Governance as a Growth Engine

In this episode of Unleashing Genius, NetApp Chief Marketing Officer Gabie Boko is joined by Thomas “T-Rob” Robinson, CEO of Domino Data Lab, for a conversation about one of the most persistent challenges in enterprise AI: moving from promising pilots to systems the business can trust and run every day.

T-Rob explains why many AI initiatives struggle long before they reach production and why the underlying challenge often has less to do with models than with data, governance, and operational processes. The discussion covers the hidden cost of data scientist time, the growing importance of bringing AI to the data, and how organizations can build trust into their systems from the start rather than layering it on later.

Gabie and T-Rob also discuss AI agents, reproducibility, data sovereignty, and why the most effective organizations treat their data infrastructure as a strategic advantage rather than operational overhead.

Continue the conversation at NetApp INSIGHT 2026, September 29-October 1 at the MGM Grand in Las Vegas, where customers, partners, and technology leaders will come together to discuss what's next across AI, data, cloud, and cyber resilience.

Hosted By: Gabie Boko

Produced By: Kenya Hayes

Transcript

Copy Transcript Copied

SPEAKER_00 0:05

I'm Gaby Boko, the Chief Marketing Officer at NetApp. Welcome back if you've been here before, and if you're new, welcome to Unleashing Genius. This series is a set of executive conversations about what's changing across industries as AI, data, and shifting business demands reshape how organizations are running. In each episode, and today I'm really excited about, we're going to zoom out and look for patterns that leaders are running into right now and where the pressure is building, maybe what assumptions are breaking and what it takes to respond in a way that holds up over time. A lot of people talk about what's possible. We all know that. This series is about the shift from possibility to execution and what it takes to turn new ideas into systems and workflows that keep your company and you as a person performing at scale. We're going to use customer experience as the lens because that's usually easier, but the goal obviously is a broader takeaway, what leaders should learn and what they should do differently as these shifts continue to accelerate. So, gonna get in today. So today's conversation goes straight at one of the most stubborn numbers in enterprise AI, and that's really AI pilots, because most of them aren't ever actually making it to full production. And that's usually not because the models are bad. It's because the systems underneath them can't actually handle what the business needs, which is strong oversight, results you can repeat, and speed, of course, and everything all at once, right? Thomas Robinson and PS, I've been given permission to call him by a very, very personal name. So I feel that we're friends now already. T Rob has spent his career on the operating side of that problem. And I'm excited that he is going to spend time with us today. Early in his career, he spent time building infrastructure for a demanding technical environment and was recently appointed, congratulations, as the CEO of Domino Data Lab, where he helps some of the most regulated companies in the world turn early AI experiments into systems that they can trust and run every day. So this conversation, as we talked at the top, I hope is going to be about a mindset shift, what changes when a company stops treating AI systems as an afterthought and when it starts treating them like the product. Welcome, T Rob. Thank you for joining us. I'm super glad you're here.

SPEAKER_01 2:25

Thank you so much for having me. I really appreciate it.

SPEAKER_00 2:28

So let's just go right in, right? So I'm going to start with a non-provocative question. I'm going to start with a realistic question. Industry research is telling us that something like 85% of AI projects are never actually getting past that pilot stage. That's how I let in the this podcast. But from where you sit at domino now, what's actually going wrong?

SPEAKER_01 2:49

Yeah. I mean, the first thing that I would say about this is everybody should settle down. And I am I I am not at not surprised at all. In fact, I'm somewhat surprised that it's 85%. I've heard other numbers that are that are higher for failure rates. But um I didn't want to answer this question in in just getting into um exactly what happens in the enterprise. I I want to frame a little bit first because I think um I, you know, I want to I want to bridge from why I'm not surprised to why uh everybody you know potentially is really surprised. So in this uh for me who's had experience in working with uh data scientists for about a decade and understanding what it takes to actually have um hardened, durable models at the core of the business, um, failure is a huge part of that equation. And anybody who's been attached to any scientific endeavor or any research endeavor completely understands that uh failure happens all the time. And so, in in that sort of sense, it's not at all surprising. A lot of what data scientists and researchers have done historically is disprove hypothesis that's as equally as valuable and important to proving something out. So that that's kind of you know one lens on the on the situation. And you have this other lens, which is generative AI has suddenly made AI a kitchen table topic. And I think everyone is sort of wowed by their first experiences using generative AI to like write that email they were not wanting to write or you know, read through that earnings report or, you know, whatever it might be, right? And and so there's a gap between the expectations of the individual experience and the hardened reality you need in an enterprise. And, you know, that's really where this number comes from. And and again, you know, I see these headlines and not at all surprising. That is the challenging uh thing. And as we go from AI really being a personal productivity tool, you know, like every enterprise has deployed it to end users into something that actually runs in core business processes, um, there's a really different expectation of what it takes to uh harden those models, harden the harnesses uh that are used uh to actually make real business decisions that impact your um revenue and impact your bottom line.

