents. I think that's what we're going to start to see evolving over time, more and more of what I would call foil, the repetitive work, the work that takes a bit of time. All of that will be supplemented with agents. Then, what we'll also see is innovation, things that we can't even think of today that are now possible because we've got those things in place. Then, who knows where we're going to be in a couple of years time. I think it's going to be really exciting.
Dave Vellante
>> But very practical advice. I appreciate that pragmatic approach. What patterns are you seeing in adoption? Are there any sort of industries, verticals? I mean, we're seeing a lot, obviously, financial services. They have the skills, they have the budgets to go hard after AI. We're certainly seeing a lot of real strong results in coding use cases. What are you seeing? Any learnings from internal Dell that apply there, that apply broadly in the industry? Share your expertise there if you would, Beth.
Beth Williams
>> We had our four main use cases that we were looking at for once we kind of whittled them down to the most informed ones. One was kind of supply chain optimization with AI. The other were including AI in our products, but the two I think most helpful one that we've seen recently around things like assistance for sales and assistance for support, and those are predominantly chatbot. Chatbots with data from your organization help either support your own customers or support your internal teams to get to the outcome they need. I'd say we've got about a number of proof of values that we're running at the moment with customers. 50% of those follow a chatbot, pre-trained LLM RAG model, because that solves so many problems for customers. Yes, there are certain niche use cases that we see. You can start to see a combination of RAG and LLMs which sentiment analysis as well. We had a customer recently in financial services that wanted to get some feedback from all of these data points they were getting from customers around their branches. We managed to get all of that through from those 13 million data points user, just a RAG, LLM, and a chatbot. I think in all honesty, if you get those kind of core pieces in place, so many of your processes get improved. There's lots of other things like code generation, content creation, things like digital twins as well. There's loads of other interesting use cases as well, but I think those kind of search and chatbot capabilities are cornered to most businesses regardless of what vertical you're in. Everybody needs to know where their data is. Everybody needs to get answers to questions, whether that's just them talking to customers or internally. I try and explain to people, don't go after the really niche stuff, solve the 80% problem. The 80% problem is you've got a ton of information in your organization. How do you make sure that people get that right information quickly, either to help themselves or help customers?
Dave Vellante
>> Where would you say your sweet spot is, Beth, that your organization is most comfortable, you've got the best experience, your playbook is proven? Whether it's shaping the stack, getting the data house in order, governance, security, dealing with hybrid situations, where do you see your strongest event, if you will, your strongest swim lane?
Beth Williams
>> I think we're strong in a lot of different areas. I think to your point earlier on, sometimes we're the best kept secret. A lot of people know us for our products, so a lot of the times when we start with customers, it's going to be at that layer. They understand that we've got these great servers with NVIDIA GTs and so on, and so a lot of the time we'll start off with an AI factory conversation because that's where they know us to be. "Hey, you know, we've got these fantastic servers. Now, go and build that software platform on top." We also have fantastic heritage in data, and we also have our own internal skill around data as well, because as I mentioned, we've created our own data product as a service platform, which we use ourselves. Data's cornerstone to us. Our AI factory is cornerstone to us, but we also have a fantastic security resiliency business that's been going for a very long time that's now supplementing things like the AI factory and the AI data platform. I think we have fantastic capabilities in all those areas. I think people know us to be a little bit more niche because they know us to be Dell, but I think we're starting to expand that footprint.
Dave Vellante
>> Do you think the future of enterprise AI is maybe less about the models, the size of the models, all the benchmarks of the model, and more about who can harness that data, who can harmonize that data? How do you secure that data trust, make it trustworthy so that systems and agents can take action? If you believe that, I wonder if you could expand on that and explain your thinking.
Beth Williams
>> Well, yeah, so I mean, data is the goal, and I think that for many years there's been a big pool of data that we've not been able to harness. Unstructured data being that big pool, so if you think about the fantastic amount of unstructured data that we have that we haven't been able to really get insights from, all of a sudden now that's available to us. Video search and summarization, being able to ingest lots of PDFs, being able to scrape lots of internal SharePoints like being able to suddenly harness all this data. It's going to open up so many opportunities for us to be able to innovate and improve our processes, so data is the key. I think, as you say, the emphasis on which model is best, I think is going to potentially sort of become less. I think it's going to be more about how would you bring the right context to the model, which is all about where that data is, how you ingest it appropriately. How do you clean it? How do you prepare it? As soon as you harness all of that gold and all the unstructured data that we've just been keeping, I think that the possibilities are endless.
Dave Vellante
>> Well, Beth, I want to thank you for coming on to the AI Factories: Data Centers of the Future Program. I'm looking forward to seeing you in Austin, and we could go deeper into this topic. I'd love to have you back personally, face-to-face at the New York Stock Exchange, so please put that on your wish list if you would.
Beth Williams
>> Well, I was only there a couple of weeks ago, so I'd love to come back.
Dave Vellante
>> Oh, next time you're there, let us know. We'd love to-
Beth Williams
>> Will do....
Dave Vellante
>> host you. Thank you so much-
Beth Williams
>> Thank you....
Dave Vellante
>> and thank you for watching AI Factories: Data Centers of the Future. This is Dave Vellante. For John Furrier, we'll be right back right after this short break. You're watching NYSE Wired and theCUBE's coverage right back to the New York Stock Exchange.
Dave Vellante