Building an AI Security Program: Asset Intelligence to Incident Response | Chris Jones, Axonius

View Show Notes and Transcript

Artificial intelligence is fundamentally changing how security operations centers (SOCs) defend their organizations. As threat actors begin leveraging AI to operate at machine speed, traditional, linear security playbooks are no longer sufficient to contain attacks.

However, the root causes of these incidents are over-permissioned service accounts, long-lived tokens, and poor asset visibility remain exactly the same. Security teams must integrate AI into their own incident response workflows to contain threats effectively while leaning heavily on foundational practices like robust software development lifecycles (SDLC) and accurate asset intelligence.

In this episode of the AI Security Podcast, Ashish sits down with Chris Jones, Chief Trust and Security Officer at Axonius, to discuss how leaders must adapt to the new AI paradigm. Drawing from a unique background spanning public safety, the intelligence community, and major tech enterprises, Chris explains why security executives must get hands-on with AI tools to lead effectively.

The conversation covers how to run practical tabletop exercises for rogue AI agents, the changing skill sets required for senior security personnel, and how platforms like Axonius provide the critical visibility needed to defend modern, interconnected enterprises.

Questions asked:
‍00:00 Introduction: Risk Management and Building AI Trust
‍01:50 Chris Jones's Background: From Stuntman to Axonius Chief Trust Officer
‍04:40 Why AI Security Incidents Expose Legacy Fundamentals
‍06:20 Project Glasswing: How Mythos Tests SDLC Maturity
‍08:30 Why Security Leaders Must Go Hands-On with AI Tools
‍10:50 Evolving SOC Workflows to Counter Machine-Speed Threats
‍15:20 The Shifting Role of Senior Security Experts in an AI World
‍18:30 Designing Tabletop Exercises for Rogue AI Agents and Leaked Tokens
‍22:30 Governing Autonomous Agents in Customer-Facing Enclaves
‍27:30 Proving Security Practices and Building Trust with Enterprise Customers
‍31:00 Why CMDB and Asset Intelligence are Foundational for AI Security
‍36:30 Demonstrating Business Value Through Faster Incident Response
‍38:50 Scaling Security Team Capacity Using AI Without Increasing Headcount

[00:00:00] Chris Jones: At machine speed, you're losing ground every single second. Is this even a viable product? Should we even allow this to exist? Should we burn this with fire and act like we don't even know what this is anymore? 

[00:00:10] Ashish Rajan: Everyone is using AI whether they like it or not. Maybe they're not telling you about it, but they're definitely using AI.

[00:00:15] Chris Jones: People on their laptops are spinning up five, 10, and 15 different agents doing all kinds of stuff. They forget about them. Our response time was cut down from probably four days to probably about four hours. When you bring up CMDB with other leaders from organizations large and small, everyone starts to laugh.

Everyone knows that it's not up to date. No one trusts it. The hardest decision isn't gonna be technical, it's maybe deciding whether or not you're gonna have to kill this agent. The only thing that's gonna respond well to the speed of the attacks that are coming is another AI agent. 

[00:00:47] Ashish Rajan: If you're building an AI security program for an AI world, you probably wanna hear this conversation with Chris Jones, who is the chief trust officer at Axonius.

We spoke about how do you build trust, not just with external customers that you may be [00:01:00] working with, but also internal stakeholders. How do you find the balance between plethora of AI tools that your organization may be dealing with, and what role does a CMDB or an asset management tool could play in helping you build a program that could help with AI?

All that and a lot more in this episode. If you know someone who is working on building a security program or probably looking at how do you find the balance of leadership while being technology forward and building security capability, definitely share this episode with them. As always, if you're here for a second or third time and have been enjoying the episodes of the podcast, I really appreciate if you take a quick second to hit the follow or subscribe button, whichever podcast you are listening or watching this on.

We are on all podcast platforms, including YouTube, LinkedIn, Spotify, Apple, and anywhere else you consume your podcast from. Enjoy the conversation. I'll talk to you soon. Peace. Hello, and welcome to another episode of AI Security Podcast. I've got Chris Jones with me. Hey, Chris. Thanks for coming on the show.

[00:01:51] Chris Jones: Hey, thanks. It's great to be here. I'm glad to be talking about this topic today. 

[00:01:54] Ashish Rajan: Maybe to kick things off, if you wanna give a brief introduction about yourself, your professional background. 

[00:01:59] Chris Jones: Sure thing. [00:02:00] So my journey into cybersecurity is anything but normal like a lot of people, but it kind of weaves into a little bit about what we're talking about today.

