Name: Ask Me Anything: DatasecAI 2026
Uploaded: 2026-10-13T19:15:46.172Z
Duration: 45 min 44 s
Description: Ask Me Anything: DatasecAI 2026

Transcript for "Ask Me Anything: DatasecAI 2026": Perfect. We're live. Hello, everyone. Thank you so much for joining. We're gonna probably, kick things off in a second, but thank you for joining our ask me anything webinar. It's the first webinar... The ask me anything series. We've got lots to talk to you about today. Really excited, but we're gonna give it a couple minutes, let people join, get settled in, and then we will introduce ourselves. But thank you very much for for taking the time out of your busy schedules today to listen to us ramble on, talk about AI and data security and the importance of upskilling, and maybe the opportunity to grill us and quiz us in this fun format. Okay. Hopefully, everyone has had a fantastic Tuesday. I'm ready with a bunch of questions for myself and, and Shane. We're very excited to talk to you today about this. Okay. Well, it has been a minute, so let's kick things off. Firstly, thank you very much again for taking the time to, you know, listen to us talk today and to ask us questions. This is a part of our ask me anything series. I'm really excited for this. A big fan of Reddit, have been for years. So I like the ask me anything format. I feel a bit like Matthew McConaughey here. This is my celebrity moment where I get to be grilled onto my area of expertise. And we can talk about the data sec events that are coming up. But myself and Michael Ferguson, everyone calls me Fergo. I spent far too long in Australia and came away with this, nickname, but I am the principal data security services architect over here at Sayera, and I head up the AI security school, which involves all certifications and some of the live ones that will get a DataSec AI. Shane Coleman is on the call as well. Could you introduce yourself, please, Shane? Absolutely. Welcome, everybody. Shane Coleman. I'm the chief data security evangelist here at Sayera. I spend a lot of time with customers, prospects, and our channel, making sure that, we are, enjoying the value that Sayera delivers, that we, build credible outcomes that our customers can measure, and, very excited to participate this as well. I think this is my first time, on the, on this channel as well. So You feel like a celebrity invite. as well. I do feel a little special. This is great. Okay. Perfect. Well, look. Let's get started. Talk a quick agenda. So, basically, what we're gonna talk about for the next forty five minutes, quick overview of, the event. This is... We're talking about DataSec AI. This is our third DataSec AI event, which we're really excited about. We we love these these, sessions. We get lots of people together talking about this really interesting area in security, the emergence of AI and agentic and data and identity and non humans and all this kind of fun stuff. How can we secure it? How can we be the, sort of moral compass in in this and try and fight the good fight to protect our organizations? We've talked about the agenda and the events, the talk tracks, the workshops are gonna go, those breakout sessions. Marie's gonna talk us to... Through us through that. Then we'll talk a little bit about the certifications. We are delivering live certifications. Very excited about this. Last year, we had, 250 people in one room for certs, which was incredibly overwhelming, but, fantastically rewarding as well. And that's an opportunity not just to learn and upskill, but to get some CP credits as well. And then we're gonna talk to you about to you about how you can earn and gain a free pass to, the event, how you can actually get some of this paid for, which is, you know, a credible opportunity and get a nice trip to Dallas and get to maybe do some certifications with myself and learn from the great Shane Coleman. And then we get to the main event where we ask a bunch of questions and start grilling us on what we know, or if we're big, phonies, I don't know what we're talking about. So really excited to get into it. Okay. So perfect. What we're going to talk about now, I'm gonna ask Marie to first introduce herself and then talk to us about what what is DataSec AI. Why is it important? Thanks, Fogo. Marie Reyes here, Head of Global Events and Executive Programs. Super excited to be here and answering everyone's questions. So as Fogo mentioned, this is our global conference that we have annually. This is our third year hosting this event. It's a three days packed full of content, education, certification, CPE credits. And also we've sprinkled in a bit of fun there as well, and we'll also have our sponsors there, and we've got some great speakers lined up throughout the three days. So the dates are November ninth to eleventh, so it's in a few weeks. So for those of you who haven't registered, please go ahead and register, and it will be held at the Hyatt Regency in Dallas, Texas. And, also, this is a free conference. So this is free for all of the attendees. Fogo did mention we do have a competition where we will be comping travel as well for those who, are eligible to be in the draw as well. So, again, three days full of fun and education that we have planned, for this conference. Perfect. And, also, just if you look in the, in the webinar, in the docs section, you should see a register for data sec ai events. So if you click on that, you'll be able to register now. If you haven't already, just click on click on that link. Okay. So, let's do a little, introduction about what what we're gonna be providing you with. As opposed to just coming here and listening to people talk, we want you to have some takeaways. It's that this area of, security, and we mean that