November 12, 2026

00:48:57

From The Field at VeeamON Sydney 2026 | Mika Yamamoto & Shiva Pillay on Getting Past the AI Hype Without Losing Trust

From The Field at VeeamON Sydney 2026 | Mika Yamamoto & Shiva Pillay on Getting Past the AI Hype Without Losing Trust
KBKAST
From The Field at VeeamON Sydney 2026 | Mika Yamamoto & Shiva Pillay on Getting Past the AI Hype Without Losing Trust

Nov 12 2026 | 00:48:57

/

Show Notes

Recorded for VeeamON Sydney 2026, the one-day Sydney stop of Veeam’s global tour, this From The Field episode sees KB sit down with two Veeam leaders to explore what happens when AI enthusiasm runs ahead of trust.

Chief Marketing Officer and Customer AI Officer, Mika Yamamoto, explains why companies are now all in on AI but still searching for value as token costs climb. She also covers the gap between efficiency and efficacy, and why accurate answers, correct actions and protected data decide whether customers keep trusting you.

Chief Sales Officer, Worldwide Solutions, Shiva Pillay (Previously SVP Worldwide Data and Trust, as well as General Manager for Americas), unpacks AI FOMO, and how over-privileged agents can surface sensitive files nobody knew existed. He also explains why trustworthy AI needs data classification, access context and precise recovery working together, and how attitudes to AI risk differ across regions.

Key topics: VeeamON Sydney, AI FOMO, AI value and ROI, token costs, customer trust, agentic AI, over-privileged agents, sensitive data exposure, the data and AI trust layer, precision recovery, AI governance, regional approaches to AI risk.

View Full Transcript

Episode Transcript

KB [00:00:10]: What's up everyone? It's KB dialing in from the US for Veeam On Sydney. This is Veeam's global tour touching down for one day only, and the brand has come a long way from where it started. Veeam made its name on backup and recovery, the stuff you were quietly glad You had when something broke, really. Today, the room splits into 2 tracks, resilience on one side and security on the other, and both of them keep circling the same worry. By Veeam's own numbers, something like 90% of enterprise data is dark, unclassified, and ungoverned, and now AI agents are being handed the keys to it, moving faster than any team can audit. Which means the thing that you are backing up might already be the thing that's being quietly compromised, and you find out when you go to recover. Across the day, I'm sitting down with the people who run resilience and security for a living to work out what's a real shift and what's just another agentic conversation bolted onto last year's pitch. This is From the Field. KB [00:01:20]: Let's get into it. Joining me now is Mika Yamamoto, Chief Marketing Officer and Customer AI Officer at Veeam. And today we're discussing, before AI scales, trust has to hold. So, Mika, thanks for joining and welcome. Mika Yamamoto [00:01:39]: Thank you for having me. KB [00:01:41]: So, Mika, I know you joined Veeam relatively— you're new to the role. So, I've looked at your background and your pedigree. So perhaps I want to start there with you and understand your visibility into more than half a million organizations globally. But then what's the reality of enterprise AI adoption today? And then how different is it from the hype we're sort of hearing in the market? Because you've obviously got a lot of experience, so I want to hear your thoughts. Mika Yamamoto [00:02:11]: Well, I think that what's fascinating about the time that we're in right now is that I've, you know, had the opportunity to be able to see several waves of innovation, whether it's mobility or whether it's the advent of the internet. And what's amazing right now is how fast customers and companies and individuals are evolving in terms of their stance on AI. So, if I back up to a year ago, standing in front of a set of customers, and regardless of where I was in the world, whether it was Australia or Germany or France or, you know, the US or Canada, a lot of companies were saying, Hey, I'm kind of dipping my toe into the world of AI, but I'm not really sure what I'm doing. And now as we ask customers, and I'm here in Australia now looking at, you know, in front of hundreds of customers, everybody is putting their hands up and they're in AI with both feet. And so I would say from a reality versus hype standpoint, it's been remarkable over the last year how many companies and individuals are saying, I'm all in with AI. I think the challenge is that they're still figuring out what the value is. So, we're no longer dipping our toe in the water from a company standpoint. That's not a reflection of Veeam, but if I think of the collective of companies that are out on the planet. Mika Yamamoto [00:03:35]: And so, I think companies are exploring this with both feet, but I think that what they are now— what's happening now is, you know, there's a lot of hype versus reality in terms of the value that customers are experiencing or organizations are experiencing with AI. So, everyone is deputized to use it. Everyone is in with both feet. But I think that right now there's the, you know, there's a lot of activity, there's a lot of spend. I mean, token spend is, you know, increasing for a lot of companies, but I think they're really trying to figure out, you know, what is the value that we're getting out of this? KB [00:04:08]: And then a quick question I have there, how would you define sort of hype around AI versus the reality? Is it more so that people are just ripping the bandaid off, just getting into it, trialing it out, seeing what works for them, and then going from there versus talking about it or even like listening to podcasts. Because I've spoken to a lot of people on the show and they're like, it's one thing listening to the podcast, but it's another thing to literally just get amongst it and trial it out. Mika Yamamoto [00:04:37]: I would say we're into more trialing it out and getting into it. I think that, I think that it's not either or in terms of trying to learn, listening to podcasts. Trying to read as much as possible, you know, figuring out what prompts might work for you as an individual. So, I think that we're seeing, you know, things happen in parallel. So, people are trying it out while they're listening to podcasts, while they're reading as much as possible, while they're getting in touch with as many peers as possible. We have, you know, we host a ton of customer exchanges. I participate in several exchanges of my peers. And what we're trying to do is figure out like, what are, what is everybody doing as individuals and as leaders? And what's working and what's not. Mika