58:38 Webinar
Data Intelligence Impact: Know Your Data and Trust Your AI
Today, AI has to reason and act on a fragmented data estate that most organizations struggle to understand: what is this data, where does it live, can it be trusted, and who (or what) is authorized to see it?
This webinar first aired on September 3, 2026
The first 5 minute(s) of our recorded Webinars are open; however, if you are enjoying them, we’ll ask for a little information to finish watching.
Click to View Transcript
Hello, everybody. Good morning, good afternoon, good evening, wherever you may be joining us from on this lovely Thursday, and welcome to another edition of Everpure Tech Talk series. Per usual, I'm your host, Jason Langer, and today we're gonna be talking about know your data and trust your AI.
And we've got a fantastic guest here, Ashish. But before we kinda dive into that, I just wanna do some quick housecleaning items. Obviously we wanna try to keep this as interactive possible, so if you have questions, please throw them in the Q&A window or throw them into the chat. I'm gonna do my best to keep my eye on that as well, as well as Ashish
and I go back and forth. So if you see me looking over here, that means I'm looking for the chat, make sure if I'm cov- covering everything that you guys have got to ask or to say about this fantastic topic that Ashish is gonna cover. And Ashish, without further ado, I'll let you introduce yourself to the folks that might not
be familiar with who you are and what you do here at Everpure. Yeah. Thank you so much, Jason. It's just an honor to be here. Good morning, good afternoon, good evening, everyone. My name's Ashish.
I recently joined, Everpure with the acquisition of OneTouch, where I was the CEO. Now I'll be the general manager driving the data management business unit. Our goal is to help you understand your data better. And with that understanding will come the ability to comply, govern, and secure that data so that you can make it AI-ready, in the sense that you have the right context.
And that's what we'll talk about today. So without further ado, Jason, with your permission, I'll just get into it. Yeah. I just Before you do that, I just wa- I, I made this joke and I wanna say it out loud is, like, for those that might be new to Everpure or maybe you're an existing Pure Storage customer and know about, obviously, about our,
our name change about, I don't know, was it six or seven months ago? You all are on the call with the f- one of the people that are primarily responsible for our name change, of moving away- from just being a storage company, but to being a data management company. And the, the, the OneTouch or the Everpure data intelligence stuff that we're gonna go into is a key aspect of that.
So y- I wanna make sure everyone gives, gives Ashish the credit that's dues because you are definitely bringing something new and unique to, to our product portfolio that folks might not just be used to. So with that, I'm now gonna turn it back over to you, Ashish, because I think you have a lot of great ground to cover. Yeah. Thanks, Jason. You know, it's a truly a privilege to be part of Everpure.
And as we were going through the acquisition discussions, even prior to that, one of the things that was really key to Itzik and myself as we were having most of the meetings was, is this the right company for us? Because we were growing pretty well, and as a startup. Mm-hmm. And what's really clear was that the complementarity of the products was so strong,
and the genuineness and the intellectual honesty of the people was so strong that it made perfect sense for us to become part of Everpure. At that time, Pure Storage. Yeah. So when the name d- change came around, it really talks to the point of where the puck was going. Yeah.
And to a certain degree, it's already there. And, the executives at Everpure really thought about that while they were thinking about the acquisition as well, and the name change obviously reflects- Yeah where the puck is going. Yeah. I- it's chocolate and peanut butter.
You know, it's a great example, two things that just go together. So yeah. Absolutely. I'm excited for it. I'm excited for it. Fabulous. So let's jump into it. Why is this important?
You know, for decades at end, we've seen the walled gardens of applications, and walled gardens really mean what it means. It's a beautiful place. However, it's a walled garden. You have data that sits within the application.
You have context that the application builds around the data, and you have processes that secure the data within the application. What's really changed is that those walled gardens are being broken down, as most walled gardens broke down over the years. And it's been broken down for good reasons, because now you need to unlock that data in a
way that it's accessible, not only to applications, but to people and to the agentic world as well. And guess what? If those walled gardens won't be broken down from a strategic point of view, AI is gonna jump in and go access that data anyways.
And the biggest challenge that comes with AI is that it needs to have the right kind of context underneath it. But the challenges that come along with understanding your data, because AI is statistically smart, but not semantically informed in the sense that it needs to first understand where is that data.
And we can all agree to the fact that data is fragmented across different infrastructures. Even at a small startup like OneTouch, we had data sitting in Salesforce, in NetSuite, in our HR management system. And I actually had to introduce a universal ID for employee to be able to connect all the data across all three systems.
And that's something that a data company is doing, and a small company in those days. Today, if you look at all our customers, they see this across multiple different sources of truth. And when you have multiple sources of truth, you lo- lose context. Similarly, unstructured data is growing about 90, 80 to 90% faster than structured data.
What is unstructured data? These are emails, PDF files- Yeah downloads from Salesforce. Right, Jason? We all- Yeah have done that. I download Salesforce into Excel file. I'll do my pivot tables.
But guess what? That's very sensitive data that's sitting on my computer, and probably I put it into Slack and sent it to someone. More unstructured data. Yeah. More copies of that unstructured data. And sometimes it's modified unstructured data.
So now you don't know how to govern that, but you don't even know if it's the right data that AI should be accessing. So the top three levels are by themselves a big challenge, but what becomes the largest challenge is the fact that- AI requires the semantic knowledge, the relationship between data entities to really make the right kind of inferences.
And I'll get into that a little bit here on this slide. With traditional solutions, which are very SQL query based, we'll go out and look at the data, and they come back and they say, "Oh, you know, here's your data set. We'll do a ETL on it or ELT, whichever one you agree to do." that's, extract tr- you know, transform and load. Yeah.