by, like, AI and agentic and and, nonhuman identity and how it couples with data and data life cycle. This is all fairly new. It's always evolving. Like, no one on there had in their, yeah, job description AI security architect, you know, maybe three years ago. This is all fairly new. Most people wouldn't have had this. So there's a lot of skills for us to, to already start, you know, learning about and thinking about. So we wanted you to not just listen, but come away with something. Come away. And that's what we at Sayra have been, delivering. We created an AI security school, which is a sort of, a training partner for Sayera, completely vendor agnostic, where we can start thinking about what are the problems when it comes to AI and AI security and how do we structure it and provide, you know, complex and really hard, you know, courses that allow us to get some hands on experience and get yourself certified so that you can walk away with a shared vocabulary with not just, yeah, the industry, but, you know, with your team around the areas of, yeah, the risk inside of AI and then become certified. We then provide you with frameworks, not just a structured approach for thinking about security in AI and whether it's generative and embedded or building a blended and all this fun stuff. But how do we match that with the, industry standards and frameworks that we've got in place? We'll just map them to the... You know, whether it's NIST or the EU AI act, the, you know, ISO, OOS, MITRE. There's all these different frameworks that are available. We wanna map that because we can talk about the theoreticals as as much as possible. But if you don't have the evidence, you're in trouble. You can't prove your level of maturity, so we wanna provide you with that so that you understand what I need to collect when so that I can not only ensure that I'm on the right path, but that I can provide the evidence when required. And, hopefully, by doing this in person, you get to learn from the people that are doing this. Like, one of the things that we started with when we were building this out is we had all of these office, the CISO, and we had services teams, and I sat there going, we've got about three hundred years of, you know, experience in this team. We should be teaching about evidence. We should start thinking about the, you know, the training people up. So all of these courses were built by myself and Shane and and the office of the CSO and Sol Rashidi and Jason Clark. We pulled all this brainpower together to build something that's gonna help you structure security for this emerging area in the best way possible for you, make it nice and easy. That's kind of my role in it, trying to translate it into something that's easily digestible from all these, super brains at the company. But that's what we wanna try to learn from the practitioners. So what are we gonna learn? Right? Basically, this year round, we've we've actually dropped the DSPM practitioner course, which is our data life cycle management course because we wanna focus on data... On AI. Sorry. And we've got two AI courses that you can take. So there's two tool tracks, and we're doing them twice in... You know, across the the the three days. You can do Monday or Wednesday. Just choose when is best for you. Maybe do both about that preferable for myself, But there's a big difference between security and trust. Security is essentially going, is that date... Should that data be in there to, ensure my organization isn't liable? Is it gonna be comparable to to, lawsuits or to, you know, sort of legal issues. I wanna be secure in in that I am not putting p I... PII data into that, AI system so that it doesn't generate outputs that are gonna compromise my organization. Yeah. That's, you know, data loss and all of those types of, issues when it comes to security. But trust is interesting because now we've got these incredibly autonomous environments or systems that are creating and doing work for us, and they're connecting to data and they're connecting to systems. So I've got to be able to understand the intent and the boundaries in which they should operate. And if they overstep that boundary, then it might be outside of the the trust decision targets that I set for it. And, therefore, maybe I need to, you know, escalate this to someone else or or provide some mechanism to stop the AI so I can provide... I know that it's always got this trustworthy balance and trustworthy boundary in there. So, it's making sure that I trust that it's going to be delivering on this intent. It's not really security, it's just it's like having a, a mechanism to make sure that my employee is doing what I set them out to do. So there's AI security, which is primarily where we've always been, the AI trust practitioner is is another one as well. So there's two different talk tracks, really powerful, really both interesting, and, yeah, we'd love you to sign up for for one preferably both, whenever you go and register for the course. Now as, we said before there are going to be the opportunity for five people on here today that that are gonna be drawn at random to have their hotel accommodation at their Hyatt Regency in Dallas paid for for those three days. It's gonna be paid for in its entirety. So, well, what's your inbox? We're not gonna do a draw. Okay. Yes. Lots of craps. Lots of love hearts. This is great stuff from, from, Marie. She's a very generous human being. I think she paid for it out of her personal credit card. She's she's that kind of human being. So, but keep an eye on your your mailbox, and, yeah, we'll we'll... Hopefully, you'll be one of those lucky winners to do this. Cool. So, if you haven't already, and you're okay with taking a picture of a