Yamamoto [00:05:17]: And so, I think that it is listening to podcasts while people are trying things out. And so, I think the hype versus reality is really, there's a lot of hype about the value of AI, and I think it's exacerbated, frankly, from technology companies. But, you know, I think that companies are still trying to grapple with where is that value? And it could be, you know, value in terms of we have accelerated our growth, or value in terms of we are more efficient, or both. But I think companies are still trying to figure out, or organizations, you can include universities and schools, et cetera, organizations are trying to figure out like, where can we actually see benefits from AI? KB [00:05:54]: And then in terms of history, given your pedigree in this space, if we look at even virtualization, cloud, people thought like, this is all hype. Even when the internet sort of came out, you know, in the '90s, people thought, oh, it's all hype. But then that reality starts to to settle in once the dust settles. Do you still think, given what we've been through in terms of technology iterations, it's much the same? Obviously, it's a lot more enhanced and things like that, but is it going to be, yes, there was hype at the start, but now people are finding their groove with where they sit in this AI game? Mika Yamamoto [00:06:24]: You know, I think that people are finding their groove. I think that because there's democratized access to AI, so they're There's been some waves of innovation where, and many waves, like the advent of the internet, mobile computing, you know, that was consumerization of IT, right? So, you saw a lot of people buying those devices. So, there's maybe Luddites within an organization that were saying, oh, this is all a fad. But you saw people as consumers actually, you know, consuming, you know, experiences on the internet or consuming mobile experiences and bringing them into work, which became a reality. for the organization themselves because they had to deal with the fact that, you know, in the mobile computing world that people were bringing their phones into work and accessing and doing work on their phones. And they had to think, okay, well, we have to make sure that we deal with that reality. I think that's what's happening to a lot of organizations now where some organizations are being mandated to use AI by their boards or by executives because those bodies are saying, look, I think that there is some there there in terms of efficiencies. Usually when I see these mandates, it's typically to be able to drive efficiency. Mika Yamamoto [00:07:31]: So we expect you to lower your cost basis by 10% by using AI. AI isn't free by any means. And so it's either a mandate or there's a bunch of companies that are, sorry, organizationally, there's, you know, people within the organization that have self-deputized using Claude, using ChatGPT, Perplexity to be able to, you know, create their own, create their own, create their own agents and do work on their behalf. And so I think you're, There's different pockets of how innovation is rolling out, not dissimilar to, you know, other ways of innovation like the internet or like mobile computing. There's people that even if, you know, if it's mandated, if it's not, even if it's not mandated, people are bringing AI to the workforce, maybe not en masse depending on the industry, but they're still doing so. And so that's creating a need and, you know, and also an opportunity for companies to start exploring, you know, the benefits of what AI can do for them. KB [00:08:26]: So then my next question would be, would you say organizations are seeing genuine business value and where are those expectations getting ahead of reality? And I ask that because recently what's happening, a lot of my interviews, people are saying we've over-indexed on AI KB, we're trying to find the value. But then other people I've interviewed have said, yeah, but it's going to take a little bit of time to really see that ROI. It might take another year before it's like we've really seen that benefit. Is it just going to be staying the course and riding out the proverbial wave through this so people can really understand in fidelity where those value lies within these businesses? Mika Yamamoto [00:09:05]: That's a great question. You know, just because there's a lot of time and money. I mean, AI is not free, right? I mean, you open access to Claude, for example, to your entire enterprise, whether you're small or large, and the consumption of tokens that are getting more expensive is ever increasing within an organization. And I don't know that, you know, costs are going up, and I don't know that the value or benefit to those organizations is going up as steeply as the costs are going up. And so I would say that, you know, there's gotta be some, what you don't wanna do is tamper people's curiosity and, you know, an interest in AI. But I think there's a couple things that are happening in terms of, you know, making sure that the employees that are using tools are recognizing that they're not free, that they're recognizing that, you know, that they're thinking about what good looks like or what great may look like. So, an example would be, you know, if you're creating an agent to be able to interface with a customer, that you recognize, number one, are you accessing the right data? Should you be accessing this data? But also, are you creating a good enough experience for the company itself? And is that agent actually talking to other agents or other humans that may be engaging with customers as well. And so there's a governance element there I think that's required. Mika Yamamoto [00:10:23]: And then there's information sharing, information sharing within the organization in terms of where value is being created or not, because we're testing, we're all testing and learning, but also, you know, sharing among peers, whether it's peers in terms of functional areas, salespeople to salespeople, IT people to IT people, security to security, Or peers in terms of, you know, financial services organizations to other financial services organizations. There are pockets where there has been an extraordinary amount of value that has been created with AI. And so I think that that will continue to grow. KB [00:11:00]: And Mika, you mentioned something before which I want to explore with you. It was around temper people's curiosity in AI. So what does that look like in your eyes? Mika Yamamoto [00:11:10]: Like not to tamper it? KB [00:11:12]: Yeah, exactly. Like, what is, like, I'm curious, like, what do you mean? Mika Yamamoto [00:11:17]: What I feel like is that, you