I started off life of right out of high school going in to be a motion picture stuntman, and worked as a stuntman for a few years on a few low budget shows that were, Okay ... produced down in the Orlando area. So that introduced me to the idea of, in risk management. I got injured doing stunts, so I thought, "You know what?

Need to find another career." So I moved into public safety, became a police officer, thinking that was gonna be safer. You know- ... spoiler alert, it wasn't. So So became a police officer and also was a medic EMT out in the field working a rescue truck. Did that for a number of years. Was injured, had to retire out from that, and found my way into technology to try to stay current with things.

And found my footing into doing computer forensics and digital investigations. And that's my core technology background [00:03:00] comes from that type of work. And disappeared into the government, the US government, and supported the intelligence community and some of the agencies there, where I really refined risk and risk management and what does that really mean, and from a standpoint of high stakes risk, like life safety issues and how we're dealing with people, processes, and technology, and the-- how this all intersects together.

Finally moving back into corporate world, I got to spend some time in companies like Nike, building out their very first cybersecurity capability. Spent time at Dell, Clear, Cisco, and now currently I'm at Axonius, where I'm the chief trust and security officer. And the whole thread there is really a refinement of how I view risk management and leadership across the board- and how to lead through extraordinary circumstances with ordinary people. 'Cause we're all ordinary at the end of the day, I've found. 

[00:03:56] Ashish Rajan: Yeah. Who would've thought? Also, such a great transition [00:04:00] from stuntman to safety officer, and now you're chief trust officer as well in a way, so you've come up full circle.

[00:04:08] Chris Jones: I certainly have. It's been no end of excitement and fun. I'll tell you one thing like I tell a lot of people, I've never had a boring day in my life. ... It's always exciting, it's always interesting. There's always a new challenge to be had. And, and that's what we're dealing with today, with this, the new paradigm of AI encroaching on all of this and causing all these questions where there's, all these supposed experts out there talking about something that's really only impacting the community for the last year or so.

So we're all experts, but none of us are expert all at the same time. So it's an interesting place to be. 

[00:04:41] Ashish Rajan: 100%. And I was gonna say that considering you came from a forensic incident background as well, a first question that I have for you is around the security playbook idea. Okay. And this is probably fa- fundamental for a lot of people.

They know what security playbooks are, but a lot of that has been challenged with the Mythos of [00:05:00] the world and everything other frontier model that's been coming out with, "Hey, we're gonna... We are basically breaking through all barriers." But no one's really paying attention to incidents. Maybe to get your understanding of this, considering you've done this for years, what do you see is still fundamental with incident security playbooks, even if it's AI that's causing the incident as it plays out 

[00:05:22] Chris Jones: Right.

This is a great question. S- Everyone is struggling with this right now, and really at the end of the day, AI changed really who is attacking and how fast they're attacking What it doesn't change is your blast radius and dealing with that piece of it. It's what you have, what it can reach, and who owns it.

If you look at any AI incident write-up the root cause will sound familiar to most people. Like it's a long-lived token, an over-permissioned service account, an asset that no one knew existed, open egress points. The attacker might be new, but it's the same gaps that everyone has been dealing with [00:06:00] for years.

It's the same thing. We cannot, we can't get out of our own way it seems like. And it really gets back to the basics of are you really doing the right things every day for your network? A- and I love that you mentioned the issue with the Mythos problem and Project Glasswing. 

[00:06:16] Ashish Rajan: Yeah. 

[00:06:16] Chris Jones: I had a unique position in being able to see that play out early on, and- Nice

there were some things that are starting to come out from that, that I'm really happy to hear that's permeating across the industry. One of the main things was that development teams that were using good SDLC processes, they had good code reviews, good QA, and the, and these were, mature processes that they actually followed, and they had a culture of good engineering.

When these code bases were put up against Mythos looking for these vulnerabilities, sure, things were found and responses came back to them, but these were not game-changing levels of vulns that came back. These were things that the [00:07:00] team could handle. It may be a little extra, but nothing was put out there that was like caused into question, like what are we doing here?

Teams that weren't following good best practices, that didn't have good SDLC processes, that forewent good QA, and they were just seat of the pants push. Let's get new releases out as fast as we can. At the end of the day, leadership was looking at the output from that wondering, "Is this even a viable product?

Should we even allow this to exist? Should we burn this with fire and act like we don't even know what this is anymore?" Because the, the results that came back were so overwhelming. How do you deal with this? And that's what Mythos has pushed us into is how do we deal with vuln management at this point?

And the teams that have good processes, they're able to deal with it. The teams that aren't working from a good baseline of fundamentals, they're in real trouble at this point. 