random QR code in a security organization, take a screenshot here. This is, gonna help you register for not only attending the, you know, attending DatSec AI, '20 twenty six in Dallas, Texas, and meet us, but also, hopefully, to do some of their certifications and become upskilled and a, a big brain in the industry around AI security and AI trust as well. Perfect. I'll leave that there for a second. Hopefully, you've got that, and then I'm gonna maybe hand it over to Marie who I've already been told has got a lot of questions for us. So I'm I'm praying they're all for Shane. But so... But let's see. Let's see. Let's put this put this to the test. Let's ask those questions and see how. good. Yeah. I've got lots of questions coming leave this. up. I'll, leave this up. yep, as organizations move toward a Gentic AI that can take actions across enterprise systems, what governance controls need to evolve beyond those used beyond traditional generative AI? Great question. Okay. So, there's a couple of interesting ones here that I think are are fairly interesting. There's common ones that we'll have across the board. Mapping, what, know thyself is is really important, and there's a be a bunch of controls within NIST, and ISO that align to to mapping and understanding this. But, governance is interesting. So having, like, AI ownership, is is really interesting. So understanding intent. What if they... We don't have that for data necessarily, do we? We don't have that for endpoints. What is the endpoints intent? Well, it's what I do with it. But AI has to... I have to understand the intent of the AI because then I can understand if it's deviating from its original intent, then I can escalate to the AI owner. So those types of processes are really new and really interesting. You build this system and you wanna inventory everything, and then you wanna understand the intent of those different agents so that I can ensure that if they deviate from that intent, someone is going to be notified or some action can be taken. So that's absolutely something that's brand new. So, I think that's a really interesting one. The other one I find quite fun and interesting from a, you know, sort of governance is weaving in things like, use our pen testing. Well, now we have red teaming of AI. So it thinks kinda cool. Like, way, you know, myself and, Marie can sit there and and see if we can break Shane's AI agents. I mean, he's, you know, constantly vibe coding, building wonderful projects, in in himself. So United team to systematically go through, understand what the intent of those systems were, and then see if we can connect to a different types of data, see if it actually deviates. You know, the probabilistic nature of AI can mean that if it has been over scoped its access to data or to tools, it might actually, you know, do do more than it should have done. And therefore, that in in and of itself is untrustworthy AI. So having those types of mechanisms can be quite interesting as well. So, hopefully, that can answer your question. There, Shane, you'd like to add anything to that? No. I think... Listen. I think you did a great job. I'll share with you what a lot of peers are thinking about when they think of that question. Is, many... And I say many, more than a few, are talking about the next stage. So there's that framework of, let's be honest, you're just amplifying, all of the challenges that you've had prior to the adoption of AI. It just continues to go, down that path. But what's interesting is we're finding individuals willing to say, okay. Baseline, let's assume that I have a full understanding of my data, the agents that are interacting with it, the identities that they use, and I have a good understanding of intent and intent drift, whether that's a jailbreak motion, whether that's a, you know, a natural drift, or whether it's a shadow AI accessing the data. I wanna have that baseline. And then the next level, they're actually saying, hey. I'm gonna I'm gonna do a one, two, phase one, phase two approach. I'm seeing a lot of aggressive conversations around looping. Like, how do I how do I ensure I'm about to launch a multimillion dollar application? I need to make sure that the application has a level of controls that can govern. I hate to use the word kill switch, but for lack of a better word, depending on where the the issue is closest. Do I need to put the control at the data layer? Do I need to put the control at the endpoint layer? Do I need to put the control at the, LLM layer? Yep. Maybe the answer is is yes to all depending on on the gaps, and we want to leverage AI to make those decisions more responsibly. I find it's a very interesting conversation, I would be very honest. I have yet to see this really done yet. The conversations are taking place. But I think where that that question's coming from this is after the baseline, what's the next phase? What's the next phase? How do we how do we run at the same pace of AI? Or how do we make sure our AI applications are running at the same pace of of everything else that's AI related? Well, actually, on that, there's another question from Steve Gray. Sorry, Marie. I'm just jumping in because I think that's. quite interesting. Go for it. Yeah. Yeah. So Steve Gray has asked, how should CSOs balance giving employees and AI builders fast access to data whilst preventing autonomous agent workflows from overprivileged data exposure? Great question. And AI will very quickly, you know, catch out poor data governance because it has that ability to stretch far and wide across the organization, read everything, and provide you with, those insights, good or bad, biased or unbiased, poison or unpoisoned. It doesn't really know. So you wanna be able to make sure