MWC25 | Phil Kippen, Snowflake & David Sedlock, Zayo

Clips
More from MWC25

Phil Kippen

Head of Telco

Snowflake

David Sedlock

CDO

Zayo Group

play_circle_outline Maximizing AI Performance with Zayo's Direct Fiber Connections to Data Centers

play_circle_outline Scale of investment by hyperscalers and telcos in data centers

play_circle_outline Speed of innovation in telecom industry due to AI advancements

play_circle_outline Maximizing savings and revenue with AI through data harmonization for operational success

play_circle_outline Unlocking Business Transformation with Snowflake: Leveraging AI and Unifying Data Structures for Acquisitions

play_circle_outline Snowflake Joins Forces with Telcos to Enhance Network Planning and Call Center Management for Improved Customer Experience

play_circle_outline Use of Python Libraries for business rule and logic abstraction, enabling flexibility.

play_circle_outline Bringing AI Applications to Life: The Role of Small Steps and Democratized Data Systems

Info
Transcript

Phil Kippen, Snowflake & David Sedlock, Zayo

Phil Kippen

Head of Telco Snowflake

David Sedlock

CDO Zayo Group

The video presents an insightful discussion on the impact of connectivity within the artificial intelligence (AI) landscape. David Sedlock, head of strategy at Zayo, and Phil Kippen, data science lead at Snowflake, join theCUBE at MWC25 Barcelona to explore the intricate relationship between AI development and connectivity infrastructure. This discussion, led by theCUBE’s experienced analysts, delves into the foundational aspects of AI and the essential roles both connectivity and data strategies play in the evolving technology landscape.

David Sedlo... Read more

explore Keep Exploring

What aspect of the AI data center industry is often overlooked by many? add

What is the estimated amount that big three hyperscalers and Meta are collectively spending this year, and what percentage of that spending is going into connectivity components? add

What types of capabilities and acquisitions has the company been working on, and what areas have they been focusing on such as AdTech, MarTech, finance, compliance, telecom, and network operations? add

What were some of the biggest challenges faced in the project mentioned in the text? add

What strategy did you propose for integrating data from acquisitions using the TM4 model and eTOM SID form in a Snowflake environment? add

What are some examples of use cases involving network operations and call center management that have been implemented recently? add

What is a common mistake made in regards to business rule and logic when it comes to core applications, and how can it be avoided by utilizing Python Libraries? add

What approach should be taken towards implementing AI technology in order to ensure successful integration and practical application in business processes? add

bolt Powered by CUBE AI

Phil Kippen, Snowflake & David Sedlock, Zayo

search

Savannah Peterson

>> Good morning, CUBE community, and welcome back to beautiful Barcelona, Spain. We're here kicking off day three of our four days of live coverage on theCUBE. My name's Savannah Peterson, here in the cockpit with Dave Vellante. We were just high-fiving, we're so pumped up for today.

Dave Vellante

>> I feel good today.

Savannah Peterson

>> You look good today.

Dave Vellante

>> Oh, thank you.

Savannah Peterson

>> I can tell you got-

Dave Vellante

>> Ditto.

Savannah Peterson

>> Thank you. We got our spring shades on.

Dave Vellante

>> We're still awake. I still got my voice, sort of.

Savannah Peterson

>> I know. I was actually shocked about the voice thing. I was thinking about that. But you and I have been really good about hydrating.

Dave Vellante

>> Yes, we have.

Savannah Peterson

>> Which has been absolutely critical. I can't think of a better way to start off the day than with our first two guests. David and Phil, thank you so much for taking the time to join us.

David Sedlock

>> Excited about it.

Savannah Peterson

>> I know you've had a very busy week. You've been making the rounds, you've been doing a lot of talks together. You've got a really great partnership. David, I want to open up with you because you said something really fun while we were chatting before we went live. You said that you are enabling the AI mirage. What does that mean?

David Sedlock

>> Yeah, so we're at Zayo. We're the largest broadband provider independent in the United States. And we're looking at two things. Our largest customers are the hyperscalers, and the telcos, and the satellite companies across the US and Western Europe. And so when we're looking at this, everybody's talking AI, they talk around compute. And everybody's talking about NVIDIA and the servers. They're also talking about, well, the real constraint's power, and then that's also determining where those AI data centers are going to, when they're going to be built. And then the other piece that nobody's talking about is connectivity. And that's our world. Our world's connectivity. And we're bringing fiber to all those locations. And what's different, even in some of the big telco players, is we're more bespoke, in the sense we'll build directly to those data centers. So we're catching that AI windfall, or wave, that's coming from all those hyperscalers and telcos as they're building out those data centers. So what nobody's talking about is that connectivity point and that-

Savannah Peterson

>> You're absolutely right. We haven't had this conversation very much, and we've had literally thousands of conversations about this. So let's hang out there just for a second, because now I've got you... And, Phil, don't worry. I'm not going to forget about you.