[00:07:52] Ashish Rajan: I think it answers my question that I was gonna get into as well as to at where, what points are these arguments about the [00:08:00] fundamentals breaking and to what you said when you put it against a broken process or something that should've been fixed ages ago, but was kept, swept under the rug for lack of a better word.

I'm curious in terms of the are there any AI risks that are probably top of mind in terms of- I know how security leaders have personally done AI agents in terms of different use cases. But what should security leaders think about, or how should they approve... I guess where I'm going with this is that the software development life cycle, which is the, the left-hand side- that's also evolved quite a bit. And a lot of that has also been triggered by developers using their own AI agents. They're building their, bringing their Claude Code, Codex, the list goes on. Those are just the popular ones. They, I'm not even talking about the "shadow AI" ones. What's the thinking that you recommend to security leaders when they're approaching trying to solve that problem of how do I approve the use of AI agents in my organization?

What's the right way to look at that? [00:09:00] This is a place where we're really starting to separate out leaders and the types of leaders that are in place in companies now. At the end of the day, the leaders that are gonna be successful, that can answer that question and answer the idea of what is my team doing with it, these are leaders that are Having to go back hands-on.

[00:09:18] Chris Jones: Maybe they don't have to go deep hands-on with it, but they need to be using it. They need to break it. They need to watch it, have to emergency stop something, and they have to understand who owns it. If they're not in there understanding what it is that they're dealing with as a leader, they're not gonna be able to effectively lead the team.

This AI is not, this is not a new tool that's come out to meet some sliver of a segment of the market. AI is a transformational thing of how we're doing business as a whole across every company out there. It's impacting all the processes. And if the security leadership is not going hands-on with it and understanding it at an [00:10:00] intrinsic level, like they don't have to be the expert in it.

They c- have your team make sure they're getting the time to do it, but they've got to be using it. They've got to understand how it's being used. They've got to understand what can it do, what's it getting access to, what logs is it touching, what can I do to stop it if I have to, and do I really understand who the owners are of these different agents that are out there?

And these are questions to be asking themselves. They have to train like you fight at this point. You can't approve something that you've never rehearsed or looked at yourself. If you're not training along with your team to understand what this is, you're gonna be in trouble, 'cause this, this is a wholesale change of how business is operating today.

This isn't a new tool. This is a new paradigm. 

[00:10:46] Ashish Rajan: I think to your point about the new paradigms, another paradigm that's shifting is for SOC workflow, which probably was the first AI security use cases a lot of people wanted to use AI for in security. Yeah. How do you see [00:11:00] SOC teams a lot of people may be hesitant to think AI can even play a role there.

Some people are maybe overambitious as well. What's the reality there, and what do you say, and how should they approach this? 

[00:11:12] Chris Jones: That's a great question. This is something that is on the front of every leader that's running a SOC right now, and if you're not thinking through this question as to how is AI impacting my team, the makeup of my team, you're gonna be in a really bad place in another year or so.

AI is transforming what is available to the SOC. For me personally, and how I, I deal with this question of how AI is impacting things you have to make sure you truly understand your incident response process. Do you understand your company, the market that you're operating in, the business value that you're bringing as a SOC?

Because if you're not looking at this as a way to move faster, because people are not gonna be able to respond as fast as your agentic systems will be in [00:12:00] this new age of the, the types of attacks that you're responding to. So you've got to be looking at this. For me personally, having done this at a c- at a couple of other organizations, I try to structure upfront for these changes that are coming, and I'm honest with the team about what's changing and when.

I had the opportunity to for my early tier one, tier two SOC people, I brought in contractors for that. So they knew that this was gonna be a short-lived thing while we're trying to build out the new processes and reserve my higher level SMEs, tier three, four, and five type folks to drive the new generation of the SOC that we were building for it.

Because at some point in time, the only-- Like I said earlier, the only thing that's gonna respond well to the speed of the attacks that are coming is another AI agent who can see what's happening, respond quickly. You still have to have some humans in the loop. ... But as a leadership question, you've got to be upfront with your team about what's going on or you're [00:13:00] gonna lose credibility.

You can't promise things are not gonna change because they're gonna know you're lying about it, 'cause things are changing. 

[00:13:07] Ashish Rajan: Yeah. 

[00:13:07] Chris Jones: If, and if you're not upfront with your team, like some of the leaderisms that I use with my team to talk about this are embrace the suck and the company taketh away.

And I try- Yeah ... to be realistic with people that the company has to make decisions to keep going and you're here supporting the company. Hopefully you as a leader have had the time to grow your people to start seeing this coming and being able to elevate people into different roles and put them into places where they're supporting and you can start pushing your lower tier analyst into these more automated positions.