And the thought process there was good many, many years ago, and it was required- Decades ago, quite because of these application wall gardens, right Jason? Yeah. And, you needed to extract that data. Then you need to transform it and load it, and now you had, like, multiple copies of it. Which is the right copy to use?
And in doing that, remember what I said about the wall garden? That context was left behind in the application. Yeah, you've got the data. So AI struggled with the fact that it dun- didn't understand that somebody could be an employee of American Express and also a customer of American Express.
Right. Right? Because if you're a loyal employee, you probably carry a American Express card as well. And to that end, you need to know in the context what that query is all about. Is it my paycheck that I'm querying about, or is it my customer service on a American Express card that I'm querying about, right?
So you need that context, and these transitional tools that came along when you start thinking about them, they were just pouring information into the context engine, but they didn't have the ability to relate that information. Yeah. So as a result, as we've talked about, hope is not a strategy. Hope is not the right strategy for reasoning as well, and you need to have the ability to
give these LLM tools, these agentic AI tools, the context that's underneath the data. This is also the reason a lot of AI projects seem to be failing, right? Now, AI is going to make its way through the technology adoption curve. I would say that the chasm, most people will say it's still in the chasm. I think it's past the chasm.
It's in the early majority. Look, we are doing things with AI here at Everpure that are saving a tremendous amount of time for our teams to go out and do innovation with the time that they're saving. Now, you need to be understanding of the data to make those projects come together correctly, right?
And that's what a lot of analysts are talking about, is that the challenge on the AI front has been the ROI has not been certain, and a lot of the projects are failing 'cause they, people just don't understand their own data. And to that end, what's crystallizing is the fact that customers are now saying that they have to catalog their data.
They have to find where is that relevant data. Mm-hmm. Today, 68%, Jason, of all financial services data still goes through the mainframe. Wow. Did you know that? Like, AS400 type stuff? Even other mainframes that are- Yeah, wow.
That's- You know, because the AS400 gets to the mini side of the things. Yeah. But the bigger mainframes- That's my mainframe experience, like, in the early 2000s, so. Exactly, and you know, I've got job security. You know why? Because I'm a COBOL programmer, and I know- I'll bet you are those mainframes are not going away, right? Yeah.
And- They were in high demand about 1998 Yeah, indeed. And so outside of dating myself, uh- Yeah the fact is that, COBOL is still very, very important. But all those mainframe programmers are gone, so how do you get that information to come out? Not all, but a lot of them are gone. Yeah.
And getting them is pretty hard. But understanding the data across structured, unstructured, on-premises, in the cloud, data in motion, data at rest has become really, really important so that you can prepare the data in-flight, if you will. Not because of a ETL process, because you wanna do it in-flight and get the context
coming with it, get the RBAC rules, the, coming with it so that you can secure the data. And this requires you to understand the data very, very well. Yeah, you, I have to say this. I get nightmare flashbacks when you talk about the unstructured data when I was an Exchange guy. Talk about dating ourselves, and, like, people
downloading their own PST files, and I'm sure people maybe on the chat have seen that too. But it's like when you talk about, like, just this data just being everywhere and then governing it. Like, it's like stop downloading your PST files. Like, we need to make su- like, we can't make sure that, that, just that email's archive
walks off the prem or, you know, out of the business, I guess. But anyways, yeah. Yeah, I mean, the data's just everywhere, right? Everybody knows that there's copies of copies of copies. And, in my career, both at working at Pure, but also as a consultant or even being a
practitioner, nobody ever said to me and said, "Hey, our storage is..." or, "The amount of data we're storing is going down." Yeah. Like, I just, I don't remember anybody ever going, "Yeah, our storage, we're, we're, we need less this year." Yeah. "We don't need more." So.
It is just growing up, going up and up and up. And Andrew, thank you for that note. I will keep the COBOLCO cowboys in mind. I'm gonna be an empty nester soon, so just to save my marriage and not bother my wife too much, I might get back into it.
But so thank you for- I'm gonna get you a, get you a hat and a lasso. Exactly, right? Data wrangling, right? Yeah. So, you know, we announced this at, Accelerate in Las Vegas a couple of months ago, and it's super exciting. Everpure has been thinking about this problem
for many, many years. It started off with the whole idea of having a evergreen architecture, then creating this unified data plane, providing the intelligent controls to actually understand how is the data being used within applications. With the acquisition of OneTouch, now we are adding a data intelligence plane.
And you would say, "What exactly is it?" I'd say there's three really important things, and we'll get into data intelligence specifically, but from a architectural point of view, this is completely heterogeneous. So we work on PureStor-- Everpure as well as we work with any other storage device because data needs to be understood universally- Mm-hmm in a continuous manner.
Secondly- It is something that is not limited by the kind of data, the type of data. We look at all sorts of data. And third, we provide the context so that you can do the right kind of analytics, the right kind of, agentic AI on top of it, and the applications can get that much more informed about the data. And I'll just give you an example.
A major hospitality customer has a particular requirement to meet US compliance laws, and they basically say that if you're a US resident or a US, citizen, your data cannot be held outside of the United States by a hotel on their devices. Now, how many of us I was just in Belize, and the person made a photocopy of my, passport, made a photocopy of- Yeah my driver's license, and even my credit card, and I said, "You know,
there's a law which says you can't keep it here." And they're like, "Oh, we didn't know that." And then they sent a note out to their sist- to their, you know, general management, in the US and they said, "Yeah, you can't keep it there. You can have it there for a certain amount of time, but not keep it there." Yeah. Now, the customer really was having this challenge.