Phil Kippen

>> Oh, no.

Savannah Peterson

>> Why do you think people aren't talking about connectivity or thinking about it?

David Sedlock

>> I think they take it for granted.

Savannah Peterson

>> I think that's right.

David Sedlock

>> It's assumed that connectivity is going to be there, right?

Savannah Peterson

>> Yeah, like some magical thing, like we've solved that problem forever now indefinitely. And even though we're doing different workloads and everything else, it's going to be fine.

David Sedlock

>> Yeah. You just go like, "Oh, well, yeah, we just have it." But just think about where we're having to put the data centers. They're in remote locations. They're in Nevada. They're in Idaho. They're in West Texas. And a lot of people don't want to invest, taking fiber to those locations, but that's actually what we do. And so it's the right time, right location, and we can extend that out to them. And we're seeing the headwinds. We've signed billions of dollars of deals in our funnel, and with all those hyperscalers and telcos.

Dave Vellante

>> Do you mean you see the tailwinds?

David Sedlock

>> I'm sorry?

Dave Vellante

>> You're saying the headwinds or the tailwinds?

David Sedlock

>> Well, I'm sorry, the tailwinds. I'm sorry, good correction.

Dave Vellante

>> I was like, it doesn't sound like a lot of headwinds.

Savannah Peterson

>> It's billion dollar headwinds, yeah.

Dave Vellante

>> Sounds like you're taking off, like a rocket ship.

David Sedlock

>> Yeah, we're taking advantage of it, taking advantage of that tailwind.

Dave Vellante

>> Yeah. And so between the big three hyperscalers and Meta, our estimate is they're spending about 332 billion this year. It's insane.

Savannah Peterson

>> That's a serious-

Dave Vellante

>> 332 billion....

Savannah Peterson

>> amount of capital.

Dave Vellante

>> Now, some of that goes to Amazon warehouses, but most of it is going into data centers. So of that 332 billion, what percent of that is that connectivity piece that you're talking about? Single digits? Is it even bigger than that?

David Sedlock

>> Well, I can't tell you exactly, but it's more than you probably think. It's more than you probably think. Power and computer are going to be the lion's share of it. However, I can tell you that there's hundreds of millions of dollars of deals that are coming in the doors at once. Here's the way to think of it. Before, because they've been our customers a long time, they'd be ordering 100 strands of fiber at a time. Now, they're ordering tens of thousands, or hundreds of thousands, at a time. So the order of magnitude is 10 to 20x.

Dave Vellante

>> And the telcos add to that 332 billion.

David Sedlock

>> Absolutely.

Dave Vellante

>> Not to that scale, but they spend a lot on CapEx as well.

David Sedlock

>> They do too, yeah.

Dave Vellante

>> I don't think it doubles that number, but it probably goes up to 400 billion when you add those guys in. Massive.

David Sedlock

>> It's crazy.

Dave Vellante

>> So, Phil, we were just talking, three years ago, you guys were just getting into the telco space, Snowflake, and you said that seems like a long time ago.

Phil Kippen

>> It does.

Dave Vellante

>> And you said something that was interesting. You said, no longer are we thought of as just the data company. The customers now know we understand telco. So explain the journey that you guys have been on?

Phil Kippen

>> Absolutely, and it's a great question. It seems like yesterday but, also, we've been doing a lot of work in that time. And so there's just been a consistent role of new capabilities. So we brought AI in with the Neeva acquisition. Our new CEO came from that acquisition. We've started building onto that, working on multi-agent frameworks now. We started really focused on AdTech, MarTech, finance, compliance. And when we brought telecom, when we launched telecom here at MWC, we knew that the focus would now be on things like telecoms services and network operations. I didn't exactly know how fast that would happen. And it's amazing the speed at which, not only we've brought out capabilities for telecom, but the speed at which our customers and partners are working with us on transformation. And I think that's been amazing to see. So, now, we're focusing, last year, we're focusing a lot on network operations, for instance. And then, with GenAI, accessibility, so using GenAI to help operators interact with data in a much more meaningful way. Now, we're focusing on multi-agent frameworks where you're looking cross-domain. And just the speed at which this has all happened has been absolutely amazing. I'm thinking two years in the future, and wondering where we're going to be. It's going to be wild.