You're never gonna get rid of people in this. You still have to have the human in the loop. But you've got to be upfront about what you're doing. You've got to be upfront with yourself and honest with yourself first that this is where we're going, and own that and move forward with it. It's very difficult.

But we're in a new time period and this... If you're not embracing this, things are gonna pass you by and you will pay [00:14:00] the price later. The adversary will overtake your defenses. There was a great- What- there was a great- Yeah ... I was doom scrolling this morning, as I will do sometimes periodically, and there was a great quote that I saw.

It had nothing to do with cybersecurity, but the quote was, "Aggression and intensity defeat complicated techniques every time." And A- and I was like, "This is so true- Love it ... for what- Yeah ... we're doing." Because, nowadays SOCs are dealing with these guidebooks and these playbooks and guidelines that they have to work against that are very process, linearly driven with time-bound things, and with the new complicated attacks that are coming in, the aggre- and those are with aggression and with intensity.

And our complicated techniques of following this X, Y, Z, one, two, three is not gonna work with that. It's gonna fail every time. And if you're, as a leader, is not dealing with that, you're gonna get overwhelmed, and it'll be a self-correcting [00:15:00] problem in your own career at some point in the future. 

[00:15:02] Ashish Rajan: I'm in agreement with that.

But also another similar conversation is with the change of role for seniors as well. W- to your point- Yeah ... AI agents today, they may not be solving this, a level one problem today, but in the future, in the near future or maybe even very near future, it's gonna happen. But what does the senior people become in this future?

'Cause you obviously mentioned the human-in-the-loop part Is that I don't wanna use that as an example, but essentially there is this transition period where the seniors are probably the, the human in the loop in the beginning and everyone's kinda going, "Okay." But once it understands the human in the lo- loop part is now turned into a trusted workflow, what is a senior person doing in a security?

W- what do you advise there? So at that point, some of the skills are actually starting to change. Your senior leaders, not only do they need to be possessive of the technical skills and acumen of what's going on with the AI agents, but now they're really truly filling [00:16:00] the gap of the, the conversations that happen outside of the flow, especially during an incident with all the different teams that we have to interact with from legal to corporate comms to engineering teams, all these different folks and i- including leadership, the senior leadership.

[00:16:18] Chris Jones: So you're really laying more communication skills, which typically is not a strong suit for a lot of your SMEs that people rely on today, as you've got your, you've got your team members that are break glass in emergency, otherwise you keep them back away from normal people because they- Yeah

they can't interact very well with them, but they're critical to your team because of the experience, the knowledge, the skills that they bring to the table. It's a very interesting problem, and it also gets into the whole how do you even get into cybersecurity anymore if we're losing the defined pathway that a lot of people use to come through the SOC as an early SOC analyst to get into it.

The skills are changing there [00:17:00] too. We're gonna have a higher level of expectation of skills in AI, a higher level of skill expectation in human communication and being able to translate information into these other teams and to have problem-solving and being able to work through issues, not from a standpoint of this is the only way, but to try to find consensus and working together while still providing leadership.

It's an interesting place, and we're in a new frontier here with a lot of this. 

[00:17:29] Ashish Rajan: So maybe to you, what you're saying as well, I guess there would be other skills that seniors would have to evolve, and it's just, it may not just be enough for you to be technically sound about everything. It's about, okay, what's the higher level task that's coming next after this?

[00:17:43] Chris Jones: Yeah, that's exactly what I'm saying. In addition to the technical skills, they've got to bring their human skills. They've got to bring the human interaction. They've got to bring their ability to communicate effectively across different skill sets and even into leadership, which is very difficult because communicating [00:18:00] into leadership is a lot like, moving from someone who likes to communicate in details and technical acumen and show how deep they understand something to actually turn around and now you've got to turn around and basically tweet that concept to someone in senior leadership, and you've got 180 characters to do it.

[00:18:18] Ashish Rajan: Yeah. And 

[00:18:19] Chris Jones: that's difficult. That's really difficult for these smart people to do that, because they want to make sure that they're being understood and the depth of what they're communicating is felt. 

[00:18:29] Ashish Rajan: Do you feel, made me think of to your point about communicating to the board or just in the general executive audience as well The way tabletop exercises have traditionally been done, and there's a-- and going back to the whole security playbook thing here again there's a playbook that people have followed for a while, and these days when it's an incident, let's just say an agent which has a leaked token being used, what are some of the things that people should have figured out before walking into an incident like that?