that you can move quickly, but you also be able to be cautious. And the way that we would think about this is, a really good approach, that I would take is trying to upscale as many people within your security team around AI and AI security processes as possible. Throw them through the certification so that you can start thinking about structuring your AI because not all AI is bad. Not all data is important. So having an understanding of critical data, high impact data, and the AI agents and systems that are mapping to them will allow you to understand what levels of governance you want to put in place. And what you can then start to do if you have the right mechanisms of great security tools, Shane will talk to you about some of those in a bit, you can start applying things like, policy as a code. This high risk, AI system that has access to high risk data, therefore, needs to be, you know, more stringent, might have to have more human in the loop mechanisms. Those that are just processing, you know, all of Marie's birthday invites because she's a very popular person, they're not that important. They're they're allowed to be as autonomous as possible. So you can understand the criticalities of the data, yeah, and the intent, all that stuff, but it's... I mean, it's a fundamentals of security. Know thyself. And if you can map everything and have an inventory of your AI, an inventory of your data, and map them together, you can start to apply some pretty quick policies to to allow you as quick as possible or as autonomous and as fast moving, as agile as we want to be in that space, whilst also being able to pull pull the reins when we start seeing, agents that are accessing risky data doing risky things. That'd be, yeah, great question. Thanks, Steve. We've got another great question that just came in too. What should a technical program manager like me be measuring and reporting to leadership to demonstrate that an AI program is not only delivering value but also operating within an acceptable level of data and security risk? Okay. Oh, I love. that one. Yeah. I'd love I'd love to take that. Yeah. Let. me got me. actually take that. one. No. Because you've been doing a lot of this yourself. lens. I've seen you've been mapping this. You've been... And you've got the... You've got the consoles for this. So so what I'll do, maybe I'll share a little bit of my journey, and I think that's, you you know, super powerful is through stories. And so when you think about us at Sayre, we also adopted, AI internally from an operating perspective. And, you know, we were asking some very serious questions like, what model should we use? Who should have access to those models and when? Are they using it responsibly? You know, caching versus optimization. At the end of the day, as a technical program manager, again, generally speaking, throwing through our lens is internal usage meant, optimization. Potentially, revenue generating activity if we could build something and model it and get it to the field quickly. But, really, your your primary focus was, how do I operate more effectively? How do I make sure that... You know, I'll throw this through the lens of our sales team. Again, kind of a sales motion here. But are we... Are deals progressing along? Are, you know, are are we preserving margins better as a result of this? Like, is AI making us better, or is it, slowing us down? And so being able to measure and have a baseline before adopting AI is what we used in conjunction with users of AI and how they used it to to to start measuring the productivity gains. And, again, that's just one that's just one motion. I can tell you some of your peers, if you have a customer facing application, like... I mean, you can almost look anywhere. I'll give a public example of the Adrian platform for, we lost him? Think we... Experian. That's tragic. Right? That's was really interesting. r... Am I off? Oh, I'm back. Okay. You're. back. You're back. sure where. I lost. on. you. Some some... Okay, some public examples are... Of of leveraging and how do you measure it are productivity gains. So look at the Adrian platform with Experian. I think their numbers are around 14% improvements by adopting a, you know, a customer facing application. So I think at the end of the day for you, it's get the baseline of without AI and then be really critical on the measurement. And don't expect gains in the first few weeks. Like, full transparency, it was really about three months before we could confidently say, hey. These are our productivity gains Yeah. internally. And that... It's... That's interesting because we, we're talking about the the productivity side of things. It's something that, in the certifications, perhaps, we haven't talked about. We spoke... Speak primarily around, maturity modeling in the various stages of the journey so you can set prioritization, to build a cohesive security plan, whether it's around discovery, classification, governance, protections, controls, and detection response, and then assurance and trust. But you wanna set, like, what's your baseline of where am I at right now when it comes to security? And based on whatever my organization is, my, yeah, my industry, so what what compliance standards do I have to adhere to? What's my budget? What's my risk appetite? What's the culture like? All these other things. I need to get to this level of maturity when it comes to classification of AI. I need to come to this when it comes to detection and response, you know, of AI. This when it comes to protect... Controlling and protecting the usage of my AI. And you wanna know where your current level of maturity is and your target level, and you have to track that over time so that you are, not only, oh, you