Savannah Peterson

>> And you're talking about speed in an industry that is not historically known for being the fastest industry on the planet. Do you feel like the attention that AI has gotten has accelerated that desire to innovate, from your experience?

David Sedlock

>> Definitely.

Phil Kippen

>> Absolutely.

David Sedlock

>> Yeah, definitely.

Savannah Peterson

>> Yeah. I bet you're seeing it a lot.

David Sedlock

>> Oh, absolutely.

Savannah Peterson

>> And if you could wave a magic wand, what would you tell your potential customers or community to dispel the biggest myth around that? I know you've got some hot takes on what's going on.

David Sedlock

>> On what specifically?

Savannah Peterson

>> Well, just surrounding AI, right?

David Sedlock

>> Yeah.

Savannah Peterson

>> Everyone gets excited. "Oh, I want to go spend my money."

David Sedlock

>> Okay, so, yeah, I have something to say about this-

Savannah Peterson

>> I know you do....

David Sedlock

>> because I believe it. So for probably the first year, year and a quarter, I was known as Mr. No, around Zayo, around AI. Because our chairman, our CEO, came to me second day, he goes, "What are you doing on AI?" I said, "Nothing. I'm not doing anything with AI." And he looks at me, and goes, "What are you talking about?" And I said, "It's not no, never. It's no now." And what I mean by that is, we were built up through 49 acquisitions, and we had no strategy, no strategy on architecture, no on data. So I said we had to first build the foundational capabilities and the governance around that. And then, once I get the gold data sets and that's where we're at, we'll go all through that. That's where we point the AI capabilities to. And that's where we're partnering with Snowflake to help us do that. So we're at that point now where we're using all our AI capabilities, and start applying it to all of our partners and vendors that we're working with. So then, we can accelerate. And another unique thing different for us than some of our larger competitors in the telecoms space is that a lot of theirs is to cut things. Ours is, we're in a unique position because we need to go hire 300 to 500 people, but we don't want to hire 300 to 500 people. We want to hire 30 to 50, and use AI and technology to make them feel like 300 to 500 people. And so we can extend those capabilities out. And, frankly, not just because Snowflake's here, is Snowflake's our key partner in establishing that governance and that foundational gold data set, so then we can go and accelerate into that, and that-

Savannah Peterson

>> Which is critical. It's table states.

David Sedlock

>> Which is critical. And that's also something that's overlooked. And you think about all the AI hype, and everyone goes, "AI, AI." And we're going, "It means nothing unless you can have it pointing to high quality data in a location that you can then take advantage of."

Dave Vellante

>> So I like the answer to your CEO because it was controversial, but I also could put a different spin on it. Because you could have said, "I'll tell you what I'm doing. I'm doing the most important thing for AI, which is I'm harmonizing our data and consolidating all these stovepipes because that's the first step before I apply AI." And that's exactly what everybody should be doing. Otherwise, it's a waste of time.

David Sedlock

>> I followed on, I wanted to have a little bit of the shock factor.

Dave Vellante

>> It was great. I like it, yeah.

David Sedlock

>> So it got his attention, and it allowed it. Now, the other thing that people are missing here too, we were talking about this yesterday, is along with doing that foundational stuff, I came from a place where I had to monetize things very quickly. So I initiated what's called data and AI quick strikes. So I've been using, for the last two years, a small team, it's four or five people, that we said we're going to focus on the biggest problems we have operationally, data, churn, revenue, whatever it may be.

Savannah Peterson

>> Biggest problems everybody has pretty much, frankly, yeah.

David Sedlock

>> Yeah, right.

Phil Kippen

>> They're big problems.

David Sedlock

>> And say, "How can we save money? How can we save money?" And so I would start and say... I'd have to go up to the board every month and tell them what we're doing. And then I'd turn around and go, "I'm doing this and this." And you get a little glazed eyes, and you go, "But I saved $3 million this month." And they go, "Oh, okay." And it makes it a lot easier.

Savannah Peterson

>> Now, I'm listening, yeah.

David Sedlock

>> And, now, it makes it a lot easier to say, "I'm going to have to continue to invest to build those foundations."