Now, y- airlines send information about flight attendants- Mm-hmm about, pilots, and they say, "Here's a credit card number, their passports. They're landing in the UK. Please have rooms for them and send them a van to pick them up." We've seen this on a day-to-day basis. Oh, yeah. Yeah. It happens thousands of times.
Yeah. But that data can't be kept there. So the c- the customer in this case, our hospitality customer, was using our platform to see data in motion and saying, "Look, there's PII data in here, and it actually contradicts the way the business should be keeping the information in the UK or not in the UK.
So please mask that data so that you can have the right kind of, policies that are met." So we get used for compliance, governance, security, and that allows us to get into this AI-ready data space because we provided the context in the situation. It's a US resident, there's PII data, and there's a law that requires you to not keep that data outside of the United States.
So that's how all of this comes together, and that's this orange layer that we're talking about. So in this example, it was email systems, it was storage systems, it was all sorts of, communications that were being watched by our platform, and then providing those actionable insights on an ongoing basis. Yeah, and I think that's a great example.
Sorry, should we go to another slide? Yeah, please. 'Cause, you know, we, obviously we're, we're thinking, we think of things from a, or I do maybe, as an, you know, as an enterpr- enterprise IT or corp- corporate standpoint, right? But the, the thing that you just highlighted, like, as a consumer, I would wanna make sure a
company has that, right? 'Cause obviously I don't have to tell you or, you know, anybody on the call about, you know, data hacks or ransomware. You know, it's, it's like you think about, like, where all your data lives and how well those companies are securing it, let alone if it's in the US or, you know, abroad. Like you talked about, like, there's so many other things around that other than, you know,
just AI, AI stuff that we're gonna talk about. But, like, to me, I hear that and I'm like, "That's amazing," because I've, I've had my data stolen before. I'm sure people have had their credit card stolen. You hear You think about that and, 'cause I've traveled to Europe a lot, and I'm like, "Well,
I'll always use a credit card because that way I've got more recourse," right? But it's a very amazing practical example. I mean, sometimes we focus on, like, our, our careers and our jobs, but when you think about just the world we live in, that's, that's amazing, at least as far as I'm concerned, so. It's true. It's true. You know, and before I ran OneTouch,
I ran a company called BugCrowd, and we were a crowdsource cybersecurity company, 250,000 hackers, because you talked about data hack, 250,000 hackers on our platform, and we would go to customers and we'd have our hackers, ethical hackers, come in and tell them where their security risks were. 80% of all of that was data sitting in the wrong place and singing when it shouldn't be
singing, right? Wow. What I mean by that, people thought it was encrypted sitting behind a firewall. And guess what? Somebody had made a copy and put it outside the firewall to share with their partner. Totally, you know, un- not a nefarious act, but they did
that and it was now available, and our bounty hunters are finding those kinds of instances and providing that guidance. And now we're doing it automagically, right? And I never thought I would use that word- but there we go. And the automated nature of this is we'll scan the data, and we'll provide you at a physical
layer the information about your files, and we'll give you hotspots on where there's AI risk. And how do we do that? We do that by creating specific scores about your data and looking at all sorts of data across structured, unstructured, on the cloud, on-prem, even in mainframes, and then
we create a landscape, a chessboard for you. And now that chessboard is rated so that you can take actionable insights on the really dark red colors so that you can apply that. Now, this screen was made before we were acquired by, by Everpure. So I think we need to change these colors to orange, which I will let our UI people know.
Yeah, they're probably like FE 5000 or a shade of that by now. But yeah. Exactly. Yeah. Similarly, you wanna know where the opportunities are because it's not about locking the data down. In, in fact, it's about unlocking the data, unlocking it in a responsible manner, in a
compliant, in a governed, in a secure manner so that you can make this information available for training of these LLM models and making the agents that much smarter. And third- Is explaining the data. This is where the context comes in, right? We started the company with the idea of enabling GDPR use cases, right?
For GDPR use cases, you need to know context. You need to know this is a cu-customer of the, of the company, and where all the data sits and why is it sitting there. And then you need to prove that it's being used in a certain business process because that means that it's being replicated in other places.
And all that context allowed us to ensure that our customers were staying on the correct side of the GDPR law. Yeah. It gave us this ability to build business relationships between that data so that you can make the right kind of choices if you're deploying AI tools with the explainability behind that data.
So, Shash, I'm gonna, I'm gonna interrupt you. Pa- Sorry, pause you. So I think the, the If you wanna backup one slide- Sure or two slides, if you don't mind. I think the red slide, I think that probably made sense. I mean, it made sense to me from, like, heat map, right?
Like, the darker the red, the higher the number, more at risk. What I'm just curious on and maybe spend a little-- Maybe I didn't get it here. On this blue slide, you've got these darker blues with these numbers, meaning this is better data to be using? Or maybe can you just give me a better- Yeah for myself, maybe- Maybe- a better
understanding of what this heat map is trying to tell me. Yeah. Think about it like this. If I'm in a finance organization, and on the red slide, if my data is being looked at by, let's say, the sales organization and the marketing organization, which happens to be invoices, sometimes it's okay, sometimes
it's not okay, right? Yeah. But if there's other financial information that is being looked at that should not be looked at, okay, our invoice sales should look at it. But if it's a submission to a SEC filing, maybe they need to wait before the filing is made to be looked at, right?