SPEAKER_00 5:25

You know, so I I heard that, I hear you say, and I'm gonna paraphrase, we have to find a way to normalize failure as part of uh and and de-stigmatize it potentially as part of what we're doing here, right? The what AI is helping us do is to learn, as in any scientific and technology and endeavor. I really appreciate you calling that out. I think the second thing that you're really kind of digging into is that that failure becomes pronounced if you're not really kind of leaning into why you're doing the the experimentation in the first place, right? That's what you said. So both of those two things together are, hey, listen, we get we got big challenges, but right now they're not scaring you.

SPEAKER_01 6:08

No, not at all. And I'd I'd put one more lens on it as well, which is um I think we're also in a hammer-to-nail situation where uh a lot of people are experiencing generative AI and using that as a tool or driving projects where generative AI is the is the vehicle they're trying to get to uh uh to improving a business process. And and our take at Domino is um you need all of AI. You need statistical models, as boring and as silly as they are. I love to remind people that the Federal Reserve sets interest rates based on kilobyte data sets and uh statistical models. You need machine learning models, you need specialty models like computer vision. All of these things need to come together into AI systems, and and each of those models has a difference in its variability, its reasoning versus its sort of govern parameters, and and building those things together into more complex systems, I think is the future of uh of how we deploy at scale.

SPEAKER_00 7:11

So basically don't run your stuff in isolation, right? That's that's what you're saying.

unknown 7:15

Yeah.

SPEAKER_01 7:16

Exactly. Exactly. And and write tool for the job.

SPEAKER_00 7:18

Exactly. So let's talk a little bit about, let's go dig in on that isolation problem, right? I think um companies are obviously moving a lot of their data to AI, right? They're saying, this is all my stuff, go turn it into something, right? And most of the time, most of the time, they're doing it in the cloud, right? Um, which is uh un understandable, technically low cost. Um, but you're pushing for the opposite, right? You believe that, and correct me if I'm wrong, you're saying if we bring the AI to where your data actually lives, um, then that could be far more impactful for you. Can you tell me why that matters so much right now?

SPEAKER_01 7:58

Uh let me tell you, it would be a really different answer if we were talking two months ago. Um because yeah, I I think I think there's a new um I I love when I feel right when my predictions are sort of played out. And I and I feel like uh I've I've been made right on this one. I mean, the new the new biggest concern for a lot of enterprises is sovereignty.

SPEAKER_00 8:21

Um, you know, I knew you were I was thinking that was gonna come later, but I love that it happens here, right now.

SPEAKER_01 8:27

Yeah. I mean, so you know, whereas the answer uh a while ago would have been a little more simple around cost and data mobility and like ri really it's sort of um just given the the raw physics of how AI works and how data and storage need to be attached into GPUs, there's a lot of like um sort of base case things that make it very sensible to put your AI to your data, not necessarily your data to your cloud, if you will. But sovereignty has sort of stolen the scene, and I think lots of organizations are now thinking about uh risk in the sense of moving their data to a shared infrastructure or pumping it, if you will, into a foundation model provider. And again, this, you know, we talked in in sort of the the first part about um the failure of uh uh of AI experiments into production. I mean, this this is like where the risk ramps as well, because it's really different when you're you're dealing with um email synopsis and our project plan for this thing or that thing, versus um patient data in a clinical trial, or versus um your early stage research pharmaceutical pipeline, or um your core customer data, you know, all of these things, the risk ramps at the same time. And that's a big part of um, you know, why uh sovereignty has become much more of a concern for folks.

SPEAKER_00 9:53

Yeah, I love that. When we talk sovereignty, when we talk um data, what's becoming increasingly interesting is that we are looking at this uh through the lens of are we applying old habits of how we want to deal with our data to something that is completely new and we're experiencing the risk or the even the potential security gaps by applying old thinking to a new methodology. Is that kind of why this is becoming such a uh an issue, do you think?