Dave Vellante

>> So can we unpack that business case for... You've mentioned four things, operational, data, churn, and revenue. And I want to understand how they contribute to monetization. Operational means you're more efficient so you're saving money. Data, obviously, you're preparing for AI. Is there a revenue generation piece there, or is that the churn piece? Those are related.

David Sedlock

>> Well, it's both. So our biggest challenges were operational aspects of it. First one I did was churn because it's the easiest. I just said, "I'm going to unleash one data science, go work with the churn team, see what we can find out. What can we do?" And within six weeks, we made some modifications to what they were working on, built a machine learning model, and then we saved millions of dollars in six weeks, just by looking at it. We did the same thing on revenue assurance. We had a customer, everybody has it, you have leakage in your revenue. So you're billing services, and you're going like, "Hey, are we billing for everything?" And I was sitting next to our CFO, because I'm working with him hand in hand, and saying, "Where should I focus these high quality resources for data scientists?" And we said, "I have a customer of mine, a friend of mine, that's a customer. We set up a service, but he is going like, 'I haven't received a bill yet.'" And we're going like, "Well, why don't we go look at... Let's just build a model, use..." Again, it's machine learning start, and then we went in predictive modeling. And say, "Are we billing for everything?" Millions of dollars that we found that we weren't billing for. And so it's starting there, and then you can use that to then accelerate to other stuff.

Dave Vellante

>> So you got the below the line, which is the operational cost savings, then you got above the line, which involves, if I understand it correctly, your churn is a function of your ability to analyze the data. And that drives revenue because you have lower churn. But I'm trying to understand your business, because I'm imagining just massive deals with hyperscalers. But, now, you're talking about churn in the weeds.

David Sedlock

>> So don't think of churn as if you have a hyperscaler and they churn. It's churn within their different... What they're picking and selecting, so the different... They may say, "Hey, well, we're not getting what we need out of this area." And they'll churn off. So it's a subcomponent of churn. Now, where we're going now, and I think this is where you're hitting on, is around the front end of the engine. And this is where we're starting to get, and really using AI to say, "How can we look at the sales funnel, the sales channel? How can we be more efficient? How can we predict where people are going to want more and new services from us, or add-ons to us?" That's where we're going now. We're just getting to that foundation where we can get there.

Dave Vellante

>> And how long have you been at Zayo?

David Sedlock

>> Two years.

Dave Vellante

>> And then did you bring in Snowflake?

David Sedlock

>> Snowflake was chosen right before I got here.

Dave Vellante

>> Okay, why? And then, what's that journey look like?

David Sedlock

>> Yeah, so I used Snowflake in my previous world.

Dave Vellante

>> Which was what kind of-

David Sedlock

>> I was at AT&T in the chief data group at AT&T. And it goes way back to '13, but we were a big Snowflake users.

Dave Vellante

>> Wow, '13. That's early days.

David Sedlock

>> Oh, we started as a big data meme. Well, yeah, we started up the big data group, myself and four others, at AT&T as a start-up business unit. And we started and grew that to end up being 2,000 or 3,000 people over 10 years. So I was part of that journey, that's a different one.

Dave Vellante

>> So you're familiar with Snowflake and the value.

David Sedlock

>> Yeah, and the value. And so, here was my second day on the job, I went directly into the board meeting. And they said, "Well, what do you think?" And I'm going, "Okay, I could probably get fired the second day on the job." And I just said, "Here's the deal. It's not just about me coming in." I go, "I can tell you what to do. I've built this before, and I can build that. But it's also around a business transformation. You have to change the way you do business for inputs to go into the source systems to then build it, and then extend and build its foundational capabilities." I can build the Ferrari, but if you have crude oil going through the Ferrari, it's going to blow up. It's not going to work. You have to go extend that. And what I said about Snowflake, I go, "Here's the beauty around it." Because I had a good experience with them. Is I said, "Because we've had all these acquisitions, and we're likely going to have other acquisitions, what you do is, you use the TM4 model for your data structure, you use the eTOM SID form for how you line up the data. And then you can say, if you ever go either to a merger or other people you've already acquired, 'Hey, just put the data in that format, and then have it sitting on top of Snowflake. Put in a Snowflake environment. And then when day one comes around, or if we want to integrate, we can do it day one, Bam.'" It makes it easy because it's agnostic, but it's also right there in the cloud so we can actually leverage. And that's where they've been a huge partner of ours.

Savannah Peterson

>> And it gives that unified experience-

David Sedlock

>> It gives a unified experience from day one....