'Cause I guess [00:19:00] at this point in time, people assume, 'cause going back to what we said in the beginning, that, hey, how different is the playbook and bring that back to a tabletop exercise, which a lot of people may be doing these days, especially end of year coming close by. In the first 30 minutes of a tabletop of for an incident where an AI agent perhaps has a leaked token, what's the thinking there in terms of what are some of the basics that should be covered as part of in responding to an incident which is related to an AI?

Or is it similar, it just ownership is changing? Or what's the difference that you find in running tabletops that are probably AI driven, quote unquote? 

[00:19:34] Chris Jones: First of all, I love tabletops. I think they're one of the most underused, unsung heroes of processes, especially in the SOC. To me, you should be doing tabletops every week as a team.

Back doors and breaches, something. Sh- you should be doing that as a fun exercise to sharpen and challenge your team because, stress testing your playbooks is critical. And this is exactly what you're talking about, like, where are we finding the pitfalls? [00:20:00] 'Cause the hardest decision going through your tabletop, especially dealing with an with an agent, with a I think you said a leaked token.

Yeah. The hardest decision isn't gonna be technical it's maybe deciding whether or not you're gonna have to kill this agent, shut it down, and it's a business process. It's customer facing. What is it doing? And depending on what you know it's doing or what you think it's doing before you know what's actually going on, these are some critical questions.

So like in, in your first time period of working through this tabletop, some of the big questions you've gotta answer are like, which agent is this? Who owns this agent? And what's the blast radius? What is this token giving us access to? And if those three questions can't be answered quickly- You've found your first findings, and these are critical findings that you've got to be on top of, 'cause this, that's the tone up for everything else that's going on. And from that point, you're dealing with things like is someone from the outside [00:21:00] have access to this token now, or has this turned into an insider threat because the agent itself somehow is being manipulated into acting badly? Or is it taking steps on itself to do things that it thinks is the right process, but it's not, but in its world it's doing the right thing?

And these are all, all questions that you've got to work through, you know, understanding the scope of what the token has touched. Check is the agent creating anything outside of its processes that's durable scheduled jobs, auth grants, all, all these different things that are out there. And then like toward the end of it, you've got to have your decision, within 30 minutes, and you really should be doing this even faster at this point hopefully with the right tooling in place.

You got to revoke the token and stop the agent at some point in time. It used to be when I first started, you would hear a lot of people talking about, "We got to watch the attacker, watch where they're going, watch where they're moving, see where they're going." Yeah. "Let's gather some intelligence."

Those days are gone. [00:22:00] You can't- Unfortunately, 

[00:22:02] Ashish Rajan: yes. 

[00:22:03] Chris Jones: At machine speed, at the speed of these AI potential attackers are working at, the watching it and what is it doing, it's expensive, and you're losing ground every single second. So you've got to be able to understand what your agent's doing, what's the blast radius, understand the fallout from it, especially if it's customer facing and you got to put a stop to it and figure out what that looks like.

[00:22:24] Ashish Rajan: I think agent facing customer is an interesting one as well, right? If something does go wrong, who becomes accountable for it? And w- how do you even draw the line for what should be a human approval action? Like we spoke about human approval earlier as well. So is there like a framework that you think of when thinking about agents that would be operating in the customer environment?

[00:22:46] Chris Jones: So it goes back to basics still with agents operating within the customer environment or customer facing support. You've got to know what it is. There's a lot of things that get spun up within maybe your corporate [00:23:00] world. Everybody's got Codex or Claude Code or something. They're spinning up, three to five agents per person in your organization to read their email, check their calendar, deal with Slack, handle all these communication issues.

So you've got this spread there. But that's a different scenario than if you're within your customer facing enclave. Within that enclave, you've got to understand what everything is doing in there, and you've got to understand what is its purpose, what's its scope. What tokens does it have? And you've gotta have a couple of things like a process of how to deal with these things if something is detected.

You gotta have the authority to stop it, number one, immediately without asking for permission, especially if it's starting to act in rogue or suspicious ways. You need to have the authority to go through and change its access, either up or down through a defined process. You've gotta make sure you've got visibility.

This was something that was strange that I've run into some other organizations, where [00:24:00] getting visibility in the logs and actions is-- it can be really difficult to get to because you're starting to get into areas where maybe your SRE teams or some of your other infrastructure teams are like this is our playbook that we're working with.

This is our area that we want, so we're not gonna send you these logs." And it's just like very questionable, like, why are you holding these things back? 

[00:24:22] Ashish Rajan: Yeah. 

[00:24:23] Chris Jones: It really comes into the place where you've gotta have visibility. It all comes down to visibility, processes, and authority to execute on those processes.

But you gotta know what those things are doing, especially if it's customer facing, because if it's revenue impacting, you've gotta be crisp and clear about what you're doing. 