SPEAKER_01 10:25

I think that's certainly a part of it. There's a a paradigm breaking here. Um and uh I mean, in in part, it's because so many folks have warmed up. Uh here's here's the era we went through, right? Like we started to have the AI thing, and then basically enterprises started to warm up all their data in anticipation of I'm gonna translate data into value through the funnel of AI, right? And then and then now the the other shoe is sort of dropping and we're kind of in the spot of, oh boy, well, who's actually now looking at my data and what data should I actually put in there? And I think I think that's part of what's shifting it. We're moving from sort of excitement to reality. And and I don't I don't mean that it should be sobering. Like I just want to be really clear. I'm incredibly excited about the power of AI, but you know, now we're into a realm when we're trying to uh when we're trying to do it in the enterprise with customer data and and and uh you know sort of important competitive uh data, you know, now there's a different set of concerns. And I I think that's really what's driving a bit of the change.

SPEAKER_00 11:28

What I'm paraphrasing again, what you say, is like, listen, there's risk either way, but don't create more risk by not thinking about alternative ways to deal with it.

SPEAKER_01 11:38

Right.

SPEAKER_00 11:38

Exactly. Yeah. So you've talked, we talked a little bit about sovereignty a little earlier than what I thought, but I love the fact that we connected it to that. But this also brings in when you talk about kind of protecting that data and thinking about your data, there is um there is a cost to AI and and it's not just the compute, right? It's not just the power to do the compute, it's actually people's time, it's the data, it's it's um how maybe your data science team is rethinking access to data, security of data, even their time with the data. Uh walk us through what this looks like. Like, what does the data science team of the future look like taking all these new things in? Are they inventors of the future, bringing that back to an organization who might be 50 years behind them? Quote unquote. I mean, they might not be, or are they optimizers of the present to position to the for the future?

SPEAKER_01 12:34

First of all, let's let's talk about the work of a data scientist in particular, because that's being very much disrupted by AI. And there are um parts that are going away, there are parts that are becoming more important, and there are new jobs to be done, right? There's a there's a real big shift here. And and I will say data scientists sort of parallels a little bit of software engineering. You know, we talk a lot about how the coding assistance is a killer use case for generative AI, and it's sort of changing how software engineering works. I mean, a similar thing is playing out with data science, and um uh really it's changing the role of a data scientist. So here's a couple of things that are going on. First, what's going away is like coding. Um, a big part of a data scientist's job is like going from uh the science to something that is computational, right? So writing Python code or R code or different things like that. And man, a lot of that is being compressed. And we say, we see tools like Auto ML and and drag and drop tools going away because suddenly everybody can just write a lot of code. Now, that creates a dynamic where something that's growing or changing, you know, if you sort of broke down the work of what a data scientist is doing, now there's more time in specing out exactly how they want to do work. So so the care, the care that they used to need to provide in the literal coding is now moving to how do you set up an experiment? How do you understand what we're actually trying to accomplish? How do you guide the AI to make sure it understands statistics, which everybody knows generative AI is really not good at like logic and statistics and that sort of thing. It it you know, we're getting better, but that's still that craft has become an important new thing that data scientists do. Lastly, the biggest thing that's really changing, and and it again goes back to the notion of prototype to production, we see the role of a data scientist changing. It used to be a function where you would do analysis, you would create a model. I mean, and and some of the when I say model, this can be like life's work, this can be a new drug, this can be deep financial analysis. So important things. But really, the boundary was that sort of stopped with the data science team and either got sort of handed off to IT or something like that. Um, the new superpower with generative AI that data scientists have is they can build enterprise applications that leverage their models. And so the space between a data scientist and a business user is compressing to zero. And um, you know, you're not building those and hosting on your laptop. You still need an enterprise foundation. That's a big part of what we do. We help, you know, give them the platform to do that in the in the context of regulation and governance and and access control and audit log and all those things that nobody wants to think about. Um, but but that's really how we see this kind of changing. And so, you know, to to kind of put it back to your question, I see the role of a data scientist changing and thus how they interact in the enterprise is is changing materially.

SPEAKER_00 15:43

I love that you said that. Is the data scientist based on that almost becoming shadow engineering or shadow IT based on that skill set? And does that is that materially changing the role of an engineer or changing, materially changing the role of IT?