Savannah Peterson

>> across the data, whether that's-

Phil Kippen

>> All three hyperscalers, yeah. Same platform, same operational model. It makes a lot of sense from an operational efficiency perspective.

Dave Vellante

>> So you do that as part of your due diligence, is that... You go into their data room, and say, "Okay, guys, fill this out. We can what if what it looks like."

David Sedlock

>> Well, that's what I said, I said-

Dave Vellante

>> That's awesome if you can do that.

David Sedlock

>> Well, what I said to them is that, if and when we ever have another acquisition, or we had just acquired some other companies, and I said, "This is how you do it. You come up with a common data model, you do that. And then you put it right here together. And day one, Susan flips, you can have standardized reporting off of common defined industry standard protocols and data models."

Dave Vellante

>> So, Phil, you were talking earlier about you're surprised how fast it was. In some ways, I'm not as surprised, only because you're not relying on the telcos to transform their fossilized stack, your data. And I'm sure, like every company, their data is a mess. And so you can help-

Savannah Peterson

>> A siloed mess, probably.

Dave Vellante

>> Yeah, you can help clean that up. They don't have to say, "Well, we're going from closed RAN to open RAN. You don't have to worry about that hardware stuff." And so I wonder if you could just talk about the progress you've made in telco. Maybe you can name names. If you can, great, or just describe the types of organizations, the use cases. What are they doing with Snowflake?

Phil Kippen

>> So I mentioned we moved into operations last year, network operations. So we've been working with network planning, network engineering, even working with call center, for instance. So when you think about network planning, their job is to get the most out of the assets that are being deployed. So we've done a few use cases with a couple of tier 1s on understanding the capital expenditures, and optimizing that with business intent. So when you're making an investment to upgrade the network, for instance, there's a business intent, obviously, that needs to stay aligned, but that changes over time. Do you want to build more availability into your networks? Do you want to bring in revenue faster and go where there's more demand? So where you place that investment is critical. Those use cases we've been doing quite a bit of, which is really the business side of the network. Then from a call center management perspective, we've been doing some really interesting things. We've been working with Amdocs, and as well with T-Mobile, on a model where agents, for instance, as somebody's interacting with a chatbot, the information on network, service, billing, all of that is being integrated into the chatbot. So we're cutting down the time that that person, that consumer, interacts with that chatbot, which is raising customer experience. So broad range of things across the board. We're also still very much working with clean rooms. And, as well, we've been doing some real-time RAN visibility and optimization, where you put towers. Two years ago, I didn't think we'd be that far into it at this point.

Dave Vellante

>> Frank Slootman once said to me, "I like you guys, you're not gotcha media." And I'm like, "Yeah, that's true. We're not gotcha media." At the same time, we are media, and we like to talk about fun things in the industry. So the-

Savannah Peterson

>> Can't wait to see where you go next, Dave, can't wait.

Dave Vellante

>> No, I know, look out for this one. The war between Nadella and Benioff. So Nadella and Benioff are sniping at each other. Benioff calls Microsoft Copilot Clippy. Nadella says, "You're not going to need SaaS in the future. You're just going to have agents talking to CRUD databases." So it's fun. But there's some truth in each. It's interesting. So we've unpacked that. Where I'm going with this is, last summer, we saw two industry titans in the data world, Databricks and Snowflake, challenged each other on things like governance, things like open table formats, and it created a lot of confusion in the industry. Snowflake open sourced Polaris. Databricks, on stage, open sourced Unity. There really wasn't much code in the GitHub when they open sourced it, but it was still a shot across the barrel. They bought Tabular, which created Iceberg, and it created a lot of confusion in the industry. My question, David, is, how do you think about all this stuff? What are your thoughts on open table formats? It sounds great, but you got to govern it. What's your governance strategy? Can you paint a picture of how you're thinking about this whole issue?

David Sedlock

>> On the governance side?

Dave Vellante

>> The whole open table formats, is it good thing? Is it something you're leaning into? If so, how are you governing it? What's your overall strategy there?

David Sedlock

>> Yeah. Well, I'll tell you that I need to develop it more. That's the first thing. What we've really been focused on and, again, it's partnering with Snowflake, is how do we get that governance internally and operationally done, and get that... That's what we've really been focused on. And, now, we're just getting on the backside of it, and saying, "How can we take advantage of that?" So I don't have a strong position on it. I've used Databricks in the past too, and Snowflake at the same time. So I don't have a strong position.