know, because we, you know, we're trying to figure out what value exists in all the functions at many levels, you know, worldwide, on what the possibilities are with AI, is I think that what we need to do as leaders and as organizations is just ask our you know, our employees, you know, what do you do? What do you think you're doing this in service of? And so, number one, what you're— the resources you're using aren't free, whether it's your time using AI or whether it's, you know, using up tokens from an AI standpoint if you're using Claude. But also thinking like, what's your end in mind? And start driving that practice or that like rigor, if you will, to say, you know, what's your end in mind and what what benefit are you actually seeking by, you know, by creating this, by, you know, creating this agent that reads your mails and then, you know, and then spits out responses on your behalf? It's like, well, are they quality responses? Are they, is that doing it effectively? Is it saving you time? And so, I think we have to start defining what, you know, what the hypothesis is for the value. And we may be wrong, but I think we have to start drawing a line between, hey, I'm doing this and, trying to define what benefit or outcome we're seeking by doing that activity, whether it's done by an individual or, you know, a team or a function. KB [00:12:39]: And you know what's interesting about that? When you said using, you gotta obviously leverage time to make these agents, et cetera, work for you. So, I've been hearing from people in the market say they're trying to find efficiencies everywhere, whereas perhaps you don't need an efficiency with every single thing, and it's even costing more time to then get an agent to do something. Mika Yamamoto [00:12:59]: Exactly, right. KB [00:13:01]: So, are people like, are people just not perhaps spending their time in the right routes around AI? They're trying to optimize for everything, which I get. I'm just curious then on that, like, what are you sort of hearing? Mika Yamamoto [00:13:14]: You know, I think that often what I've found is that there's organizations where the board has mandated use of AI to drive efficiency. And so, they feel this sense of panic. They're like, well, we have to take 10% of our costs out. And that, you know, some boards have actually, you know, even, you know, if I think even 3, 4 years ago, I started to see this in boards mandating the use of AI and just saying, look, I'm gonna give you a target and I'm gonna give you an organization-wide target to take 10% of your costs out. And so it is a, you know, panic situation where organizations want to respond to their, you know, to their boards who represent their shareholders. And so it's a, it's almost like a, a blind use of AI or using it for the sake of AI so you can come back to your board and say, hey, we're using AI and we've taken out these costs versus saying, taking a step back, because efficiency is one benefit of AI, but there's also efficacy and you can actually use AI to increase customer satisfaction, which isn't just an efficiency metric. It's actually an ability to meet a customer and employee's need in a better way. And so I think that companies that I've seen successfully use AI have come to it with a point of curiosity to say, what benefit could we see from this? Is it efficiency or is it actually, you know, more creativity or is it actually to, you know, use or redeploy individuals in a more successful way? So an example would be being able to use AI in a customer support scenario where repetitive tasks can be taken care of with agentic solutions. Mika Yamamoto [00:14:50]: So reset of passwords, being able to check, be able to check inventory, be able to process a return so that an individual could— individuals within a customer support organization can deal with the complex problems that may require more empathy with a customer. But they, you know, the basic or fundamental, what, you know, in customer support you'd call an L1 ticket, can be taken care of by agents. So you can deploy more human hours to the more complex problems, which And then you can use agents to take care of customers' needs in a much faster way, 'cause they can work 24/7, and they can work with an almost infinite number of customers at any one time. And so, you can increase customer, and when done well, you can increase customer satisfaction. You increase employee satisfaction because employees are doing things that add more value to the organization. They feel like they have, you know, a better and broader remit. And so you can, you know, it's a win-win. So I think it's more like, what's the possibility of things we could do with AI that both drive a top-line and bottom-line benefit and a customer satisfaction or employee satisfaction element just versus just sheer efficiency? Because efficiencies can be realized, but I think you've gotta think of the big picture. Mika Yamamoto [00:16:01]: So I think if you're just looking at efficiency, you may just take AI and deploy it on top of existing processes. If you take a step back and say like, what's just the benefit that we can drive, which could be top line, bottom line, and satisfaction, you end up redefining processes. You're not just laying, you know, not just laying AI on top of old processes, but you could create new normals that can then, you know, create more value for customers and in fact, you know, and employees as well. KB [00:16:27]: So then sticking with customers now for a moment, how would you say AI is changing what customers sort of expect from companies that they do business with? So for example, Even in media, people are complaining, saying, I'm sick of reading AI slop. So now it's like, well, they would prefer this sort of style conversation because it's 2 humans having a conversation authentically versus a machine spitting out an article. So, I want to, given your role, explore that in a little bit more detail. Mika Yamamoto [00:16:56]: You know, I think that customers' expectations are fundamentally the same. Like, if you back up when, You know, in a, in a time before computers even existed, right? Customers expect, you know, they expect value, they expect convenience, they expect personalization, they expect, you know, a, a good or like a company or an organization to be able to anticipate their needs, whether it's a hotel, whether it's a car manufacturer, whether it's an, like, whether it's a B2B transaction, those fundamentally are needs that all customers have had for the history of time. since commerce, since commerce started. I think that the advent of AI has essentially allowed the acceleration or the improvement