know, getting... Progressing towards that target level, but you're not overstepping. You're not overreaching and slowing down the organization. You can build a slow cohesive plan that covers all gaps, but does it in a way that is also going to allow your organization to keep going and being innovative. I think that's very important. This now... Yeah. If you just slowed everything down, you're gonna, like, start to run into problems with the CEO saying, no. No. No. We need to be innovative. So I think that that's something that can be quite powerful. Using maturity modeling, we have a pulse check. If you go to sierra.com on resources, you can now go to the AI maturity pulse check, and start doing your assessments there and do certifications, and you can start figuring out finding out how to hit those target levels as well with what technologies, what people, and processes. So, yeah, that's a great question. Great question. Yeah. Yep. And, again, I would go back to the key is the key is the value you expect to get out of it. Because at the end of the day, you know, the control should absolutely reflect. reflect proper usage, not just single date... Data combinations. Like, are those actually, number one, within the confines of those regulations, do we have the ability to put the proper controls to ensure, you know, safe adoption? But then at the end of the day, is your business being more productive? You'd be amazed. at the amount of insights that you're able to gain, you know, inspecting this, like, looking at accountability, looking at visibility. Like, it's not just a security control. It's a it's an optimization conversation. You're gonna... You'll gain so much more insight. There's actually a part two of this question as well, which I'm I'm interested in. So as organizations move from AI experimentation, the hobbyist, This is the Shane Coleman's building new pilot all the time. Yes. To they're moving into production and everyone's got access. Collecting more tools, more data. That causes bigger headaches. The more utility you provide AI, the greater potential risk. Guaranteed great understanding. of the problems. So how do program leaders build security and governance into delivery without slowing innovation, and what two or three controls, do you believe are nonnegotiables? Now Oh, I'm. gonna give my... I'm I'm gonna give mine, and Shane's gonna give his. I'd love to put earmuffs on him so he can't see if we we hit the same. But the the the way I would look at this is, first things first, you need data discovery classification. Data is so important. It's so entrenched in AI because it's a generative tool, and it creates data all the time. So whether or not you are training your own models, probably... May... Maybe, maybe not, but also connecting up rag indexes and what's in that index and what classification and what boundaries you want for the AI. AI will expose poor data governance very, very quickly just by the pure nature, the probabilistic nature of models. So having a strong data governance and access to, you know, what agents that data that... What data that agent should have access to, imperative. So, you know, data classification and having really strict and cohesive across the board. Then a really strong AI inventory, Before we even start thinking about, do you can't do this, and I wanna do prompt controls and prompt output, observations and and drift monitoring and all that stuff, what have I got? What is there? Whether that is Shadow AI, I'm accessing HeyGen AI video and what, you know, creations or using ChatGPT or or turning on, Einstein or was it saying AgentForce. You've got GitHub's got its own and Snowflake's got its own, or maybe you've got a bunch of guys building their own agents with Bedrock and Lava and Hugging Face, and so there's a whole bunch. I wanna see everything I've got and what are they accessing... What are they getting access to. Only then... So that's the two the two pieces. That's a DSPM type of tool and then a I s p n type tool so you can map it. And then the next one is at at more of an administrative control. I wanna assign ownership to it, And then you can start putting in place programs as to say, this is what I want to allow, the types of controls for my high risk AI systems, my medium risk, and my low risk. You can tier it. Say we have a data classification, do it with AI systems, and then we can start to think about... Okay. Well, this is a high risk, but it's for sales, and they're going to make money. Therefore, I can allow this, or this is a low risk, and it's for marketing. You can do whatever. Yeah. We can be... Start understanding the type of governance we want to apply to our, to our systems once we've got those in place. So, yes, know thyself is always my most important, I think, when it comes to security. Shane, what do you reckon? You know, it's funny. I I think we're both gonna touch on the same things. And quite frankly, as we go down our conversations with our customers, it's really a great standard, that that we leverage, which is, to your point, first, is you've gotta understand what you have in your network. Like, you know, the question I would reflect on ourselves before I even get started in an AI initiative is, what value do you expect to get out of the platform? Like, are you plan... And I've I've heard some great stories. There was a a law firm recently, and the name escapes me with... They publicly announced a billion dollars in an AI initiative. Alright. Great. I know my North Star. So as we start to move from kind of, playground usage to proper adoption, is that our North Star? We're gonna feed everything. We're gonna expect to make money. We're gonna expect to save money. That really helps because what you