[00:24:41] Ashish Rajan: I think this is so well said also because from a agent conversation perspective, a lot of people would land on, hey, every agent should have a human owner.

What authority is allowed least privilege should be things that people should be care... should try and implement. I... To your point about the [00:25:00] basics or going back to basics when something goes wrong in a customer environment, how are you thinking about this when it comes to internal AI agents running as well?

So to your point, there are two-- you guys have your own AI agents running in customer environments, and by that what I mean is hey, Exton is a product, but then there's also AI capability inside it, and how are you seeing other leaders approach that from a, hey, it's like a, is it third party, fourth party?

I think it's like at that point it's fourth party now. 

[00:25:25] Chris Jones: Yeah. It certainly can be. It's like we hold ourselves really to the same rule that we ask of our customers. Like every agent needs an accountable human for anything that we ship, anything that's going on in our platform itself, that starts with us.

And we, that accountability is really shared, but it's not vague. We're accountable for what the agent's doing. We're not necessarily an agentic platform where we're pushing multiple agents out into customer space. Our agents are, contained within the platform itself and- Yeah ... performing actions there, maybe interacting with [00:26:00] our customers MCP or any one of the 1,500 some odd connectors that they may be using to help manage their enterprise.

But these are things where we're really sharing that with the customer in that we're providing them with the actions that can be taken. We're providing them with accountability and logging and processes. But we're also pushing that out to the customers so the customers can set the dial of risk. And this is really important because every, every customer is different.

Yep. We have some customers in defense, government, some customers in retail, some in education. All of these different areas provide a completely different risk acceptance by those customers, and what works for one doesn't work for another, and they need to be able to drive that. So we try to set a really crisp and clear risk ownership as a partner with them in that we're not shirking what's going [00:27:00] on, but if something happens, we're there with them to work through it.

But the customers really have to be the one that set that risk and understand those actions that may be triggered by information that our platform is uncovering, and they need to understand the downline, ramifications of that, both the good and bad of it. 

[00:27:16] Ashish Rajan: I love that. I also wonder, as a chief trust officer You probably you're maintaining trust on both sides for the customers of Axonius, but also internally as well.

How do you find the balance for... there are thousands of tools these days, right? And privilege access is like one of least privilege is one of those dream situations that we've always wanted. How do you, raise the trust level on AI capability? Or what's your thinking behind this?

And I'm thinking more from other people who are obviously fighting a similar battle where the floodgates are open, for lack of a better word. Everyone is using AI whether they like it or not. Maybe they're not telling you about it, but they're definitely using AI. So- Yeah ... how do you approach trust when it comes to trust with [00:28:00] your customers as well as trust with your internal stakeholders?

[00:28:04] Chris Jones: So that's really a critical value proposition that we have to have with our customers. We're-- in reality, we're a high-value target, and we act like it. We have thousands of customers. We're a core piece of their enterprise strategy, connecting hundreds, sometimes thousands of tools within their environment to give them a single view of asset intelligence across their organization.

And because we're providing such a critical service to them, we know that we are target, and we act like it. Everything we tell our customers about privileged access, we have to prove on ourselves first. And this is a piece where I'm trying to build that across internally with our company too, as, as we're going forward, to make sure that we're practicing what we preach, that we're training like we fight, that a representative of what we're asking our customers to do.

So when we meet with them in our [00:29:00] customer business review meetings that I participate in, that we're talking about what we're doing, and they can hear what we're doing and what's working for us, and we can hear what our customer's doing and what's working for them, and we have that information exchange.

It's a critical piece of the business value that, that we're bringing to them, that we're not just pushing a platform to them. We're pushing a platform that provides them with a tremendous amount of visibility, accountability, and- opportunity not only across what they're seeing with their tool sets and bringing value to the tools that they already have in place that maybe they weren't getting before, but also talking with us and other customers to even further refine how they're doing it.

It's a big responsibility. It's a big piece of what we're trying to build up as we're going forward in our own organization and the next generation of who we're becoming as a company, the growth that we're experiencing, and we're trying to ensure that we keep that [00:30:00] transparency, that we're keeping the integrity that's in place, which is huge with me and being honest and transparent where possible.

Because we're all in the same boat together, and if we're not speaking and talking with that and listening to what they're experiencing and they're not hearing what we're experiencing and understanding where we're both coming from it's a very difficult road to go down. It's a critical, it's a critical piece to make sure you're fostering that trust and it's not something you do one time.

It's something you do every time with every conversation, and you have to be honest, act with integrity. Sometimes you gotta own the mistakes And work with the customers to help them with their mistakes, 

[00:30:36] Ashish Rajan: but 

[00:30:37] Chris Jones: not pointing fingers. We're there to help. 

