Teradata Possible 2024 | Meeta Vouk, Teradata

Clips
News
More from Teradata Possible 2024

Dr. Meeta Vouk

VP Product Management Analytics AI/ML

Teradata

Nvidia partnership and perks for ClearScape fuel Teradata AI offerings

This week, Teradata Corp. unveiled new capabilities for VantageCloud Lake and ClearScape Analytics designed to speed the return on investment from generative AI use cases. The new Teradata AI offerings will enable customers to add their own large language models and take advantage of small or mid-size open AI solutions.Meeta Vouk, vice president at Teradata, talks with theCUBE about Teradata’s AI offerings

play_circle_outline AI sustainability and energy consumption considerations

play_circle_outline Productizing AI and analytics capabilities for customers

play_circle_outline Enhancing Analytics and AI with Best-in-class Frameworks: Open Approach for Cloud and Integration

play_circle_outline Focus on AI production with world's most confidential data on Teradata

Info
Transcript

Meeta Vouk, Teradata

Dr. Meeta Vouk

VP Product Management Analytics AI/ML Teradata

The event at Teradata's Possible event in Los Angeles, California discussed interesting use cases for AI, including combating anti-money laundering. Traditional AI and GenAI are often used together to solve complex problems. One retailer saw a 28% increase in sales from hyper-personalized marketing emails. Sustainability and operationalization of AI are key focuses, aiming to make it simpler for organizations to adopt AI solutions. Teradata offers capabilities like AI Unlimited to make AI more accessible and cost-effective. The goal is to help customers move ... Read more

explore Keep Exploring

What specific use cases are emerging in the market around domain-specific models, and how are they being utilized by a financial institution in terms of sustainability and efficiency? add

What has the company done to help customers with the process of productizing AI and realizing its potential? add

What strategy has your company been using in order to bring the best frameworks and toolings to your customers for analytics and AI implementation? add

What would constitute an absolutely phenomenal success for Teradata next year? add

bolt Powered by CUBE AI

Meeta Vouk, Teradata

search

Savannah Peterson

>> Good afternoon data lovers and welcome back to Teradata's Possible event here in Los Angeles, California. My name is Savannah Peterson with theCUBE. Absolutely stoked to be joined by Rob Strechay today. Rob, this has been more fun than I could have anticipated so far.

Rob Strechay

>> Well, data, AI, analytics.

Savannah Peterson

>> We had magic tricks.

Rob Strechay

>> Magic tricks.

Savannah Peterson

>> We had some cool stuff.

Rob Strechay

>> We talked security during Cyber Security Month. I think we've gotten a lot already, but now we're going to get into the meat of it and where the rubber meets the customer and the rubber meets the road here, so.

Savannah Peterson

>> We're going to talk about making AI real, which I know is a passion of everyone on this stage. Meeta, thank you so much for taking the time to be here with us.

Rob Strechay

>> Thank you for having me.

Savannah Peterson

>> This has got to be a very busy week for you.

>> Yes. It's been very busy, but very exciting.

Savannah Peterson

>> Yes. Oh, I mean, extremely exciting. What is it that excites you most about where Teradata's at and the engagements you get to have with your customers?

>> Yeah, we are seeing such interesting use cases, and just to share a few with you. For example, we talked to some of the largest financial institutions in the world, and one of the things that they want to do is build out anti-money laundering solutions. The world lost $3 trillion to anti-money laundering activities last year.

Savannah Peterson

>> Wow.

>> Which a big part of it goes in funding terrorist organizations. It's a mission I am passionate about, but when we talk about anti-money laundering, you have to do, to detect people who are laundering money, you have to do two things. You have to look at graph neural networks to figure out who sent money to whom and who did they send to, and how many hops the money made. And you have to look at just troves and troves of information to see if there's any negative news about the person. And it's very, very interesting because you see traditional machine learning and deep learning and generative AI coming together to solve for an extremely important and an extremely complex use case.

Rob Strechay

>> And that's kind of know your customer, I mean, is part of that whole thing.

Savannah Peterson

>> KYC?