do wanna avoid at some point in time, it's costly. You know, you'll you'll start to see individuals. When I talk about visibility, I'm not just talking about complex data combinations that are feeding AI. To your point, Virgo, I'm talking about what AI assets sit out there, what agents sit out there, what are they getting fed by the data, what, you know, what data is is feeding the AI, and how is it using that information? Then you need an audit in control motion. Like, quite frankly, like, I... Again, I'm not gonna use the word kill switch, but if I find something that's outside of my risk tolerance zone with AI, all of a sudden a drip. This is not proper usage of of the agents starting to do something it shouldn't. And individuals trying to jailbreak it, or, human interaction with data is being misused. Not only do I need to be able to audit, but there's levels of risk that you're not gonna be comfortable owning and just going back and manually addressing. You're gonna wanna kill it. You're gonna wanna address. it quickly, and then you wanna take action on it. So I go back to these things. You gotta have visibility of your entire landscape, agent intent data. Then you need some type of audit trail, and then you need some kind of accountability control motion. Those are the three things eye level can't go without. Oh, it looks like Bogle might be breaking out a bit. Might have lost him there. But, Shane, there is a question for you that did come in, through our inbox. The event brings together identity and AI security. Can you give an example of why those teams need to work together? I think... Oh, wait. Yeah. I don't know what was going on. I'm I'm back. Hopefully, you can hear me. Oh, here we go. Okay. Yeah. I don't think I'm on stage now. lost... Okay? Yeah. We just lost you there, Shay. Hey. No worries. I I promise I haven't pushed any buttons. Incredible. Yeah. No. Okay. So let's, let me let me see if I can answer that question. So so what I just heard is is why your data identity and AI security teams need to work together. and and. and and certainly encourage participation there. So let's let's kinda break it down. If you look at the AI teams, they're building the agents. Right? They're building the interaction, and they may even be building the platform. You may have multiple AI teams. Like, we... Our team, for example, we have an AI go to market. We have an AI platform team, and then we also have a data team. So we really have three different. departments working together. Then you have an identity team that is deciding what service accounts these identities, or these agents have access. And then you have your data teams and and, you know, potentially, security teams are are controlling the data and the classification and so forth. They're deciding to what to what extent we're gonna allow these agents to access the data. And now, by the way, let me make it even more complicated, data combinations. Data by themselves that may not be very critical, but together can amplify benefit for the business or risk for the business. It is it is, you know, each of these teams see a picture of the puzzle. So if something goes wrong, nobody owns all of it, and then they take responsibility in a fragmented view. Having all of these teams participate, And even if you're only from one organization and the rest of your team can't go, how amazing is it to attend Datasec and listen to other peers and how they're interacting and dealing with these challenges? You get to take some of that insight home with you and kinda help build a more robust security program and AI initiative than you had before. Right? That's my that's my take on it. It's great. Also, like... So, further to this, it's really interesting because, like, if you've got everyone speaking the same language, the same vocabulary, you're gonna build a much more cohesive, governance structure across across the board. You know, that you're gonna be becoming a AI AI security from multitude of different avenues of, yes, there's data. Yes. There's compliance. There's privacy. You've got operations. The organization wants to move forward and fast, and there's there's just... There's a whole bunch of different areas. That's another reason why upskilling your teams now with the AI security schools certifications, which again are not, you know, about Sayera's technology. It's just having a common vocabulary in these issues, will assist you in being able to build a cohesive North Star together and say, this is what we need. So when somebody says, I need to have, protection against prompt control... You know, prompt injections. I need to be able to look for, you know, intent drift. You'll know exactly what it's looking for. I need to be able to observe end to end, accountability inside of my, AI environments. Okay. Well, you you start to understand why that's there and what they're for, and that will help you to build a strong, you know, plan for moving forward safely and securely, as well. So that's something I think is really important. Great. There was one thing I wanted to talk on. Sorry. Quickly because that was a follow-up from it was Jerry d, Della, mother. But he was talking about, like, how, tight scope data and, you know, having autonomous agents means you need least privilege and short lived permissions and controls. There's a really interesting thing on the data side and data trust piece as well, especially when it comes to AI. We talk about, like, compliance controls. We'll talk about this if you come to the AI trust practitioner. But when it comes to trust, I wanna make sure that only the right data is going into the right AI agents, right system... AI systems. And you can't just rely on saying, well, it's got access to that