of, you know, value and convenience and personalization and anticipation of needs. And it's allowed not just companies like Amazon who have a ton of resources to drive agentic or AI solutions, and they've been doing so for a long time, well before, you know, 3, 4 years ago when, you know, ChatGPT was released and you know, it unearthed this huge, you know, AI wave. A lot of companies have been exploring AI far before that, but now because it's been democratized and because even small businesses can use AI to be able to drive, you know, value, convenience, personalization, and anticipation of needs, there's an expectation now that these things can happen faster. And there's an expectation, for example, that if you are dealing with a retailer, even if it's a small retailer, that they can actually answer a question because they can use an agent to answer a question, check inventory, be able to process a return. And so I think that the expectation is that these things, these things that have existed since the beginning of commerce in time, can happen faster and with any size company, not just the biggest of big companies like Amazon. KB [00:18:46]: And then if I flip to the customer side, as we know, people are not as loyal as they were back in the day. They get annoyed. I think even here in the US, T-Mobile went down, which I was impacted a couple of hours. People were raging online. So, do you think that people's tolerance then has gone down, which forces companies to leverage AI more because they know they need uptime, they need to respond, they can't have, you know, outages, etc.? So, I'm curious to see it from the other side then on how AI then can be leveraged to ensure that customers are happy. Mika Yamamoto [00:19:19]: Yeah, I mean, I think that it can definitely be leveraged to make sure customers are happy, but it adds this level of complexity for all companies. Because in order for AI to make customers happy, you have to make sure that AI is accessing the right data. AI is actually providing the right answer. AI is actually taking the right action. So imagine if AI was actually helping you change your flight and it went from it, you know, you wanted to go from, you know, wanted to change from Atlanta to Boston and you were actually going to redirect to New York. Well, what if the agent actually just canceled your flight altogether and you went to the airport and you realized you had no flight? And so, making sure it's actually taking the right action on your behalf. And so, while there is this opportunity for speedier resolution and 24/7 resolution on your behalf, it's massively complicated just because you're dealing with lots of data, you're dealing with privacy concerns, et cetera. And you wanna make sure that, again, if it's, you know, you think of taking a draft, you mentioned AI slop, if it's actually going to take— Some inputs and actually create either a visual that actually, you know, is, you know, creating a visual with someone with 3 hands, for example, or a hand that's missing, or creating AI slop in something, you know, in a tone that actually humans don't speak within, and then you lose credibility. Mika Yamamoto [00:20:40]: I think that there's an opportunity for speed, but there's also an opportunity for, you know, degradation of trust by having the wrong information and someone takes action, or taking the wrong action, like your canceled flight, for example. Or actually, you know, with agentic solutions, you know, driving the extreme risk of, you know, breaching data privacy or data security, which is even worse because that erodes trust. There's nothing that erodes trust more than, you know, I provide an organization with my data or information and they breach my trust by actually, you know, enabling or not securing my data. So there's a leak of my information that I trusted an organization with. KB [00:21:20]: So, Mika, final question, staying with trust. What do you believe trust looks like now in an AI-first organization? And then why do you think it'll really matter now over the next few years? Mika Yamamoto [00:21:35]: You know, I think what trust looks like with an AI-first organization is an organization that essentially says, look, where can I insert AI, whether it's a, you know, a customer engagement or employee engagement or with a partner or an investor? How do I insert AI to be able to, into my workloads, to be able to drive a better experience, right? Whether it's, you know, more efficient or more timely, et cetera. You know, I think what trust looks like is that you, if you're seeking information, that you get accurate information, right? If you're seeking, for example, or if you're using AI to help diagnose, if you take a picture of your eye because it's swollen and you upload it to your computer, you upload it to your doctor and he or she is going to, you know, use AI to diagnose you. If you get a misdiagnosis and your situation gets worse, you know, that's a terrible outcome. So you have to be able to expect that you get an accurate answer. The other piece is that you need to, you know, so there's nothing that erodes trust more is if you get an inaccurate answer, which happens a lot with AI currently. And it will always happen, right? I mean, people make mistakes and people have, you know, incorrect answers as well, but people expect more from an AI. from an AI response. The other one is that it will take the correct action. Mika Yamamoto [00:22:48]: So, will it take the right action? Will it, or will it inadvertently delete your account, or will it inadvertently cancel your flight, or will it inadvertently route you to the wrong location? So, is it going to take the correct action? And if it's not, you can have the opportunity to erode trust. And lastly, will it protect you? So will these solutions protect, is your information that you're sharing and trusting with an organization, is it protected? And again, that's a massive opportunity for earning trust and it's also a massive opportunity for losing trust when your data is used in an incorrect way or shared incorrectly or allowed access to a bad actor. And so those are situations where trust again can be earned or it can be lost. And so, I think that because there's so many opportunities for an organization, big and small, to be able to use agentic solutions to be able to help their customers, it increases the surface area for opportunity and goodness, but it also massively increases the surface area for degradation of trust and reputation. And so, those are balances and trade-offs that organizations need to consider when they're thinking about leveraging AI, especially customer-facing AI versus