Rob Strechay

>> Yes. Part of the whole anti-money laundering and things. So when you see that, because I think this is a very important thing, and we talk about it a lot, is the fact that GenAI gets all of the airtime, but still there's a lot of traditional AI and ML that's behind the scenes. And it's really interesting that you brought about them working together. Kind of expand on that because I think that's really very interesting too, how organizations are really going to open up and connect into even analytics going forward as well.

>> Yeah, and for most of the use cases that we are seeing, we are seeing traditional AI and GenAI coming together and we call it ensemble architecture. So the use cases are getting so complex and to solve for them in really meaningful ways, you're going to need multiple models and multiple capabilities. And it's not really AI for AI's sake or GenAI for GenAI's sake. It's really fit for purpose AI. Are we solving the real problems? Are we solving them in a way where they can be productized? Are we solving them in a way with trust and security? Are we solving them in a way where regulators are okay with it? So we have to look at a lot of factors to solve for these complex use cases. I do want to share another interesting use case.

Savannah Peterson

>> Share them all with us. We want all the interesting use cases.

>> So we work with a very large retailer and they have about 160,000 products, and we send-

Savannah Peterson

>> Wow, that's quite a few SKUs.

Savannah Peterson

>> Yeah. They send marketing emails to about 4 million of their customers, and they hyper-personalized those, and they had a 28% uplift in just their sales. Now, the same group did another project where they looked at 20,000 of women's dresses and used GenAI to look at images of those dresses and generate copy. So their copywriting went from 30 minutes per item to three minutes.

Savannah Peterson

>> Wow.

>> And they also realized that the return rate dropped down to 6%, right?

Rob Strechay

>> Wow.

>> So we seeing really impactful work around use cases like that.

Savannah Peterson

>> That's massive. That also eliminates waste on the other end with returns, a lot of returns end up in the landfill. So you're not only solving that problem and increasing sales in the capitalistic sense, you're actually doing good for the planet by being more sustainable within these business practice. 6% is extremely low.

>> Yeah. And anything, I think you touched on a very important topic with AI sustainability. I don't know if you guys heard the news, they're opening up the Three Mile Island.

Rob Strechay

>> Yes.

Savannah Peterson

>> I did hear that this morning.

>> Yeah, yeah, because AI is power hungry, and we really have been maniacally focused on are we delivering the right building blocks, the best in class capabilities, and do we have our eye on sustainability as we do that with trust and security? So it's a passionate topic, and there's emergence in the market right now around domain specific models. So small and medium language models tend to do really well for certain use cases. So we have, working with another large financial institution, where we are looking at troves of emails between their customers as well as their employees to see, do they own a regulatory compliance? Is there undue pressure, sales pressure on the customers? And we can do that really well with a small language model on Teradata just using CPUs. We don't have to use any power-intensive compute capabilities for those use cases.

Savannah Peterson

>> You bring up a really good point. I'm giggling because I have a run GPU sticker on my laptop right now that the audience can't see. But I think most people rush to think they need the most powerful Nvidia chip to get something over the line. It's not necessarily the case. You need to be able to do it with the most cost effectiveness, faster inference, and to really make AI real for that customer in that moment.

>> Yeah, and it's, again, it touches on fit for purpose AI. We are saying we will meet our customers where they are, whether they're on-prem or in the cloud. We believe in hybrid AI, and if we need GPUs to solve for something, we will use GPUs. If you need CPUs, we'll use CPUs. You heard the announcement today. We are very excited to partner with Nvidia.

Savannah Peterson

>> Yes. Exciting collaboration.

>> Yeah, yeah, yeah. The possibility of what are we going to do next with this? There are certain use cases which will always need GPUs, fine tuning of models, creating vector embedding, so we really wanted to support our customers for those needs, and that's the idea behind it.

Rob Strechay

>> So I think you kind of went down one of the use cases, and I think we've kind of wandered around a little bit to the point that you were getting at, which is really interesting, is the operationalization of this and how you make it simpler for organizations. Because not everybody's a data scientist, not everybody's a data engineer, and the customers you deal with have differing levels from their platform engineering to their data engineering to their data science. How do you see all of this coming together and what is really Teradata helping with from an operationalization, if I can get it out, perspective.

Savannah Peterson

>> You got it.