[00:30:38] Ashish Rajan: So maybe if I were to ask you, someone's building an AI security program from scratch.

They've resisted AI for a long time in their organization. Now the, they're at that point where floodgates have opened up. What would be, say, the top three things that come to mind that people should definitely get their arms around? And what role would a CMDB or an [00:31:00] asset management play in this partic- if, 'cause we're, to your point about the attacks happening in minutes, not in days or hours anymore, a lot of people obviously may not have the latest version of CMDB.

They may not have the latest asset management just as well. So- ... but keeping that aside, assumption that, hey, they have some kind, control on this, I'm curious as to how do you start building an AI security program, and what role would a CMDB or an asset management play in it? 

[00:31:25] Chris Jones: So it's funny, in the meetings I've been to when you bring up CMDB with other leaders from organizations large and small, everyone starts to laugh.

Everyone knows that it's not up to date. No one trusts it. But it's a critical cornerstone of what they're dealing with, especially in an incident. Incident, 

[00:31:43] Ashish Rajan: yes. 

[00:31:43] Chris Jones: It's hysterical to me that everybody is like, "I don't trust it, but it's the only thing I have to work with." I'm like why aren't you updating it?

How come you don't have a workflow process to deal with this?" And they're like we don't really have a good way to do that." So if you're starting your organization new and you're starting to [00:32:00] build, there's, you've gotta put some things, some core components in place to make sure you're keeping that visibility because security and administration, your enterprise, it really shouldn't be separate from each other.

You really shouldn't be acting like y- you're in an adversarial relationship. Start early to build that relationship with those other teams. If you're not building security as a core competency of your engineering team, security as a core competency of your information technology team, of your cloud team, you're gonna be behind the bar.

That, to me, that's the first thing, is building a culture of security to go forward with. From there, you've gotta have your core things of your identity. What are you doing in your cloud service providers? Are you getting the logs? Like I said, sometimes it's difficult to get those from some of the, Yeah

cloud teams. What are you looking at with endpoints? Endpoints is always the really interesting Wild West because that's where people on their laptops are spinning up five, 10, 15 different [00:33:00] agents doing all kinds of stuff. They forget about them. They'll push them out to an EC2 instance somewhere to let them live their own life doing their own fun things, and they lose control of it.

And it's like, if you don't have good visibility across your endpoints and then your network and your egress points, where are your touchpoints that actually get out of your network to the internet? Do you understand that? Are you even thinking about finance? Like th- this-- finance is a critical partner in all of it.

These are the people who own the p- the purse strings of how you're spending your money to do these things and your subscriptions and stuff. It's... These are all critical pieces of it. And I think the final critical piece is your code repos. Do you have the right auth and controls around your code repos?

There's some real interesting things I've seen there where people are like, you know, you've got your Git set up and people are gonna log in to start pushing their code bases up, but it requires them to use their personal email to log in the first time, and then maybe they forget, and all of a sudden they're pushing your source code up to their personal Git [00:34:00] repo instead of pushing it into your company.

And there's just all kinds of weird things out there that people don't think of. And when you're small, you can deal with all these things. You can come in with an Axonius, and you can get a layer of visibility that becomes baked into how you're running your company to get the information you need to respond quickly to things.

You can see what's going on. You actually know what's on your network. Axonius is not a CMDB we don't claim to be, but- ... we'll make your, we'll make your CMDB knowledge accurate because we can take all this information that we've talked about, whether it's AI as an asset class, which is a new thing for us that we're looking at and treating it as such, where all these different agents that are out there become something that we're tracking.

And you're not gonna get that from a single place. I think CrowdStrike just put out their a new feature release the other day where they're talking about their AI kill switch. And that's great CrowdStrike is only seeing a sliver of things in the network [00:35:00] from their viewpoint, and we're in a place to see that same sliver.

CrowdStrike is a great partner and is a great endpoint tool, but we're able to see across everything you have, and we can triangulate across all these different pieces of the puzzle, whi- which is your enterprise, and help you identify things that you don't know is there. Because what you don't know, you can't protect.

What you can't see- Yeah ... you can't control, and that's where your downfall is gonna be. And if you start small when you're small, you can maintain control of that and have a well-engineered, well-oiled machine to run your company with, and security becomes baked in as part of your way to do business and not something that's a frustrating cost center.

[00:35:42] Ashish Rajan: I think, though, I love that also because, to your point, I have noticed something you did not say a lot about new things that are AI related. A lot of this relationship building, soft skills, hey, gets a security capability in an organization. These are not AI things. So the AI [00:36:00] security program still is sitting on the foundation of what we have built with cybersecurity programs in general.