SPEAKER_01 16:00

Yeah, I mean, I I love the term shadow IT. I mean, it's it's a um, let me say it's a double-edged sword, right? Like from an IT perspective, understood completely, CIO or CISO hat, uh, shadow is bad and there's risk, and often at times folks on the shadow business side, whatever, don't understand why the context, et cetera, right? And so I understand that side. But I really feel um, you know, the other half of it too is often an IT team is very busy with large projects, um, focused on massive platforms, their cloud, their data architecture, building new ETL pipelines, like like all these heavy things. And, you know, the the the thing that's needed is the business users need to be have self-service um and and be able to build things. And so there is a beauty to uh the shadow IT that's valuable for a business. Like as a CEO, I want a lot of shadow IT going on, um, but I also want you know balance and care to sort of risk. So it's it's I think it's a hard thing to balance. I will um shamelessly plug. I mean, this is the world we've always tried to live in. Our core belief is you need to um be able to move fast, but also do that on uh stable infrastructure and rails. And so we've always thought about actually when you do those two things together, that's where you get your best work, both innovative but but lower risk.

SPEAKER_00 17:31

I I think that that's the right answer. I appreciate you calling that out. You know you know, and that kind of drops into kind of that second kind of portion of that conversation with how do you how do you build trust and policy into the infrastructure so that you're not necessarily doing it on the back end, which policy and governance on the back end to manage changing roles, to manage speed, to manage all of those things actually creates a level of friction for the data that results in violation versus bringing bringing it back in. So how do you how do you think about building those types of things? It doesn't not looking to restrict, but looking to invite, maybe in time. Absolutely into in the infrastructure and the platform.

SPEAKER_01 18:15

Yeah, absolutely. And um five years ago uh as a company, we were focused on helping data scientists get access to infrastructure and data. Like we we always do surveys and we heard the top concerns are those sorts of things. Um we saw a new trend emerging a few years ago, which was hey, I've I've got a great place to work, but now the biggest impediment to my work actually having value to my organization is going from something I built to it being in production. So we dug in, we thought that was very interesting. What's actually going on? And we heard time and time again that um what you said, governance often happens after development. And um, we took a pretty significant stance on that and said, can we reinvent that? Could we put governance alongside the development process so that I like that alongside?

SPEAKER_00 19:11

I like that a lot.

SPEAKER_01 19:13

Exactly. And you know, part of our critical observation, we've got we've got this beautiful diagram of the data science team and the and the governance compliance team, whatever. We saw a ping pong match going back and forth of I build something, I send it over, somebody sends a question back two months later, they you know, put it in in a Excel tracker, they send something back, they want some data, blah, blah, blah. And so for the data scientists, you know, they create something in a month and then 11 months go by um with interstitial questions along the way, and it's like a terrible way to work. You know, and for the governance team, it's equally bad. I mean, a lot of empathy for the governance team because they're trying to like peer in and figure out what are they doing over there and what like help me help me know that it's okay.

SPEAKER_00 19:57

You don't want them to be a gate, right? And that's right, exactly.

SPEAKER_01 20:00

Yeah. So we we really put it alongside. We we observe that if you have uh tracking of the work that is done, if you have governance gates, if you are building in um reproducibility, and I think we'll come back more to uh in depth what that means in a second, um, you know, if you're building those things in along the way, you sort of satisfy both worlds. You don't have repeated work on the data scientist side and and all the materials are uh are are built along the way. I'll give one analogy at the end. You know, I think it's a lot like uh building a building, right? If you built the entire building and then the um uh then you came and like pulled the permits and had somebody come observe how you built it to make sure it was built correctly, it'd be chaos, right? Uh the way that building a building works is you build the foundation, then you get it inspected, and then you get approved, and then you do the next step, right? And you're you're checking along the way that things are are working well. We think that's the analogy for how it should work rather than waiting until the building is built, if you will.

SPEAKER_00 21:01

So the pieces stop fighting against each other, like security and speed and risk and like all of it stops fighting each other and it it really becomes something that's built in, not bolted up. That's what we like to say. Hey, I'm gonna give you a shameless plug moment, right? Because you've described the partnership between Domino Data Lab and NetApp as something like peanut butter and jelly, um, which I quite like, and I think that that's a great description. Um, I want you to spend a few minutes talking about the pairing between the two of us as something that we s solve better together, but also would love for you just to give us the the 50-foot view of who Domino Data Lab is.