folder because anyone can move a file from one folder to another, and then the AI agent just inherits that permissions and goes, oh, perfect. I have access to this. So there's a there's a way of thinking at the moment that every single file should have additional metadata added to it to ensure provenance and trust inside of the files. To say, well, this particular agent was created to do pricing information. Therefore, I wanna make sure that the pricing information is updated. It was created by the finance team. It was last verified at this point. It was not modified at any time. There's a whole bunch of series of checks to ensure that I trust the data that is going into these systems is accurate and trustworthy and relevant for today. So we should be embedding every single file with certain, additional metadata types to ensure, provenance, who created it, when did... Was it created, for what system, for what intent, and when was it last verified. Was it verified across different documents to say, yes. This is, you know, been sourced by everything else as opposed to just being picked up. So there's a whole bunch of different things that we can talk about rather than just least privilege. It's now AI data trust calculations to ensure that every single data file that goes into the system is matching that trusted data target, which, again, we can talk about significantly in the AI trust practitioner course, Yeah. Yeah. Garrett, It just... Dallas. And just to add on that tight trust, you know, it's it's really the focus of of the service account, the token, you you know, aligned with the AI agent too. So this is where you heard, Furga talk about that complete scope and each person having, a way in which they would combat. Right? Like, if I were on the identity side, I would be focusing on, you know, type trust motion. Right? Focus on the identity, the service, the token. Having all of this information from the identity to the agent to the data that it has access to, this is where I think it's so critical for for those, not just in the security world, but those that are are focused on, I... You know, you know, I'd say traditional data security from an SIRE perspective to the to the AI security, to the, identity teams. Even the the the the business development teams that are trying to launch these applications responsibly, having a... Having all these people coordinate a unified effort around visibility, audit, and some type of action and control, I think this is where you start those conversations. How amazing would it be to have someone on the identity team talk, you know, a tight, you know, a tight trust scope model to someone on the data side right after one of our sessions or engaging in those conversations? It's just... It's very meaningful. And and where the rubber meets the road is absolutely it's important, and it should be part of that conversation, those three elemental frameworks. There's a really interesting one. I'll be interested to hear your your opinion on this, as more of a tech head than I am, Shane. But, it's a question by, Jindo four, about how would AI be used to proactively identify and prevent recent high profile agentic AI hacks, such as Mythos and Meta? That's really interesting because I I did a couple of videos recently on the hugging face incident. I find it fascinating. So, yes, this... The area of security in which we are operating in is ever evolving, but there are certain technologies out there at the moment that allow us for auditability... Increased auditability of what's happening. One of those is this new area called AI detection response, AIDR, if you want another acronym to, you know, add to your your alphabet soup. But, ultimately, what it's doing is if you're able to plug into the platforms that you're operating in the way where you're creating your agents, connecting to the various models, it should be able to then provide you an end to end observability of every single action that goes from the agent to the model itself when it goes to the core application service, when it goes to any rag index, when it goes to the tools. So we're just observing it. And at the very least, if we can provide a baseline of what it should be doing and then see where it's ended up, that could be useful. Now, of course, most of those environments were in a sandbox environment where it's asked to hack into something, which I hope most of your staff are doing. But, you know, they, they they... At least having some form of technology that is designed to provide you with that observability of every single time that agent does any movement, any orchestration. I wanna see it's thinking. I wanna see what it's doing. I wanna see everything and then, check that against a baseline. So AI, DI was a really interesting technology. It's evolving and it's moving faster all the time, but that's something I think, we're gonna see much more out of in the future. And you can start to then build policies from it. But, yeah, it sounds... I think it should be at the end, but, of course, we don't know how some of these things operate. So if you plug in AIDR and you can observe it, you can go, oh, this is where the risk is. I'm gonna set a policy to prevent it from accessing these tools. I'm gonna prevent it from sending emails externally to my competitors. I'm gonna prevent it from accessing these data types. You can start to, you know, shore up your your AI governance program after you've plugged it in. So that is one approach that it can be done. So I think it's a really interesting space, that... That's evolving. Great. We have another comment here. Jerry also mentions I loved the AI, school certifications, and I think I have seven certs as of today. So that's fantastic. Question that did come