Because that's the whole idea, to be able to be transparent within the filing regulu- regulations. On this slide, the idea is that if you have domain information, if you can understand in that example of being an employee or a, customer or both at the same time, you need to be able to, in agentic processes, be say- saying, "What is the data for this
particular use case that is applicable?" Okay. Right? So you wanna be able to say, "These are all the customer, files." It's coming from your CRM, it's coming from your financial tools. It's also coming from, specific email chats with the customer, and make that all relatedly available so that you can make the right kind of determination of the use cases around
customer success and customer first, use cases. The- So this is really the, the payoff slide to the sixty-three percent of projects fail because the-- I forget what the, the ex- AI- 'Cause AI data wasn't clev- This is where, this is where it's like this is kind of the, this jumping popcorn where we fix that is kind of what I'm- Got it. Okay. That's right.
That's right. And then what this does is it takes that data to the next level and says, "How can I explain the HR policy document sitting here?" And then how it applies to all those employee-related files as well. Got it. Right. So- Maybe that was clear for everybody else in the audience, but I'm-- I just wanted to make
sure I wanted to tie all those together, so. I really appreciate you asking- Yeah the questions, and keep them coming. And you're keeping an eye out on the chat as well. Yeah, for sure. Yep. So please ask questions from there as well.
So one of our major insurance companies, and I can tell you, I learned COBOL programming working for an insurance company right out of college. And we looked at this insurance company as the gold standard, and it continues to be the gold standard. Obviously, getting into, into security use cases,
our customers are very careful about not sharing their names- Yeah for good reason. But the best practices are something they're very open to sharing with others, and here is a best practice. The-them being the gold standard in the industry, they were, many, many years ago, looking at how their data was gonna be used by LLM models. And what they found was that they had a lot of
production data sitting in non-production environments, right? Now, all of us have this. Why has it become more important to keep an eye out on this? It's because AI tools are indiscriminately- Yeah looking at all sorts of data and pulling that in. Now, let's think about it like this.
You've got a non-production environment. I do this all the time. Or not anymore, but I used to do it all the time. I would change the data so that I could see if my logic, application logic, was working. Now, if that changed data gets into LLM tool, you start to see inference challenges because
it's not the right kind of data. So this company used our platform to not only understand their data across structured, unstructured, o-on-prem and in the cloud, but then they also said, "Where is this non-production data?" And start creating enclaves and tags saying, "This is non-production. Don't let it go into AI." And now with
Everpure b-having this full capability of storage all the way to discovery and classification and then even vectorization of the data, is creating enclaves of these data, sitting so that AI tools know where to go, which enclave is the best- Yeah enclave to go to and where not to go to. So it becomes a really important thing to get total visibility of your data so that you can
make it compliant. You can then make it AI-ready as well. So how do we do that? We first discover the data, as we've talked about. It's continuous and it's dynamic. It's not sampling. It's complete.
We look at the entire data. Now, we have algorithms that are- Built over the years, so they can scale to exabytes of data across our customers. It's already working at large enterprises, financial services, hospitality, retail, government today, that we can do discovery and classification across plentitudes of data.
Right? We then classify it for what are the specific sensitive elements in that data, because not all data is sensitive. In fact- Sure there's many studies that say about 15% of a data landscape is sensitive, and especially if you start to de-dupe that sensitive information, take all those files, Jason, you and I download from Salesforce so we can do our analytics on where sales deals
are going or opportunities are going, you know. You can take all of that out, and it's 15% that's really important. Wow. But understand- I'm surprised that number's so low it, it's amazing, right? Because- Yeah a lot of that data is still sensitive, but it might be dark and wrought and not being used.
Yeah. So one of the things that we give as, one of, my colleagues says, as a outside of the real platform value, is we'll tell you where's your sensitive data and when was it last accessed over the last number of years. So you can actually say, "Is it really necessary to keep that data- Yeah in the form
that it is at, or should we just archive it so that it doesn't create a issue for all those ethical and the non-ethical hackers as well?" Right? We then Kontxtual- That goes back to my comment before of, like, nobody deletes stuff. But, like, if you don't need it, like, you're, you're, you're limiting your risk, right? Because if it's not needed or if it's not accessed- Yeah why keep it?
So. Reduce your attack surface. Yeah. It's very important to continuously reduce your attack surface, but not do it in a way that reduces the ability of the surface to give you- Right better business outcomes. Yeah. We then contextualize it for that reason, and
this is the secret sauce. From day one, the company started contextualizing information. We said, you know, "Who owns it? How is that ownership, of the data manifests itself into domain information, financials versus HR versus sales versus, other functions in the organization, and who should be
accessing it or not accessing it?" But most importantly, we start creating relationships. You know, Ashish, the brother of Atul. Atul happens to work in financial services and has certain kinds of information, access to certain kind of information, so I shouldn't be making those kind of trades. And specifically in healthcare, it becomes so important because- there's
hereditary diseases, right? Yeah. Now if you can create those connections across an insurance company, you can provide, proactive engagement in terms of addressing hereditary diseases. So our platform is being used to provide these actionable insights because we create this knowledge graph that allows you to make decisions across relationships between the
data and the entities within the data. Not, doesn't have to be a person. It could be a, piece of hardware. It could be a valve, in GE's case, running underneath a dam, and understanding if that is working correctly or not and people are accessing that valve,
because you need it to work when the floods come, right? Right. And so you, we tag and label all of this information, and then we automate all the reports underneath it so that you can make those decisions on a ongoing basis. What that allows us to do is address multiple use cases within the journey customers have for the data.