SPEAKER_01 21:42

Great. Um, I'll say I think uh, you know, I'm the dad of twin three-year-old boys. So I think when I said that I had like making lunch on my mind. Um I'm okay with it. It's a great analogy. I mean, so um uh you know, first, what does Domino do? So we build A platform for advanced mission critical AI use cases. So hence why we're talking about, you know, what does it mean to be in a core critical business process? That could be something like mortgage underwriting. That could be the work we do with the U.S. Navy around uh underwater mine detection. Uh that can be helping build new uh cancer therapies. I mean, a variety of different things, but but the underpinning is their mission critical. And so the stakes are high, the rewards are high, um, and the uh the AI work is tough. It's not it's not simple. So, you know, these are this is where we live. And the way that we solve that is we build a platform that takes care of a lot of the infrastructure integration, takes care of a lot of the tracking, governance, regulatory concerns you have, um, but ultimately makes it a beautiful place to build and deploy your AI work uh in your business. So in terms of the peanut butter and jelly, um, you know, NetApp, so uh maybe I'll tell a story of how this came about. Love that.

SPEAKER_00 23:03

That's great. Yeah.

SPEAKER_01 23:04

You know, we didn't, we didn't sort of like uh invent it on the fly. We actually worked together at several customers for for several years. And um uh I was sitting down at some of the NetApp team having dinner, and you know, we're just brainstorming. We're like, wow, this, you know, we sort of see how this could be really great. I mean, the two main things that really stood out were um NetApp's uh ability to help with reproducibility and governance, largely through snapshotting, um, was just sort of seemed like such a clear fit for data scientists. And and and I think you guys had this realization on your own and were sort of down that path as well. Um, and then, you know, the the hybrid and multi-cloud, uh, you know, putting the AI where the data was, uh, you know, these two things sort of like started to come together. So, you know, we actually sat down and sketched out and said, we could really see a way that that uh these things could work, work together. Uh I'd say critically we'll put it in front of customers. Um, of course, you want you want customer feedback, right? Uh we don't do that, don't do this in vain. Um, and then set about a journey of integrating our products together. So, I mean, to describe it in the tasty peanut butter and jelly sandwich way, when someone is in domino, um, a data scientist doesn't have to think about uh being a storage admin to use their data in NetApp. And by the same token, a storage admin doesn't have to worry that the data scientists are clobbering the storage. We've basically put those together so that data and the development um, you know, go together incredibly well. And and I mean, I'll say the reception amongst our customers has been fantastic. We um we have many, many customers who love the tight integration, but most importantly, the capabilities they get uh in their in their workflows.

SPEAKER_00 24:54

Yeah, I think that that's I love the fact that you talked about the shared belief because that's that's what we love in the partnership, right? Is that it's it's not just enough to slap each other's logos on things, but you're doing it for a customer. And we started that with you at customers, right? Where AI compute and AI storage coming together to say we can build a multi-cloud world from day one. And I think, as you said, I think customers, and correct me if I'm wrong, the customers that we have worked together at, they're seeing the results of our partnership in in really easy terms, right? They're running faster, right? What how are they seeing that that partnership come to life?

SPEAKER_01 25:32

Look, I'm a I'm a believer that um great software should sort of disappear in in the user sense. Right? You know, and and so um uh you know, the things that they are getting are um sort of uh a magical experience where they don't have to think about reproducibility because they have a precise snapshot of the data they use. They know all the code inside dominoes so those are kind of how the two pieces are working together. And this especially becomes critically important back to reproducibility and and governance that we talked about, both for internal, but also for a regulator. Um, you know, what you can basically do is go all the way from a regulator question all the way back through and take a pharmaceutical use case where maybe there are five or ten years of research that have gone into bringing new therapy to market. Yeah. You can trace all the way back through to, you know, the specific experiment that a particular scientist ran and exactly what the data snap uh data set is because it's snapshotted in NetApp, all sort of put together in one piece as a as a bundle for you to go look at. And I mean, that's just a that's sort of a mind-bending experience. Because when you hear about how it typically works in an enterprise, they're like tracking stuff everywhere and having violations and you know, I mean, it's just it's sort of nuts, but it's we've just put it all in a single thread and it it's a really magical experience.

SPEAKER_00 26:56

Aaron Powell We know where it is, we know where it's been. Where does all this go next? Right? You've talked about moving beyond um single models, right? Uh with agents that operate on their own. What your future cast for us, what where does this go?