in in inbox is, Virgo, can you give us a concrete example of what someone could cover in the AI certifications and how they could apply it back at work? Yeah. Good question. Thank you very much, whoever. May... It's it's selling it for me. So, there's an Australian expression that Marie is very familiar with. It's not... We're not here to put mittens on millipedes. We are doing this to try and help you, elevate your security program. And so, yes, isn't just to educate you about some of the risks, but to give you practical steps to implementing and getting to a target level of maturity for your organization in this ever evolving space. So whether it is AI governance, practitioner, whether it's an AI security practitioner or architect, whether it's AI trust. There's all of these different certifications, all the DSPM certifications. We've got eight at the moment, and there'll be 10 before the end of the year. All of them work on the, pillars or or stages of a journey. These are the things you have to be good at in order to exercise good AI security or AI governance, and then we'll have levels of maturity. And what we will provide in the architect courses is actionable steps, whether it's people, so training up people, assigning ownership, processes, documentation, consistency, and then, of course, technologies of whatever it is to allow you to get there. And what we also provide is a actionable, action table to say, well, if I know what my current level of maturity is and my target level of maturity, what do I need to do at what stage, cohesively, to build up an action plan of near, mid, and long term goals so that I can get there with the resources available to me in the timelines available to me to at least get me progressing and trending forward, with something tangible. And that's hopefully what the security school is trying to do. And then once you realize that, the the technologies that you need, the best in class technologies to do this with the best capabilities, you speak to Shane Coleman, and I'm sure he'll be able to talk to you about all of the wonderful capabilities that Sayre can offer to actually implement that in in the... With our tech... The the technologies. So yes. But a great question, and thank you. It's... We wanna make sure that it's valuable for your security program, not just an education exercise. And, speaking of Shane, a question did come in for you, Shane, from an executive standpoint. What are a couple of reasons an executive should attend DataSec AI, and what should they expect to walk away with? That's great. Well, listen. I I will share with you what I am excited, to to to attend, and I'm sure this will resonate with many executives. So there's two reasons that come to mind. Number one, first, you know, the agenda is built for leaders who have to go answer for AI. Right? Either to the board or to the CFO and to the regulators. You're gonna hear... I know Yotam's gonna be talking about some of the changes when enterprise start to run autonomous agents. I'm really excited to hear that. I'm gonna... I'm really looking forward personally to the Lamont Orange and, John Yo from, Cloud Security Alliance. I think they're gonna be talking about, AI security, AI governance, and where data really fits into that motion. They're holding that conversation tight. I tried to look at the agenda, and they wouldn't let me have it. So I'm looking forward to that one. And, listen, on the executive track, there's sessions like governing AI, at enterprise scale, how to speak to the CFO, like folks, the world of individual budgets are being pulled into an AI budget where an AI committee, has has a program that they wanna introduce. It has to go to the privacy, compliance, legal teams, then it goes to the security team for sign off, then it goes to the CFO. We have to get comfortable in those value level conversations. So this is some of the some of the conversations you're gonna hear us, speak to. You're gonna hear real examples from customers who... Again, this is not a Sayera problem. This is a budget unlock problem. How do we find the biggest problem and quantify the biggest value so that we can speak to the right people to unlock that budget and accelerate our projects? Those are really exciting conversations that I'm looking forward to hear from some of our customers who have who have been successful in that. So, hopefully, that's why you should attend if you're an executive and some of the amazing things you're gonna walk away with. Amazing. And we've got a couple of seconds left just really quickly. If someone isn't a Cyre customer, what would make them... What would make DataSec AI valuable for them? Oh, I'll take it. So. I don't... Because I'm not in Sarah. So I'm not... Like, you you... I've I've worked for AI security storage vendor agnostic. This is an ever evolving space where we need to learn from each other now more than ever. This is imperative that we get together and talk about these issues. Humans are human because this is something that's gonna be increasing... Become increasingly important, increasingly complex, and we can only solve it together. So coming together with your peers to talk about this and figure out how people are solving it is gonna be invaluable. Forget the technology, it's it's about us. It's coming together to to think about how we can solve these problems. Alright. And that wraps our session up. Looks like we are out of time. So Perfect. Thank you very much, everyone. I really do appreciate it. Hopefully, it answered your questions. We'll probably take a collection of everything, and then we can send it off, to whoever's answered them. But, Right. yeah, I'll stop sharing now. Thank you very much. Cheers, guys. Take care, you. buddy.