What I appreciated is that you actually had the foundational pieces of building security capability in an organization. None of that is AI specific things, but they are foundational- ... to be able to do AI better in an organization. 

[00:36:19] Chris Jones: Yeah. It's, I think we're moving into a world where security is considered to be a business differentiator.

And- ... CISOs that are gonna be successful in this new environment understand that it's really critical to understand how you're driving business value, how these things work. E- like putting, ev- even things like putting these AI SOC agents in place and have-- be able to respond quickly. One of the things I saw was even having some of these automated security things in place, the interesting thing to me was the number of people that were needed on the incident calls [00:37:00] dropped dramatically.

So you're really saving the company significant money and time by doing this, where before maybe you had 30 people from all kinds of different organizations on this call, maybe we were down to 10 for a pretty sizable, a response for this one organization, and our response time was cut down from probably four days to probably about four hours.

Wow. With only about 10 people needed on the incident call, and that was done through the automation processes that we had put in place. But that also put a lot more criticality on our ability to communicate effectively with this team to make sure that we've got the right people on the call and we're not having to blast out where maybe we need this person, maybe we need that person.

And you could sit back and actually calculate the cost savings to the company that we didn't need all these other folks and allow them to stay focused on their core jobs, let us handle the security incident with as minimal disruption to the company as [00:38:00] possible. If you're not thinking about these things as a CISO or chief security officer moving forward, you're gonna be left behind because CEOs are thinking about this.

I know the CEO of Axonius here with me, Joe Diamond he's thinking about this. He's a very technical CEO. He understands the value proposition of what we're doing and what we're trying to assist our customers with, and I hear this from my peers. If you're not thinking about the business, if you're not thinking about how you're driving your corporate value, how you're protecting the customers and thinking about the customers we're gonna be in a really really bad place in a few years, and you're gonna be left behind as a leader and your company will leave you behind or fall to the wayside.

[00:38:40] Ashish Rajan: Very well said, man. Very well said. I think I love that perspective also on moving forward. We're not saying that the people, are not needed. We're just basically saying that there's a lot more that can be done with the same number of people if you want to as well. It doesn't mean that I have to go for extra budget.

[00:38:56] Chris Jones: And that's a big thing of what I talk with my teams about, is [00:39:00] the AI processes that we're talking about. I have the team that I have, and I think that For the most part, the teams that I've been at have been right-sized. We- Yeah ... the adjustment in team size wasn't needed. But we're looking ahead to future growth, and that growth, I'm not gonna need more people, but I am gonna need tooling because I need to give them capacity to deal with these problems.

But with- Yeah ... that capacity also comes the ability to give them time back to think about their own careers, their own knowledge, their own abilities to do research and have projects on their own to let them rest a little bit, have some mind expansion and some enrichment that goes on. But also- Yeah giving them the right tools so when something does happen, they're able to jump in quickly and deal with it. Even as the company is expanding and becoming larger with more customers, we don't need more people because we've got the right size of people with the right tooling in place to deal with it.

[00:39:55] Ashish Rajan: Awesome, man. Dude that's all the questions I had, but this was a really great conversation [00:40:00] also because I loved how a lot of the conversation that we had that would be helpful for CISOs was grounded in the reality rather than the AI hype. So I appreciate you doing that. But where can people learn more about Axonius and the things that you guys do and connect with you as well?

[00:40:15] Chris Jones: Sure. So the website is axonius.com. A-X-O-N-I-U-S, axonius.com. You can check out the website there.

Sure ... I love connecting with folks. You can also find me on LinkedIn at that Christopher Jones on LinkedIn, and I would love to connect people there. 

[00:40:33] Ashish Rajan: I will put the LinkedIn link on the show note as well as the link to Axonius as well. But thank you so much for your time, Chris. I really appreciate this.

[00:40:40] Chris Jones: Thank you. It was great being here and a great conversation. Love talking about these things, and I hope it's helpful for people and love to continue the conversation sometime. 

[00:40:47] Ashish Rajan: Yeah, it was. Thank you so much. Thanks everyone. Thank you for watching or listening to that episode of AI Security Podcast.

This was brought to you by TechRiot.io. If you wanna hear or watch more episodes of AI Security, check that out on [00:41:00] aisecuritypodcast.com. And in case you are interested in learning more about cloud security, you should check out our sister podcast called Cloud Security Podcast, which is available on cloudsecuritypodcast.tv.

Thank you for tuning in, and I'll see you in the next episode. Peace.

‍

No items found.
More Videos