employee-facing AI, whether the risk is somewhat lower because you're able to walk up to your employee and talk to your employee versus customers who you may not be able to, you know, you may not be able to reach them just based on the relationship you have. And so I think that, you know, there's a massive opportunity, but there's also, you know, a massive responsibility to make sure that you're, you know, you're earning trust every single day with your customers. KB [00:24:33]: Joining me now is Shiva Pillay, SVP Worldwide Data and Trust, as well as General Manager for Americas at Veeam. And today we're discussing AI FOMO, which meets data reality. So Shiva, thanks for joining me and welcome. Shiva Pillay [00:24:47]: Thanks, KB. How you doing? KB [00:24:49]: Okay, so Shiva, I just really wanna start first cab off the rank. Like, what is like AI FOMO? Like, what's your version of this? Shiva Pillay [00:24:58]: Yeah, it's a great question. Like, look, what is FOMO generally? It's a very simple topic. It's the fear of missing out. I used to think about it when I was a kid. If I didn't get invited to a party and I was sitting at home with my friends and I didn't get that invite, it would eat at me pretty deeply. And I would think about it like, you know, am I not cool enough? Am I not hip enough? And I think AI has a very, very similar twist. Am I not doing enough with AI? Am I not innovating fast enough? And Everywhere I look left or right, somebody is talking about AI. You couldn't get on YouTube, couldn't get on the news, and you couldn't get anywhere without an AI story per se. Shiva Pillay [00:25:34]: And so there's just this general feeling of like, am I doing enough? And am I participating in what is, I believe, a revolution of the highest order in terms of technology wave in the space that we've seen for a long period of time. KB [00:25:50]: And Shiva, given your role, when you say, when people are saying to you, like, are we doing enough? Like, what does enough look like? Because now we're seeing people over-indexing on AI and saying, oh, we've dropped too much money into it. Now we're trying to come back from it. So what does that sort of mean in your eyes? Shiva Pillay [00:26:06]: So the thing about FOMO is it's never quantified. So I'll lead back into your original question. The FOMO conversation is fear of missing out, but what are you missing out in? And I think your question really hits a little bit onto that comment. Like, What is missing out? Am I missing out on just playing with tools? Am I missing out on burning tokens? Because that is a lot of what's been going on. And, um, and that manifests in very different ways. For an individual, it means having access to the technology and using, um, using Claude or Cowork or Gemini and creating beautiful things. But I think for businesses, it's very different. It's like, how do I actually drive value? And how do I drive value? That is commercially sound, safe, and allows me to attract more or work with a better customer experience for the customers I serve or drive more revenue. Shiva Pillay [00:26:56]: That piece, I think, has been a little bit of the challenge. Like, I need to have the tools and almost an arms race, but I think there's a second wave where the measurement or accounting for what have I done with this, you know, set of infrastructure or set of investments that I've made. Um, where right now, at least my take has been in that beautiful experimentation phase. Everybody's doing stuff in their laptop. Everybody's admiring their own work. In the, in the old days, it took me a week to put a PowerPoint presentation together. Now I have people sending me 10 or 15 a day. Maybe they haven't read them. Shiva Pillay [00:27:30]: Maybe they haven't understood what they've put forward, but there is this joy of creation that's kind of taken us over a little bit. And that will, You kind of need that to go because I think it's such a weird technology that everybody individually should experience the ability to create. But then that's going to translate into how do I do that safely? How do I do that when it's impactful? And how do I do that that it makes an actual difference? And that is starting to sort of happen. You mentioned that withdrawal from like all these over-pivots in investments. And I think that's the challenge I think a lot of organizations are facing today. KB [00:28:04]: So then on that note, if we zoom out and look at boards, so now as you would know, like boards are sort of going on about like, we want AI yesterday, but then leaders are sort of asking whether their data is actually trustworthy to empower it, which is really the, you know, you guys are in that space a lot. So talk to me a little bit more about what's going on here and maybe some of the, a bit of the conundrum that's happening at that board level. Shiva Pillay [00:28:27]: Yeah, I'll give you a different, I'll give you a story as a parallel. If you remember the cloud era, CEOs and boards were like, hey, get everything in the cloud because of just this hype that the cloud was cool. If you think about the cloud now, the cloud is very cool and everything is fantastic about it. But if you think about it now, people are like very selective on what workloads should go into the cloud, what workloads should be on-prem. And there is a big argument to have some hybrid sort of infrastructure so you can be successful with both. Same thing I think applies, but at a huger scale, that FOMO thing that we talked about in first question, is still like, it affects the board where the board goes, hey, I can see everyone's using AI. I need more AI adoption. Now, when I sat with, you know, a good example, I sat with a deputy CIO for a large consulting firm. Shiva Pillay [00:29:11]: They have SharePoint. They want to drive Copilot access to the SharePoint. Now, unstructured data, right? That typically would've taken you and me years of clicking on mouse pointers to find file directories, to uncover data, to figure out meaning, there was this natural deterrent to like looking and unsurfacing things that you shouldn't be able to unsurface. Agents and Copilot don't reason that way. They go in and consume all this unstructured data and provide visibility into things that maybe shouldn't have been sitting in different places. And so that firm, that deputy CIO was kind of like, hey, I turned on Copilot. What came out of it? Was not malice, but just by simple queries, overprivileged agents were able to find files that they shouldn't have been able to find. And that becomes the scary component. Shiva Pillay [00:30:02]: And that becomes the, hey, hold on a second, we need some great