Dave Vellante

>> But it's not an issue of Snowflake v Databricks so much. It's more the philosophy. Let me ask it this way. How concerned are you with lock-in? I ask executives this question all the time. And some say, "Oh, I'm concerned about lock-in." Others say, "You know what? I don't care. I get value out of." So your choice would be, I'm going to stay in the Snowflake environment, and I'm going to use Horizon, and maybe I'll use Polaris for my technical metadata. And I'm going to use Horizon for my operational and business metadata. And I'm comfortable with that. And I don't really care that it's closed. Everything's closed to some degree, if that makes sense.

David Sedlock

>> No, that's better. So here's what we've done. One thing I've learned, and we're talking about lock-in, and what I've seen around this, and my experiences back at AT&T and otherwise, is one of the biggest mistakes that I've seen made is that your business rule and logic is baked into some core application. Whether it be Snowflake, Databricks, whatever, or otherwise. So the decision that I made on that is, I put that in all of our business logic rule in dbt Labs. And it's not dbt itself. What I cared about is Python Library. So I'm putting everything in Python Libraries. So what the most nuanced and custom is my business rule and logic that I use to run my business. And so I contain that all in Python Library. So if I want to switch then applications, whether it's Snowflake switching out to other places or otherwise, I take all my core knowledge and how I run my business, and I just take my Python Libraries and reapply it someplace else. Because the way I do it is, I have source data coming in. We have standardized ingestion patterns. I have identified critical data elements, which are the top... We have 819 of them. Then said, if we focus on these things, we're going to run 95% of our business. We keep that, combine with our business logic and rules, and create what's called business-ready data sets. But the rules are the secret sauce of how I bring all that stuff together.

Dave Vellante

>> So you're not moving the business logic from the app into dbt and five training it in or anything like that. It lives there. Your applications access that, what, through APIs?

David Sedlock

>> Yeah, through APIs.

Dave Vellante

>> Interesting. So you harmonize that data through dbt?

David Sedlock

>> Well, yes. And what we do is, we then build what's called business-ready data sets. So it's code that says, "Here's the source data. Here's the critical data elements. Here's how we apply our logic to those critical data elements." And then we publish the data in gold data sets, and then we consume the gold data sets off of data science tools. We use Tableau, we use Power BI, DataIQ, and we consume that. And that's what all of our business should be using.

Dave Vellante

>> Interesting. So I'm sorry, I'm geeking out here.

Savannah Peterson

>> You're having a moment. We're going to let you have it. Let's go.

Dave Vellante

>> So you chose to do that, as opposed to, for instance, doing it through Salesforce with their data cloud, which you could have done.

David Sedlock

>> Yeah, which we don't want to do.

Dave Vellante

>> Right, okay. And so that-

David Sedlock

>> And there's a reason why for that.

Dave Vellante

>> Yeah, well, explain it?

Savannah Peterson

>> I was actually just going to ask.

David Sedlock

>> So the reason is, it goes back to the 49 acquisitions. 49 acquisitions, we had a single stack. We put everything, everything, from identification of a potential customer to billing to network to everything. So we literally customized everything. We had 66 open text fields that managed our business, very bad, okay?

Savannah Peterson

>> Wow.

Dave Vellante

>> Yeah, wow.

David Sedlock

>> So we're unbundling all that. We're unbundling all that, and we're saying, we're implementing ServiceNow, Oracle Fusion, NetworkManager. So we're decoupling all that stuff, shrinking down Salesforce to what it's really good at. And then, for me, I'm looking at the analytic layer. So what I care about is not all the data elements, I care about the most important ones. I'm really focused on the eighteen or nineteen.

Dave Vellante

>> So you got metadata as well?

David Sedlock

>> Yeah.

Dave Vellante

>> Metadata, the business logic. What about the process knowledge? And have you looked at relational AI, and maybe building a knowledge graph as part of all this?

David Sedlock

>> Well, that's coming next. We have the capability to do the knowledge graph off of what these business-ready data sets... What these business-ready data sets is, basically, how we're running all of our technical and business together, and then how we're using it. So it's almost like the roadmap. We're also using profiling all of our applications through... How can I just have forgot our vendor on it? But we're profiling all the applications, pulling all that technical metadata in, and then also the business metadata associated with that, and defining those. When I call CDEs, that's the business metadata, but I also have all the technical metadata and lineage of how the data's passing through.

Dave Vellante

>> And that harmonization, let's say, in theory, could occur through relational AI.

David Sedlock

>> Yes.