It always starts off with the fact that you want to have total visibility, but that's not where the sale happens. Oftentime the sale happens because there's a particular issue. Right. There's a audit that's coming in. There's a compliance- There's a sort of requirement that's sitting out here that needs
to be met, right? Exactly. And so, you know, one of our customers, massive financial, services com- customer, was getting about 700 DSARs, and you'll see that under Kontxtual compliance right there, right? A DSAR is a da- data subject access request, part of what GDPR did.
And you can say, "I'm a consumer. I'm a person. Send me all the places where my data sits. And by the way, prove to me that you deleted it if I requested you to delete it." Oh, wow. Right? Taking them 21 person-weeks to do that. 21 person-weeks.
With our platform, because we were able to get all that information and we create this knowledge graph of Jason's information sitting in these 19 systems, accessed in these 18 ways, and give that as a report, that 21 weeks went down to 30 seconds. Wow. 3-0 seconds. Now you can think about the productivity gains for that group because they were now being
able to address these DSAR requests. So oftentimes it's one of these use cases that come in from compliance, a security response or a governance requirement. But what we're seeing is that AI has become a board mandate, a executive mandate, and people are moving to it so fast, and they don't understand their data.
So oftentimes that AI mandate, it requires customers to start thinking about their total visibility of their data so that they can understand what are the enclaves of information that can be used in AI, and how can you responsibly make it more available. And today with tokenomics-- Sorry, I was diving, so I caught something. Oh, no. Last week.
And with tokenomics, you want to manage the way you're using tokens. So you don't wanna throw so much data at it- Right which is irrelevant. Yeah. So you wanna understand your data and put the relevant data with the right context to get the right inferences. So this is the journey, folks, that most of our customers have.
We've got close to 50 Fortune 500 companies already using the platform. We've got a lot more commercial companies as well using the platform. And it would be nice, Jason, to do a quick poll- For sure in the background and understand how are people using their data today, and what is the place they are in the journey.
And it's pretty important. It's not a judgment, it's just where do you sit, you know? Yeah, it looks like most- Often times, I'll put this up in front of our customers, and our customers, being humans, also are pretty self-critical, and they understand that they don't know the total visibility. What's the blast radius?
What's the attack surface that they have? So that it allows them to under- you know, a expert is a person who knows what they don't know, right? Yeah. And they now know what they need to fix, and our platform allows you to fix that. Yeah. So it looks like the poll is up.
I hope everybody can see it. Apparently I can't vote. It says hosts and panelists can't vote. So you and I can't vote, Ashish, but- Yeah uh- We won't, we won't bias the answers then. So. Okay. I think that should be up.
If y'all can vote, that would be great. Laura will produce the results for us, here in a second. That sounds perfect. So in the meantime, I'll just keep going. Yeah, let's do it. Is that okay, Jason? So once we discover and contextualize the data, how do we make it AI ready?
So that first two steps, we just talked about in a tremendous way. Then comes the other products within Everpure, which is data stream, that allows you to prepare the data and vectorize that, and then bring security for those vector data sources to continue, and that allows us to do better training and do it at speeds that only our storage can enable through FlashBlade and FlashArray, so that you can get
not only the training done right, but also the inferences as well. Yeah. Now, what's comes down to the key point here is you have to understand your data. So discovery and contextualization becomes that much more important, so that you know what to prepare, you know what to train the systems with.
And how do you get the right kind of inferences is those Kontxtual knowledge graph, elements that we provide. So I'll keep going on building out why AI-ready data needs context. So we talked about this whole idea of a physical layer. This is what most of our customers already have a pretty good hand on.
Yeah. Which is, where exactly does this file sit in Pure, Everpure storage devices, right? That's because we have Pure1, we have Fusion, we have Purity. That gives a tremendous amount of knowledge out here. In fact, this was very exciting to me because we're taking this knowledge, buil- building it
into the data intelligence about the data inside the file, and allowing for some insights to come out that our customers can use. The next step is, is that a file that is a invoice or not a invoice? So it gives you ba- domain, information as well. This is before we've even opened the file.
We can tell you that these files are sitting from the financial or department's, domain, and probably should not be accessed by people that are not in that department- Right for example, right? We then open the file, and this is why you don't have to look at all the data, because only 15% of that has sensitive information.
We open the file, we look inside and we say, "Look at the amount of PII that's out here." And that's called entity-level understanding of the data. And Jason, you had said the audience is gonna be fairly technical in addition to business oriented. Yeah. So I'm getting into it a little bit, of the secret sauce here.
So the entity level is now saying, "Oh, you know what? I've got Jason's information. And by the way, there was a PDF where Jason's family member," or let's say your dad, "put together a insurance claim and put your name on there because your Social Security number was there." So now I can create these semantic relationships saying, not only is
Jason's information in this invoice that was paid out, but we also know that there's specific other family members in the data source, and we can connect that. So we create this one digital record about Jason and all the relationships of the data to that, right? To your entity specifically.
So now you can have this multidimensional graph that allows you to make the decision based on the use case. Whether you're, it's a u- employee use case or a, a customer use case, you can now make the right kind of decisions for the data that needs to be accessed to inform that- Yeah specific thing, right?
I just, I've, as you're saying that, I'm just, oddly enough that you brought up this slide, I was, I was just making a map. I'm just thinking of it like a mind map of, like, where you connect everything and then you click next, and that's basically what you've got up. So. Exactly. That's how I was processing it.
You know- Yeah you're thinking about all the de- the, I won't say dependencies, but the interdependence or these data markings that land everywhere, and then just trying to understand where all of that is. Like, doing that without a, you know, to your word, automagically, I, I can't imagine, you know, doing that. Like, I've been in infrastructure for a long
time and you can kind of find some stuff, but, like, to the level you're talking about, you know, you Just something that hasn't been done before, at least that I can recall. Exactly. And the thing is, now, now you take it back to, you know, the example I was giving for that hospitality company. So yeah, we created the file, which was a email.