guardrails in place before we go and unleash AI into organizations because the downstream impact is my customer information might be surfaced in a place it shouldn't. My sensitive data information might be surfaced in a place it shouldn't. AI doesn't go horizontally. It doesn't go vertically. It goes horizontally. So typically you would access information in a very horizontal fashion. You'd go into an application, you go into that. AI just goes across the whole gamut. Shiva Pillay [00:30:33]: So I think for us, that's where you're getting this pushback. I have another CISO that turned off all AI instrumentation leading into a big sequence of events in their hometown. They work in a government agency. And the reason they did that is like, hey, we don't feel very comfortable on the what-if and what happens when things go bad. It becomes a pretty interesting and scary prospect. KB [00:30:58]: Do you also think, what was coming to my mind as you were speaking there was, do you think boards are just saying, are we doing AI? Like, are we doing it? As if it's like a binary thing that people are doing. Do you think that's just a very common question that people are asking? And it's, Not necessarily, you know, it's a hard one to sort of answer because it depends what they mean by it. But it's just more, we've seen this in the past around, are we good for cybersecurity? Are we going to get breached or not breached? Like, that's a very binary sort of question. Shiva Pillay [00:31:25]: Yeah, it's a great question, KB. I think it's funny because you would expect that not to be a question. Are we doing AI? I think that's what's happening, to be fair. I think the frustration I sit with professionals that are supposed to make this a safe place to work. I sit with the chief data officers, the chief information officers, the folks that, and they've had this really hard task of like, there's a cool new shiny toy that the board and everybody wants to use. And then they are the custodians of saying, hey, hold on a minute, we might want to think through how we use this. But yeah, if you're getting a question, are we doing AI, which what does that actually entail and mean? I think, yeah, you are finding this sort of not growing frustration, but growing like responsibility. It's a little bit more profound than the cybersecurity parallel you gave me because in cybersecurity, yeah, do we have our systems in place? Do we know what we know and do we have the right things? Do I have my firewalls? Do I have my data protection? Do I have immutable backups? And do I have ability to recover, which is something that we also help customers with. Shiva Pillay [00:32:29]: In the AI sense, it's more like, am I unleashing AI? And am I almost to the caution of, am I doing it? I'd rather get there than do it safely because the speed in which this space is moving is so quick. I don't want to be left behind. So it's a really tough challenge for the professionals that protect and store and give safe data. At the same time, boards will be boards and, you know, you want to maximize your shareholder value. And to do that, there is great You know, I'm not a doomsayer. There was great automation and great, great, great innovations coming out of the use of AI. I use it every day. How it impacts people can demonstrably change how easy it is to get things of value to our customers. Shiva Pillay [00:33:11]: But at the same time, like, am I doing it safely? Good example. I mean, one that I'd give you that I thought was quite topical recently was, you know, an airline used AI to reason its refund policy and and started issuing refunds and flight challenges where it became quite problematic. They claimed that AI was the reason for the fault, but you now start getting to this really interesting place where courts are now going to rule to say, hey, whether it's AI or it's a human, it's the company that's making these decisions. And therefore you have to honor either the refund schedule of a flight or the actual price of tickets. stuff like that. And you're going to see a lot of that hallucination happen in short time. KB [00:33:55]: So that probably leads me to my next question around Veeam is talking about data and AI trust as a new category. Now, the question that I'm curious to know is we've seen like data over the years collecting all this data, then nobody wanted the data. Now we need it again to power AI. The whole trust thing, I mean, even when I was working in industry, this was a A very common thing people would talk about, we gotta build trust and all this sort of stuff. So then the question that I really wanna know is those 2 elements coming together, is this really a market shift? Because we've sort of been talking about it and I know there's a lot of people in the market like yourself talking to me about trust, but what's sort of the new definition of this? Because originally I would say people would define it as more of a fluffy word. Shiva Pillay [00:34:45]: KB, you sound like a skeptic. You've been in industry for a long period of time, and you have every right to be. I throw up a little in my mouth when I hear people say, hey, we want to be a trusted advisor, and we want to do all that. Like, I've heard those marketecture terms. Here's how I'd lay it down for you. And like, don't worry about Veeam for a second. Just think about the concept. Before, identity is a hugely important thing. Shiva Pillay [00:35:09]: Secondly, the perimeter was a hugely important thing. And security and governance are great. AI agents do not discern, and overprivileged agents do not discern which systems they're touching. They don't, there's no human logic or am I a bad person or a good person accessing data? It's just, I think of it as the analogy I've used a few times is, you know, there's bad food in the company fridge that's gone off and moldy. It might be the Chinese takeaway I had 2 weeks ago. A human would look at that and go, I am not eating. There's no way I'm eating that. It's got fur growing on it. Shiva Pillay [00:35:47]: AI and agents, they're like, yeah, I'm eating that for breakfast. And what happens when that happens is you do get AI food poisoning, but you get that food poisoning manifesting in business decisions, data that making decisions on data that are incorrect. So you need a level of trust and the only common denominator for AI AI doesn't exist without data. If it doesn't have data to work with, manipulate, decide for, and reason with, you're not going to find any value out of it. So to trust your data is a really, really important thing. And so how do we do that? And how do we create this data and AI trust layer? It's very simple. We