SPEAKER_01 27:13

Yeah, so I'll give you two two eras. I mean, I think you know, the first uh era is agents have raised the stake on everything, right? Because they're essentially autonomy. And um, I think again, back to where we started this conversation, uh, people view generative AI as magic in their like sort of individual experience, but um, we forget that agents don't have like ethics or like really great memory. And, you know, to think that they sort of one-to-one replace a person in making a decision is like crazy. So, you know, the this has sort of raised the stakes. We both want the automation, but we need um, you know, good oversight, repeatable results. And so, you know, we're we're trying to build a foundation where agents can be well managed in an enterprise context on those critical workloads. And and that's hard tech work. I mean, they're we're you know, discovering and inventing new ways of doing things and understanding how to how to guardrail and trace and monitor, you know, these are a little different than than how you we used to do that traditionally with uh machine learning models. Um, but I'd offer, I think, the second um, you know, sort of the second regime, we are strong believers, as as I said, that there are going to be AI systems that bring together different types of models, machine learning, statistical, computer, you know, all the different types with generative AI into broader uh business applications. And I think to be a really futurist, I think that's the way we go from um uh LLMs uh and sort of foundation models into a world model or a neurosymbolic type era where we're we're bringing together multiple different types of things. So if I if I'm allowed to think five years out, three years out, five years out, that's kind of what we're tr what we're seeing next and where we're trying to navigate towards, if you will.

SPEAKER_00 29:04

And that's why you're so keen on and and I think that that's really the point of this conversation, right? Build the right foundation and the right kind of data infrastructure right now because you're not building it for what you can optimize today. You're building it for, like you said, what that five-year out. And I mean, who knows at the rate we're going, maybe it's three years out. Who knows, right?

SPEAKER_01 29:25

Yeah, yep, exactly.

SPEAKER_00 29:26

Hey, so what is our partnership taught you about running AI at enterprise scale, especially for enterprises, right?

SPEAKER_01 29:35

Yeah. I mean, look, I I'd go back to what we said a second ago. I mean, um, it is uh taught is maybe a strong word. It is reminding me how important uh, you know, customer first and the particular problem they have. Um, technology in search of a problem is no good. Uh a problem in search of technology is the right way to go. And so, you know, working backwards from what are we trying to solve to what are the technology choices, um, you know, that continues to be uh important. Um, you know, the the other thing is um building uh an early architecture uh, you know, as as we're talking about here, looking towards the future, um, you're factoring in speed, agility from from day one. And so architectural choices still need to be up front, right? Using a sort of agile approach um uh to thinking about hardened infrastructures, not always a great, a great thing. So so architecture and technology choices up front are very important, including platforms, you know, at all at all levels of the stack. Um and then, you know, I I really think that um in this, you know, we sort of talked a little bit about sovereignty and the importance of data and that sort of thing, but I mean, data infrastructure is just gonna be still the linchpin of of uh what what organizations need to focus on first um for the years to come. And um, you know, really think about that as as the bedrock uh upon which things like decision platforms or or or um or application platforms sit.

SPEAKER_00 31:10

So I love that. And that's a shift in mindset, right? That the if we can get everybody to shift that mindset, then that's the real turning point for us, I think, right?

SPEAKER_01 31:19

Absolutely.

SPEAKER_00 31:20

Yeah. This has been great. I appreciate you. Um and and I just wanna I want to call out a couple of points because I think we've had a really rich conversation. I love the data sovereignty conversation. I think that that's gonna continue to be something that everybody grapples with at the speed with which we're moving, it's gonna become even more important. I want to go back to the very first thing we talked about, though, because it's really resonant with me, right? We have to normalize failure and not stigmatize it because what we're doing here is we're basically failing with a with a roadmap. If to take everything that you've said is if you do some really great planning, if you understand what your data scientists are working on, if you're helping them chart the course and eliminate the friction that comes from old thinking like shadow Xorg or like bad planning or bad operations, then we are really turning the dime for enterprises to do and focus on that data infrastructure of the future that can help them use AI. So um, I have loved this conversation truly. Um, this is a partnership that we're excited about. And I love that it's a partnership that's based on shared belief. So thank you so much for that. T Rob, thank you, sir, again. We appreciate you. I appreciate you joining me. Thank you for all of you for tuning in to Unleashing Genius again. Um, and if you'd like to continue the conversation, please go to netup.com or come visit us at Insight in Las Vegas in September. To learn more, please reach out to netup.com forward slash insight or just reach out to me. So we'll see you next time on Unleashing Genius.