Rob Strechay

>> There we go. Where did we go?

Savannah Peterson

>> Nailed it.

Rob Strechay

>> Yes.

Savannah Peterson

>> No, and when we talk about building blocks, to be able to get to that point where we are productizing AI and realizing the potential of AI for our customers, we've delivered extremely performant in-database functions. 150 of those customers can use them to do feature engineering and a whole lot of other things. And then, there is so much innovation happening in the open. We really wanted our customers to be able to bring it home, whether it's open source models, whether it's proprietary models, whatever's happening, and they want to bring their own, we enable them to bring them to Teradata very, very easily. We are saying bring AI to the data rather than move the data to AI. We have model ops capability to manage the entire life cycle of your AI models, which is absolutely table stakes right now. We actually had another use case where our customers was looking at quality of their spot building, and we were looking at 4.3 million sensor data, 4.3 billion sensor data reads every day. We trained 9 million models and we deployed 40,000 models in production. So you need tooling to be able to manage all of these and make it super easy. So we really-

Savannah Peterson

>> That's a lot of data. That's a lot of different ... I mean, you just said some very large figures right there. So we're talking about an incredible level of complexity.

>> Yeah, yeah. So it's making AI at scale real for our customers.

Savannah Peterson

>> I can tell how much you like doing it.

>> I'm a chip designer by training, so anytime you talk about scale, I love that conversation.

Savannah Peterson

>> Well, we are definitely fans of scale. How do you find the sentiment is? I mean, you're talking about working with huge customers at scale, right? This retail company you mentioned is big. I know you work some of the largest financial institutions on the planet. How are they, internally, with that adoption? Are they eager to go on the journey with you? Are they still scared about the investment? What's the culture pulse of the types of organizations that you're working with right now?

Savannah Peterson

>> Yeah, and I think everybody's on their AI journey. You have to-

Savannah Peterson

>> Whether they want to be or not, I think, at this point.

Savannah Peterson

>> Yeah, because I think AI is going to deliver value, which we can't achieve without it. So for example, I don't think we can control fraud, financial fraud without AI. It would be very, very difficult. So it just behooves all the financial institutions to do it. So they want to embrace it. I think it's just understanding how to get to that point where you can productize AI. Productizing AI is challenging and can we make it easy for our customers along their journey? So our customers are on a spectrum of journey. Every single customer we have is going to do multiple AI use cases, not just one. They're going to get started with one, then they're going to move to the next one and move to the next one. There's just a lot of news in the industry at the moment also saying that AI adoption is low, and I think we should look at it a little bit closely because AI adoption for our customers, enterprise customers, around Copilot's chatbots is high. These regulatory compliance use cases, it's getting there. For some of the tough workloads, they're starting on that journey. It's going to be a journey. We have to make sure at Teradata that we create the right building blocks because AI is going to evolve, right?

Savannah Peterson

>> Yes.

>> The capabilities and the innovation that happens is so rapid that's going to evolve. We are trying to make sure that we have the right building blocks that we can bring that innovation to our customers to really achieve the impossible.

Savannah Peterson

>> You bring up a really good point is it is always evolving, and I love it with a lot of your product line. It's a BYO, whether that be analytics models, whatever it might be. It sounds, what I've heard from everyone today is really the commitment to be with your customers on that journey every step of the way. Because if we are embarking on this investment, it's a huge amount of money up front even just to get started and to get playing. So how are you ... I can imagine you're constantly trying to think about what's coming next and how you can adapt the product to reflect that. What does that process look like for you and your team internally?

>> Yeah, and I wanted to touch on what you said. We are trying to make it not expensive.

Savannah Peterson

>> Love that. Who doesn't want to hear that right now?

>> Yeah. So we've created capabilities like AI Unlimited, where for, if you're a data scientist and you want to experience all of the capabilities that ClearScape Analytics offers, you can quickly spin up a node, you can experiment, or you can get some results out of AI, and then you can shut it down. And it's extremely powerful. And we have it integrated in Microsoft Fabric and available in November of this year, which is really exciting.

Savannah Peterson

>> That is exciting. Congratulations!