It had PDFs with, credit card numbers. It had information about, people's Social Security numbers and national IDs, right? So you connected it to the sensitive data. That's a pretty big win to begin with. Then you bring to it specific predetermined policies that we've inserted into the
platform which says, "Hey, here's a company policy, here's a governmental policy, and here's the risk attribute we don't want to take on as a company." And you insert all of that information with the business process and the user domain into that, decisioning tool, which happens to be a AI, agent, and you allow them to make decisions and inferences in situation.
So think about it. If I was creating a loyalty program, right, and I'm a consumer organization, yeah, I can do a bunch of sequels, searches, and I can say, "Okay, I've got these five segments, and these two- Yeah segments are on the higher end, and I'm going to call them platinum and gold. Everybody else is called silver and bronze." And you know, you're done.
But how good is that if you're actually making a reservation on a e-commerce site on my company, and now I need to know that Jason prefers to sit in an aisle seat, happens to be a US citizen, therefore, when I send the information out to a hotel to say I'm making a reservation, I cannot add certain elements of that data. All that becomes something that's in situation, in motion, decisions that are being made
because you've got this knowledge graph available to you, right? And that goes back to your comment about you said in flight or real time previously, and that's a perfect example where you've got these policies. I'm assuming that you've got these policies set up so when that does happen, it's not a day after it happens or whatever, they're like, that comes in and they're like, "Nope, we
can't see his name," or the, the, to your point of, like, maybe his favorite seat or whatever, because that information isn't allowed to us. Exactly right. Yeah. And especially in situation, right? So that's, really important, and now people are making decisions in nanoseconds, you know?
Right. That we used to at least when I was in college, I had a lot of time to make decisions. My kids are making decisions, like, on the fly. And kind of funny, I was just in the all hands meeting, and my daughter sent me a note saying, "I'm flying to..." She's doing a semester abroad in Spain, and, "I'm flying to Ibiza to
go see these two concerts, and can you ex- give me my, frequent flyer number so that I can make all the reservations the way it should be made?" Right? And this should just come automatically to her. Right. It doesn't need dad to get involved in this situation, right? So I'm, I, I really think that this is the future of how data is gonna be managed is this
multidimensional knowledge graph. And you hear a lot of it. We've just been doing it for three years, and so now we've built some expertise that's available to all of our customers to make that available. Now, I did see that our audience here actually responded to, the survey, and it ac-
or the poll, and it's actually across the board. It seems that all those five use cases are important in terms of total visibility and compliance, governance, security, and AI ready, which is fantastic because that's how, you know, we are helping our customers understand the value of our platform. I think we should launch the second question as well is if you had to prioritize it- Yeah
how would you prioritize it today? So I'll let, let you do that while I go to the next slide. Yeah. Ms. Laura will bring it up for us. Thank you, Laura. All right. So I gave this example of, you know, folks who are in the United States and follow
football, which I didn't until my kids followed football. Tom Brady is a pretty important person in NFL football, has been for a long time, right? Happens to be here from California. I didn't know- Yeah he went to a rival high school of ours. So, our kids were always looking to beat that rival high school.
Not that we are a competitive family. But Tom Brady's information, if you're a part of an insurance company, is, pretty critical information, right? Because, you've got all these betting sites, and if you find out that Tom Brady's having, God forbid, a knee operation, you know, he, and he was playing at that time, that would be something that could be
transactable information. So our customer said to us, "The NFL is asking us to manage the information, but you need to tell us where all the NFL data is so that only a certain group of people have access to it. So you need to discover this information, you need to classify it as it's NFL, and you need to make it available." So they saved a tremendous amount of money from first deduping
all the sensitive information for their cyber insurance. Another byproduct of our platform. We didn't realize that in those days, I'm talking four or five years ago, that cyber insurance was managed by amount of PII data that you say you have. If you have- Oh, interesting significant PI- PII data, you pay a higher premium, and we
were able to tell them that there was a lot less, you see the numbers out there, that allowed them to reduce their premium, which saved them more money than what they were paying us as a software, which is pretty cool. You always want the customer to win. Right. You want it to always be-- You don't want it to be the reverse of that, for sure.
Yeah, exactly. You know, you want the perceived benefit to be way higher than the perceived cost. And in this case, and most of our customer cases, that's absolutely the case. In addition to that, they were able to win the NFL because they were able to say that we can control the data and only give access to the right people.
So these are compliance, governance, and security use cases already, right? A beautiful return coming out. But what was pretty important, and I-- this isn't, this, Tom Brady's just a name. This is not a situation in his family. But what happened to them was because we were relating this information, and hereditary
diseases happen in populations- They were able to take the information about a, a estranged father who ended up having a hereditary disease, unfortunately, and apply it to one of the players because they were able to relate that, that those two were connected together, and they were able to do certain tests and mitigate potential future problems in situation.
And that's what AI tools do, right? Wow. Yeah. And the fact that our platform was contextualizing and connecting the data for them and enabling them to understand the relationships, if you remember that canonical model, the semantic relationships, and adding to it this new kind of attribute, which is hereditary diseases, and relating the data for them was amazing, right?
Yeah, that's- Now, our platform- pretty mind-blowing enables that. AI takes that and does the work behind it, but that ability to connect context to data is super, super important. Does that make sense, Jason? And- No, it did. And it-- And I think it's a great analogy or
story or way to, to contextually bring back to what you talked about, again, how AI projects fail because you haven't classified the data. To your point, you just said, at least make sure I'm understanding this right, all these connections you-- we were able to, to make, but obviously AI was still used to, like, find and put it all together to say this hereditary thing.