do on the left-hand side, I would say the security components. We're bringing 2 disciplines together, data, AI, and security, where you do this classification and you really understand what data you have. Shiva Pillay [00:36:33]: Second, we provide context. Who has access to that data? Who should have access to that data? And where is that data residing? And from that, you can unlock some toxic combinations. I'll use a crass example. KB, you sit in, you're sitting in Florida right now. You have access to my employee files for my salary in a location in Europe which has GDPR. That toxic combination is almost impossible to unsurface if you don't have context to your data, visibility, and contextualization. When that toxic combination pops up, you need an ability to remediate, see it, and close it out. Now, the right-hand side of this equation is Veeam's resilience conversation. Shiva Pillay [00:37:15]: We've been doing this for a very long time. The production data and your backup data now are going to drift even more than ever before. Agents are making micro changes to documents, they're making changes to All sorts of things that we may see and may not see. And the ability to have that drift close with your backup and copy data at the same time as really understanding your production data has never been more paramount. So if you think of it as trust, you just, you need visibility and you need a layer that sits as the final straw on, I know where my data is, I know who's accessing it, and I know what type of data, where it resides. And at any point in time, I can go back and reverse my position using, you know, Veeam's technology on the resilient side to that golden state where I thought things were all okay. Being able to do both in one platform before was like a nice to have. Our belief, and I feel very convicted about this, is it's the only way to solve this problem and unleash sort of safe usage of AI. KB [00:38:20]: So then what you're saying is, from what I'm hearing, is by having the appropriate data, which I know that we need to power AI, the trust becomes a byproduct because of that data, provided that it's normal, decent, all the sort of stuff you just discussed to ensure that there is trust on the other side. Because if it's not, if you're looking at a bank that's got, you know, hundreds of years of the records on pen and paper and who knows where it's gone, and we're trying to get through that to in order to— you leverage that to power it from an AI point of view. That's obviously a lot trickier than a company that's been around 5 years and it's, you know, doesn't have that backlog. Shiva Pillay [00:38:58]: Yes and no, because now AI can read that 100 years of unstructured data and very quickly make decisions from it. And great examples are like when we go and scan our customers' data, you will find in places where you thought you had no You'll find PCI DSS data sitting out in the wild. The, you know, the sort of silly examples, it was sitting out in the wild in a storage cupboard. I'm gonna give you a visual example, but sitting out in the wild in the storage cupboard out in Denver, Colorado, in a, no one's ever touched or opened that storage unit. If you just take the metaphor, AI can find that in 2 seconds, uncover it, unlock unlock the storage unit and say, hey guys, I have credit card data exposed. And it's not malice. It's just because of the way it works. And I think for us, being able to keep up with that type of contextualization of your data, giving people the ability to know where it is and also know who has access to it as things fly in real time is really, really important. KB [00:40:03]: So then on that example, would you say that people are almost shocked to be like, oh, well, we thought everything was fine. And then actually, It was, we had all this exposed customer data here and credit card data here. So it's like people may have this illusion that things are okay, but then when obviously push comes to shove at the end of it, it's like, well, actually this surprised us. Are you seeing a lot more people surprised perhaps on thinking that, oh, maybe things aren't as kosher as we thought they were? Shiva Pillay [00:40:30]: I think for folks in the space, no. I think what for folks in the space was the risk layer. They know there was always data that's sitting out there that shouldn't be sitting out there. And they also know people have access to it. The ability, what is scaring people now is the ability to do that without a human in the loop. And the laser-pointed nature of how AI touches things exposes this risk at a rate that they've never expected. You could go a year or two if you'd had like in a non-AI world, you can go a year or two, maybe 10 years of having exposed like some of this conversations we've had with customers, we're exposing data that's been sitting in places for 10 years. There's just no way they knew it was sitting behind the laundry basket. Shiva Pillay [00:41:14]: I'm using a house metaphor, but sitting behind the laundry basket because nobody's ever looked there. AI sweeps the house in 5 minutes and tells you where everything is and then exposes that data. And that's what's scaring people. Not so much they didn't think that there was, like, no one thinks they're in a great shape. That would be pretty folly if you're a professional in the space. But the rapid ability for that to be exposed and cause challenges, that's what the scary thing is. And so how do you go and do the end-to-end view of your data estate? How do you go and connect that to your resilience data? How do you make sure you can back up and recover to positions that you should have been as AI, as it agentically touches your data, changes things and changes the landscape? KB [00:41:58]: So, I wanna shift gears slightly and talk about— Sure. You've got boards demanding immediate AI access, which we started at the top of the call, but then practitioners are sort of saying, and I'm seeing it in comments and stuff, and each day people's view changes, but they're also saying like, we're moving too fast now, no visibility into consequences. So, ultimately, binary question, who do you think's really right in this scenario? Shiva Pillay [00:42:24]: Nobody. I think you have to move. So the, not to tie this back to your original, FOMO is real and FOMO does have a consequence. If I wasn't invited to the party, I wasn't very cool. And so I have 2 choices. I either get invited to the party and play in this space. And for companies, that means being relevant or not relevant. So that piece is right. Shiva Pillay [00:42:46]: The problem is this is hard work. And I've never been more