Dave Vellante

>> And then you feed that up in the future to agents.

David Sedlock

>> Right.

Dave Vellante

>> And, now, instead of having hard-coded business rules and business logic, you can actually create more malleable workflows that can change as the market conditions change.

David Sedlock

>> Yeah, and that's also, Phil and I were just literally just talking about this on the cab ride over here, is we're at that peaking point, where we're going to be looking with our strategic partners to say, "We can't have everybody as being a strategic partner. We have to pick." Obviously, Snowflake is one of ours. How do we take advantage of what they have built, and what they have? I'm not going to go build all that stuff. I need to leverage our partners for that. But, again, I abstracted the core of our business out into those libraries so that I can apply it and use it, and then scale it. And then if I need... Because we don't know who's going to be the AI winners.

Dave Vellante

>> Right. But all of this is going to work better if it's all done inside Snowflake, and it's governed and it's secure.

David Sedlock

>> Yes, that's right.

Dave Vellante

>> But, to our earlier point, it's all inside Snowflake. So what's more important to you, that it all works and it's governed, or that it's open?

David Sedlock

>> I would say, it all works and it's governed.

Dave Vellante

>> That's a very viable strategy, and it's one we think about all the time. It's an age-old industry debate. Do I buy best of breed, or do I buy consolidated? And there's trade-offs of doing that. Thank you for indulging me in that little diversion.

Savannah Peterson

>> Hey, it's fun to watch you learn. I get to learn too, obviously.

Dave Vellante

>> But it's a real critical issue that every customer faces. How am I going to take advantage of this AI-agentic future, and what's my architecture look like? How am I going to harmonize the data? How am I going to apply agents and trust them?

Savannah Peterson

>> Yeah. And, honestly, Dave, the questions that you were just asking there are the exact questions I bet a lot of your customers are asking and thinking about, as they play together their strategy.

Phil Kippen

>> Absolutely. Yeah, there's no single answer. Customers are going to have different architectures, different goals, transformation. They're going to be at different stages. We do, we work very closely, for instance, with AT&T, it's public knowledge, on the common compute layer with Iceberg. We're a huge proponent of open. It's something that we spend a lot of time and energy making sure that we do have, at the right integration points, open capabilities for interoperability. And so for those customers who are looking for open, and want the ability to standardize on that open architecture, bring in whoever they want, remove whoever they want, make it simple, that's okay. It's okay with us. It also gives us an opportunity to solve for more issues that we otherwise wouldn't be able to. If we can be inserted into an open architecture, now take advantage of that.

Dave Vellante

>> Well, and Snowflake did something that was very clever. They said, "We're going to open source Polaris," which I call the technical metadata. Benoit will correct me, but that's at least my understanding. But the business and operational metadata stays inside of Snowflake. Snowflake could choose to extend that. Because that's what Databricks did, said, "Okay, we're going to open source Unity," which, again, was very lightweight in the GitHub, as a shot across the bow. Snowflake could do the same thing. They could open up Horizon if it wanted to. But that sacrifices some of the quality, in my opinion anyway, because it's Wild, Wild West of open source. And you're doing it yourself, and you're rolling your own. And those are really interesting and hard trade-offs for customers.

Phil Kippen

>> I think what's really interesting is who owns the metadata and where that resides. That's the problem we're trying to solve now. Which is, if, for instance, we own the metadata, Databricks, we want them to have visibility on it. We do collaborate. We really do.

Dave Vellante

>> I know you do, yeah.

Phil Kippen

>> Exactly. And so the same thing applies in the other direction. So that's the one sticker right now that we really need to find a solution for, which is, how do we extend visibility, and also extend metadata, without having to completely be a part of that process. And I think, over time, we'll continue to work together to make that happen. But we are extremely cooperative in that type of model where we've got multiple catalogs, or we've got one catalog model, open architectures, specifically around things like Iceberg. We do collaborate and cooperate.

Dave Vellante

>> But what David's done, if I understand it, is you've chosen the dbt layer for that, instead of just API-ifying the metrics. You're doing that with the process knowledge and the metadata.

David Sedlock

>> Yeah, I wanted to have it abstracted. So then, even though that we're making the hard calls, I want to abstract it so I have choices in the future if I need to. Because what I saw on my legacy telco world, people had hard-coded all those business logic rules inside 10 different systems. And then you'd go, "Well, where's the master? Well, how do we know? How do we change it?"

Dave Vellante

>> And they're fossilized.