Like, so it's, it's, it's a layered approach where it's like you still have to classify the data, which is basically what we're talking about over and over here, to make AI useful. Without it, you know, that doesn't get found or maybe it hallucinates and it, it's going- Yeah to other people that aren't family members. I, I don't know if that's a good example, but so just getting that classification of like,
to your point, knowing the data so the AI tools can actually do their job, again, in that example, that's pretty amazing, you know? And then contextualize it on top of it, right? Right. The context is king in these days. Yeah. Okay. I will just say I thought you were
trolling me a little bit because, the Tom Brady picture because, my Seahawks lost to Tom Brady in the Super Bowl. Oh my gosh. I didn't mean to do that. I, I, I brought that up. I'm like, "Why is Tom Brady on my screen?" But anyways. I am football agnostic. Okay.
All right. So, and I would say even ignorant to a certain, level. Okay. So how does this go? And this is my last slide here. Hopefully folks are getting some value out of this, conversation.
When you think about AI-ready data, it is, and I've seen this for the last three and a half years working with large, large enterprises, it's not a decision that is made by one department anymore, right? It's a alliance decision. I was at a major beverage company, and I know the CI-- CISO in that case personally.
He's been my customer many, many times. And he said, "Listen, Ashish, do I have the budget to buy you right now? Absolutely," right? "I can just-- The value that you provide is huge," right? "I need to bring my chief data officer in and my CIO in-" Yeah " so that they can get the
same benefits I'm seeing out of this." And those benefits are specific to the use cases each one of these personas are having. The CISO wants to understand the data, so they know what to secure. Most data leak prevention tools have a forty to fifty percent error rate in the discovery and classification.
So as a result, CISOs can't go to the CIO and say, "I'm gonna block this data," because the business user is gonna get ex- upset at the CIO, right? And now the CIO and CISO have to work together to have a great data discovery and classification tool that informs the decisions for the DLP tool to do any kind of blocking.
So as a result, you know, CISOs work very closely with us to reduce their risk, to ensure their security engineering is done well. CIOs use our information or the intelligence to make the right kind of infrastructure choices. And CDOs are thinking about, "How do I take that context and utilize this
information that is responsibly made available by my CIO, CISO using the Everpure Data Intelligence to make the right kind of AI, investments so that I can get the right kind of inferences?" So is it a complicated thing? No, it's not, because AI is driving a tremendous amount of activity at most companies, and all three of these, personas are being requested by their businesses to go
out and do responsible AI. Secondly, what does happen is that you actually get much better alignment when you understand your data across all three of these personas, and they utilize it. Now, at a major bank, we were brought in from the chief data officer side of the site because they wanted to make sure that they were auditing sovereignty, right?
'Cause there's AI sovereignty laws saying that- Yeah LLM tools sitting in a certain country can only access certain data in that country and foreign AI-- LLM tools in this rule are not applicable and should not be used. So they were auditing this as a global bank, and they were auditing this. They took the results, and the CISO looked at it and said, "My gosh, you've got information
about your data, about our company's data that I don't have, so I can't make the right security choices." The deal tripled in size, and now the CIO comes along and says, "Oh, I can utilize that information to make their infrastructure that much more AI-ready as well." And that doesn't mean just securing it or locking things down. It's actually unlocking it in a responsible manner, and it allowed all three of them to
make a decision that enabled their different use cases to work together. So with that, Jason, I'm gonna hand it over to you. If there's some questions, answer those questions. And you've been asking some great ones, so you can jump in as well. Yeah. The audience th- this
morning is a li- little shy. We haven't gotten any questions in yet. So feel free to-- We've got some, some time to- Throw something in the chat or in the Q&A. One thing that- I wish I would explain things to my wife this well, because every time I explain something- she has 100 questions, so.
I mean, one thing that maybe I'll seed it with a kind of a dumb question, and I know we, we had some poll results and maybe we can this will kind of fold into that. If I'm somebody that's, that's hearing this or just starting out, like, and I've, I've got these AI projects, and again, I don't wanna be part of that 63% that fails, right? What's, what's the step one to, to moving forward to like even under- understand that I
have a problem and that like maybe give me the baby steps of when somebody 'Cause you had the slide with like the five, five columns, and maybe I can find that really quick. Yeah. Like this one, like, you know, if I'm in the discovery phase, like where, where do I spend the time and how do I move from these? Is that an easy question or is that a tough question?
It's a very easy question. Okay. Meaning, I think, folks, even from the poll res- results, at least 30% of the folks feel like they have a handle on their data. Yeah. But others, including the 30%, I can guarantee you when we, do a scan of your data landscape, you find a tremendous amount of
information that you didn't know. One of the fastest, decisions that was made by one of our customers was we scanned the environment, and we found board member stock grants sitting in PDFs in Google Drive. Wow. This finding went up to the board because it was a audit requirement,
audit committee requirement. Decision came back, "Buy the platform and start scanning this information." The point here is that the platform gives you a tremendous amount of understanding about your data landscape. So take that total visibility step, contextualize that data, it happens automatically, and now you can make the
decisions on how you want to make that data available to the relevant AI use cases. Because you want to also manage your cost of AI, tokens, if you will. So putting all the data in doesn't fully make sense, but you also need to give it the context so that it can make the right inferences.