>> Yeah, thank you. But for all of the AI analytics capabilities on Teradata, our focus has been adoption, so they are available to our customers for free. So if you're using any of the ClearScape Analytics capabilities and database functions or any of the bring your own capabilities, our VectorStore capabilities, there's no additional cost to our customers, right? Because we know there's a skills investment and other investments that the customers have to make. We just want to partner with them and go along on the journey of AI.

Savannah Peterson

>> Love that.

Rob Strechay

>> So how, because I think going beyond the AI, you have the analytics part as well, and you're servicing out a lot of that, and you have kind of an open framework for that. How does that really bring together the power of the cloud and the power of open and being able to bring those integrations in for that?

>> Yeah. One of our strategy has been open and connected because we wanted to bring the best in class of frameworks and toolings to our customer, be it for analytics or AI, so we've made really key third party integrations with some of the leading frameworks and tooling out there. And then, the open part is bring the innovation that's in the open to Teradata itself. And with that strategy, with our open analytics framework and all the deep integrations that we have with partners, we've tried to make it super easy for our customers to experiment on our use case, move it from POC to a pilot, and then finally move it into production. And when they move it in production, they know that they can do it at scale, they can do it with trust and security, and then they can manage the entire lifecycle of those AI models with us.

Rob Strechay

>> I think that's why it's important, because we've seen that 84% or so are kind of, when they talk about why they're not in production, it's because they're actually in POC still. So to your point, and maybe they have one case in production, are you seeing that this is helping, especially the in database functions, it's kind of chaining of different AI together that's helping move people from POC to production because now they see the ROI out of that AI.

>> Yeah. That goes a long way identifying an ROI, because then that gives them the motivation and the ability to make a case internally to go after a particular use case. Implementing AI and taking it into production has a timeline to it. We've looked at every stage of what our customers have to do to start with a POC and take it into production and looked at which part can be optimized and how can we optimize that. So if delivering a feature store so that they can reuse some of the feature engineering across different machine learning models, we want to be able to do that, right? If it's seamlessly using in-database functions, we are enabling that. We are making it super easy. We made it super easy for them to use three design patterns like we talked about earlier. They can easily bring their own model. They can use the PMML format or the ONNX format. They can now bring their large language models, use GPUs. They can even use cloud-hosted models. And then with MLOps, because we have to be able to manage the life cycle of a model, they have that ability to do that. And over the next year, we are just maniacally focused on bringing more capabilities to meet them where they are, in cloud or on-prem. We believe in multiple platform. Whatever we deliver on-prem is available on the clouds as well. And we just want to make sure that we are easing their journey and really helping them honor their data regulations, data residency requirements-

Savannah Peterson

>> Compliance, yeah....

>> compliance issues. Yeah. It's a complex ...

Savannah Peterson

>> All the sexy stuff.

Savannah Peterson

>> It's a complex landscape. Yeah. Yeah.

Savannah Peterson

>> It absolutely is. You mentioned that fervent focus for the next 12 months. I'm curious, I want to ask you a question, since you're a fabulous guest and we would certainly love to have you back on the show. What do you hope to be able to say 12 months from now that you can't yet say today?

>> I think I would be able to say that we have more customers in production with AI.

Savannah Peterson

>> Yes.

>> Right? And at Teradata, we have the world's most confidential data. We don't have it. Our customers have their most confidential data on Teradata, so it would be absolutely ... I would consider next year to be an absolutely phenomenal success, if I can say we've built some complex use cases, we've brought more customers into production with AI, with these just interesting and fabulous use cases. And then we are on a journey to get do more and more and more. That would be success to me.

Savannah Peterson

>> I love it. Well, I look forward to celebrating your success and Teradata's success next year when we're here. Meeta, thank you so much for taking the time in a busy week.

>> Thank you. Thank you for having me.

Savannah Peterson

>> Yes, our pleasure. And Rob, thank you. We're just really getting to the meat of it here today.

Rob Strechay

>> Absolutely. Digging deep.

Savannah Peterson

>> I'm excited. Strong start after lunch. And thank all of you for tuning in wherever you might be on this beautiful rock. We're in Los Angeles, California at Teradata Possible. My name's Savannah Peterson. You're watching theCUBE, the leading source for enterprise tech news.