convicted. I'm in Oz this week. I was in New York last week. I'm traveling around the world. And as I speak to people, I've never been more convicted. It's one of those, oh my God, this is such a hard problem. How do I start? And how do I start means like, it's like rewiring your house for the electricity. Shiva Pillay [00:43:07]: If you go to rewire your house, it's a pretty big project. And you kind of, a lot of customers might say, well, can I get away with just changing the the light bulbs. And like, in some of these cases, because AI is so pervasive, you do need to go and get end-to-end classification, end-to-end visibility. And then once you've done that hard work, your comfort level on how you innovate should sort of dramatically increase. And you can do both at the same time. You just have to do that very cautiously. KB [00:43:35]: So you're saying it's worth going to the party, getting the hangover versus not going and Having not a hangover because the risk is always going to be there. Shiva Pillay [00:43:44]: I don't drink anymore, but when I did, uh, yeah, it sometimes feels— the hangover is the downside, but boy, the memories are great. KB [00:43:53]: And this is interesting because even if I go back to my practitioner days, I remember it would take months and months and months for like, let's do a risk assessment, tech risk, business risk, cybersecurity risk, line 2 risk, you look at it. So now it's like, do you think people are really uncomfortable, especially if you're looking at an enterprise business? They don't like making decisions fast, but if they don't, to your point, they're going to be obsolete and bankrupt probably by the end of the 6 years. So how are people balancing that challenge? Because yes, you've got to manage it, but at the same time, you've still got to push on and get on with it. Shiva Pillay [00:44:23]: Yes, you must be innovating. I think where we, that's where we help. That's where companies like us help. We put tooling in place that gives you the ability to look at your data, contextualize data, and back it up. And the back it up component is equally as important as the context because as you make mistakes, you can undo. components of what went wrong and put yourself into a very safe state. That piece is where this becomes interesting, and it also where it becomes kind of, I have a backstop should things go wrong. I can see what data I have. Shiva Pillay [00:44:51]: I can contextualize who touches it. I can see when bad things happen, and I can undo very precisely. The precisely part is what is making the difference, because you would know in your days, and Generically doing resilience of like reinstating a whole data center or a whole department, it's quite painstaking and it's not relevant for an agent touching 10 files. The ability to look at those agent touching 10 files and say, I'm going to recover those 10 back to a state is pretty unique and a pretty unique requirement that has only existed in the last couple of years. And so that's something we solve for. Which gives you confidence on the big matters and also on the small micro things that happen on a day-to-day basis. KB [00:45:37]: So there's another stat here from Veeam that I'm going to read out. If 65% of operators feel, operators meaning practitioners, feel that they're moving too fast, to my last point, what are executives perhaps not hearing from the people that are really closest to this risk? Again, going back to the conundrum, we've got to innovate, we've got to manage risk, but we've got to move forward. I mean, that's quite high. How do you approach that then in an organization? Shiva Pillay [00:46:03]: No, executives are great. The executive's role is to push the boundaries. This is the common tussle. If you were too conservative, you'll never get anywhere. So I think this is a good tension that needs to happen. You have executives, like if I ran a company, I would be looking at how do I get the most innovation in the hands of my customers that delivers me business value? The conversation I'd be having is what is the cost of that from a risk perspective? If my job is solely to protect data, I would say everything's risk. And somewhere in between is this balance and somewhere where if you had a uniform way to look at this posture of risk and create, back to this word that you you don't necessarily love of data, AI, trust, then you can both come to the party and say, okay, we're going to innovate quickly, but we have a plan if things don't go so well. And we're able to recover, be resilient and drive that. Shiva Pillay [00:47:00]: Safe innovation is where this is heading. So I think that 60% number, the concept isn't I'm ever going to have a perfect state. It's going to be more, I need a safe way forward with a few ways to get back should things go not so well. KB [00:47:15]: Okay, so Shiva, final question. You're an Aussie and you're leading the Americas at Veeam. And I know that you said you've been traveling around New York, Sydney, who knows where you're going after this, but do you see any difference then in how regions talk about AI risk and accountability, even like in the US, Australia, Europe? Shiva Pillay [00:47:34]: Very different in different geographies, but the underlying thing is the same. Australia is very similar to the US, very bold, very innovative spirit. Let's get out there and get to the frontier and push some boundaries. And then you have some more conservative geographies where, you know, safety is a bigger priority than innovation. And you just see this different mix. That mix is great and healthy. And it's always great to see there are different regulations in different geographies, right? In Europe, heavily regulated data controls, a ton of things there that would help drive a little bit of a slower not a slower adoption, but a more thoughtful adoption process. And then there's the Wild West geographies where you could go out and do whatever you like. Shiva Pillay [00:48:14]: And there's no one really stopping you from being as aggressive as you like. So I think it's different everywhere. As an Australian going across the world, I get to, you know, tell it how it is in our uniquely Australian way. But I think there is joy and there is real good importance and understanding why people in different geographies approach things very differently. And that's a very big cultural thing. But the end net is everybody would like to innovate. Everybody would like to grow revenue and grow their business. And ideally, I think, you know, make their customers more and more happy. KB [00:48:52]: And there you have it. This is From the Field. Stay tuned for more.

Other Episodes