David Sedlock

>> And then you're fossilized.

Dave Vellante

>> But to Phil's point about who owns the metadata, you've decided that you own the metadata, which is awesome. It's a great strategy because that gives you flexibility to know what... You can't predict exactly what's going to happen in the market. It's really uncertain.

David Sedlock

>> And we want to be able to take advantage of... We don't know who's going to win in all this AI race game, in the sense of all the nuances. We've made some strategic bets on a couple of players, Snowflake being one of them, and DataIQ being one of them. But the other ones, we're going, "We're going to let this play out and then see where it goes." And we want to try it out, but we have to have it for a reason. And so this is... Phil and I talked about this three times yesterday. It's about if I'm not doing stuff to help enable the business and bring value back to the business through some of the things, saving money or cost efficiency, then I really shouldn't be doing it. I shouldn't be doing that stuff. It's cool tech, but if it doesn't bring back the value-

Savannah Peterson

>> The bright, shiny thing is not necessarily the solution, nor is it a value add, nor does it make the world a better place, right?

David Sedlock

>> Yeah.

Savannah Peterson

>> There's a lot of different directions to go there.

Dave Vellante

>> Well, it's like when you first met with the board, yeah, okay, great, you got Snowflake, but let's start with the business and work back from there.

Savannah Peterson

>> Yeah, here's all the other things we need to do right now.

David Sedlock

>> It's great that we have it because I knew as soon as I found that out, I knew I had the anchor. I knew I had the anchor that's cloud-based data warehouse, that has the flexibility that's on the cutting edge of things of where we're going. I knew I had that anchor so I didn't have to worry about it, so now I can work about everything else. And so that was very comforting for me, coming in, to know that, and have that. And they've been a great partners of ours from the get-go.

Savannah Peterson

>> It makes a lot of sense. I can see why this is a nice match made in heaven. And it gives you both strength and confidence when you're upselling, and upselling out to the largest companies on earth, quite frankly.

Phil Kippen

>> Absolutely.

Savannah Peterson

>> I have one final question for you, because you've both referenced navigating the future, things moving fast. This has been such a great conversation. When we're at Mobile World Congress next year, what do you hope to be able to say then that you can't say now? Phil, I'm going to start with you.

Phil Kippen

>> I think solving problems with respect... And I don't want to take this away from David, but solving problems with respect to communication and multi-agent frameworks, right?

David Sedlock

>> Yeah.

Phil Kippen

>> I think the way agents communicate the workflows and how, I think there's a lot that we could do there when you think about the complexity of a telecom environment. The integration aspect with legacy systems is not going to be easy. And so, next year, I'm really hoping that we'll see multi-agent frameworks and more on that, but solving the critical problems in real-world production scenarios, where there has to be integration in multiple layers, systems. And I think that's what I'd like to see, certainly next year.

David Sedlock

>> Mine would probably be, simply, practical application. Practical application.

Savannah Peterson

>> Louder for the folks in the back.

David Sedlock

>> Yeah. It's one of those things where... I came from the business side, not the technical side, and so that's where I always go. And I'm just going, great around all the AI hype, and, yes, I think it's cool, and, yes, we're going to end up doing it, but let's take small steps now. Don't go for the big home run. Take steps, get wins, drive it out, and extend. Get it out to the business side, get them engaged, and have democratized data systems out in the business, creating ideas that a person like me doesn't think about. That's what I'd like to see. I'd like to see less hype and more practical application and real life examples of what it is.

Savannah Peterson

>> I love that. That is a perfect note to end on, David. And hopefully everyone-

Phil Kippen

>> Yeah, it's more about the ride than the destination, right?

David Sedlock

>> Yeah. Yep, it's totally true.

Savannah Peterson

>> Definitely.

Phil Kippen

>> Thank you, guys.

David Sedlock

>> Awesome.

Savannah Peterson

>> Yeah, thank you both very much. This has been awesome. And thank you, Dave, lots of great questioning today to start us off here. I think this was our longest interview of the show so far.

Dave Vellante

>> I think so, that or Sheree.

Savannah Peterson

>> Yeah, yeah, yeah, yeah, yeah. I guess we always start the day with a really interesting conversation, not to diminish any of our other interesting conversations. Anyway, I hope you all have enjoyed watching this, wherever you might be. We're here in Barcelona, Spain, beginning of day three of Mobile World Congress. My name's Savannah Peterson. You're watching theCUBE, the leading source for enterprise tech news.