So total visibility is a great place to start. Okay. And then I'm gonna, I'm gonna ask a question. I might get myself in trouble, I don't know. But if I'm a us- if I'm an existing Everpure customer, maybe I'm not an Everpure customer, you did mention that we're the storage agnostic for the Everpure
data intelligence piece. Absolutely. That's kind of what I took from it. Like, if I wanna see this in action or kick the tires in my own way, is there, is there a way for me to get a trial or do I reach out to my VAR of choice that works with Everpure? Or if I'm an Everpure customer, and do I reach out to my account team? Like, what would be Like, how do I- Yeah go see this or tr- kick the tires on it for like
a better phrase? Yeah. So, the first thing is, on your trouble point, you won't get in trouble because the strategy of the company- I didn't know. Yeah. The strategy of the company from the top down is to make our customers successful, regardless if the data sits on Everpure or any other storage or in the cloud or the mainframe.
We want our datas to un- our customers to understand where their data is and understand the data so that they can make it responsibly available. So 100%. If you're a customer, there's a lot of value because we've created linkages right into Purity and Pure1, so that you can get the information.
If you're currently a prospect, which I sincerely hope you look at, the Everpure offering and in the continuum, understand your data, understand that you're gonna be putting it on the best storage out there, and you're gonna get the software working for you so you can automagically, again, get those, those actionable insights. In terms of your question, in terms of ours, we are a c- chan- channel always company.
So our channel partners, our partners are going to give us, the ability to respond to your needs through your partners as well and directly to our, sales folks. And yes, you can try out the software. And we'll obviously help you prioritize where you want to focus on things so that you don't get too many insights as well. Yeah.
Because if I had to put a poll up, there's four reasons people feel like they need to make decisions. You know, AI ready has become one of them. It used to not be, even a year ago, it wasn't the top, making the top. The second one is reducing risk. The third one is making sure that they can reduce time to market.
They wanna get those insights out so that- Yeah it can be out there. So- It's a competitive advantage, right? Depending on your industry. It's a competitive advantage. And the fourth one, which is actually the most important one for most of the folks making a decision, is they wanna be able to make prioritized actions come together.
So they want insights that can allow them to prioritize the actions that they're taking, whether they want to make a certain data set available, not available. If you want to stop a certain file from leaking out of the company, you know, you need to have those insights. So if those four come together, you need to have visibility so that you can make those
prioritized actions. You can reduce risk. You can take the data and build the right kind of time to market for anything that you do internally or externally, and most importantly, address the right kind of data comes to build the right kind of inferences at the right cost efficiencies if you're deploying AI tools. Wow. I think-- I'm looking at the time.
I think that's a perfect way to wrap. I think that was a perfect way to summarize, and that was a great bow, 'cause I'm also just being aware of the time. Absolutely. But I think that was a great way to wrap it up because, I mean, I just-- I, I don't think we can say anything better than than that on why, why the, the value and the,
the importance is there. Ashish, I do wanna thank you for joining us. I wanna thank the folks that are on the call that joined us that, that made it to the end. Hopefully you f- I try to view this as edutainment, so hopefully you found it very educational, but also I got some, some laughs or some entertainment out of it as well on
your Thursday morning. But if you wanna know more about Everpure or Everpure Data Intelligence specifically, I've got up on the screen, we're doing a roadshow for-- or Everpure Summit, sorry, lack of a better word. We're visiting Dallas, Chicago, and Boston. So if you wanna know more about this or anything else that Everpure is doing, scan the
QR code. You can register for either one of those three cities. We'll have folks there, presentations, demos, all that great stuff. Maybe Ashish, I don't know if you're joining any of these, but we'll have AI folks there. If you're not, but hopefully maybe you're at some of these.
You did mention travel, so- Yep I know you're on the road. And then finally, make sure you join us on our community site. If you're an existing customer or still, you know, maybe you're a prospect or you're looking into it, go to our customer community. We've-- You hear from other customers, ask questions.
I monitor some of the forum posts. You know, we have experts are for AI, databases, you know, backup and recovery, resilience. So it doesn't have to just be AI. Make sure you check out our community page. Lots of helpful information there, as well as kind of a, an access point to Everpure Pure//
experts or folks that can help you out. I will say do not treat this as like if you have a PRI One production issue, that is not a replacement for support. Please call one eight hundred, you know, Everpure storage or whatever for support numbers. But community site, great place for good
resources and just having an open conversation with Everpure employees as well as other customers. But again, like I said, Ashish, again, thank you. Thank you to the audience. Really appreciate it. And I hope everybody enjoys the rest of the day and the rest of the week.
Three-day weekend in the US coming up, so that can't be all bad. That's incredible. But thank you for the op- opportunity, Jason, and I really appreciate the audience, staying with. Thank you. Yeah. Thanks, Ashish. Everybody else, have a great day.
Like I said, enjoy the rest of your week. Cheers. Talk soon.
Ashish Gupta
General Manager, Data Management BU, Everpure
Jason Langer
Senior Technical Evangelist, Everpure
Today, AI has to reason and act on a fragmented data estate that most organizations struggle to understand: what is this data, where does it live, can it be trusted, and who (or what) is authorized to see it?
The answer to all these questions lies in data intelligence. By embedding discovery, classification, context, and governance directly at the data layer—at the source—data becomes self-describing and trustworthy wherever it goes.
In this webinar, you’ll learn:
- What understanding your data actually requires: discovery, classification, context, and lineage applied at the source
- How source-level governance flips from compliance brake to AI accelerator, especially for agents acting autonomously
- A way to assess whether the data your AI reasons and acts on can actually be trusted
We Also Recommend...
Personalize for Me
My Updates
No updates at this time.