TRANSCRIPT 🔗
WEBVTT
1
00:00:00.000 --> 00:00:47.000
The protocols that have again existed for a long time, like mtls won't. You can't even connect, you can't even knock on the door without a strong source identity. And company owner ratio that I'm building is focused on not just delivering the source identities within a trust domain, but also making sure that trust itself is a continuous process. Because we all know that like connection time only trust is not that good. Trust changes over time. And then also making sure we have policies to enforce that so that we we don't have to rely on overwhelmed socks to detect things that take a long time to figure out when we can put in place policy engines and we can put in place enforcement points that will just automatically act to disconnect things that aren't behaving properly.
2
00:00:47.000 --> 00:02:10.000
You're listening to PrOTect IT All, where Aaron Crow expands the conversation beyond just OT delving into the interconnected worlds of IT and OT cybersecurity. Get ready for essential strategies and insights. Here's your host, Aaron Crow. Thank you for joining me on another episode of the PrOTect IT All podcast where I get the opportunity and privilege to sit across from a lot of really amazing people that have built things and are building things and creating things and fought the fight and all the things. So today is no different in that space. And I really enjoy the fact that many times as I'm on the screen, I'm sitting in the room with people. I'm usually not the smartest person on the call and I'm okay with that. Right? So I love being around people. Every one of these times I get to learn things and many times it's not something unnecessarily expect. For those of you that have listened for a while, you realize that my podcasts are not scripted. I don't have a list of questions. I'm not trying to get you. I'm not trying to prove I'm smarter. The whole point is to have interesting conversations and talk about real problems that we have in cyber and how we can solve those things. And it's not always about a product. Sometimes it's a product, sometimes it's a product and a policy or process or people and everything in between. So with all that said, Frank, thank you very much for joining me today. Taking time out of your day and why don't you introduce yourself to the audience, tell us who you are and how you got where you're at today.
3
00:02:10.000 --> 00:02:53.000
Absolutely, Aaron. Thank you so much for having me. I'm. I'm honored to actually be here. In the first place, my Name is Frank Goodman. I've spent about 25, I'm just going to say 25 plus. So I don't date myself too hard in technology. Mostly leading engineering teams in Enterprise Journey software. Went through lots of different technologies in time, which has actually been very. I started in networking, I went into medical devices and then back into security. I've done the engineering leadership, I've done the product management. So kind of a full spectrum of the development life cycle and through that kind of come up to realizing that there's some problems that we kind of know how to solve, but nobody's really gotten around to solving them and hoping to, you know, put a company together to get that taken care of.
4
00:02:53.000 --> 00:02:59.000
So let's dive in. Like, what problems are you guys solving? What are you seeing in the space? What is it? Let's go.
5
00:02:59.000 --> 00:05:02.000
Absolutely, let's dive in. So if you're paying any attention to the cybernetic news lately, you're seeing all kinds of attacks coming through, a lot of them supply chain, a lot of them based on this kind of weird dependence that we've grown on static keys and tokens and, and it's really kind of weird to me because we've sort of understood how to prevent this for a long time, but there just aren't platforms that address it. And the main thing that I'm talking about here is that we, we kind of grew up on the Internet and we're like, hey, I need to know exactly who I'm talking to, right? So when I go to my bank, if I don't see that little lock icon at my screen, I'm in trouble. And I need to be very careful. And we've sort of trained the world to understand that. But for some reason that I can't quite put my finger on, nobody has really solved the source identity problem. We have user identity and we have some machine identity, but a lot of this is implemented in static or static siloed systems. So it's not kind of a generally available technology. There's no platform to deliver strong cryptographic source identity. And it's interesting because we've fallen back on static keys and tokens and all of these things that can be stolen and replayed. And with static, with cryptographic source identities, these whole attack vectors just disappear because the protocols that have again existed for a long time, like mtls won't. You can't even connect, you can't even knock on the door without a strong source identity. And company owner ratio that I'm building is focused on not just Delivering the source identities within a trust domain, but also making sure that trust itself is a continuous process because we all know that like connection time only trust is not that good. Trust changes over time. And then also making sure we have policies to enforce that so that we don't have to rely on overwhelmed socks to detect things that take a long time to figure out when we can put in place policy engines and we can put in place enforcement points that will just automatically act to disconnect things that aren't behaving properly.
6
00:05:02.000 --> 00:05:54.000
Yeah, I mean, dude, if you think about, and I'm sure you have, which is why you're here while you're doing what you're doing, but if you think about the full stack, you know, all the layers of, you know, from physical all the way up to application layer, and every bit of authentications and credentials and things that we have along that path and how many of those
Load full transcript
things can break down and then you start getting into complex systems and we've already seen it all, you know, with AI with Open Claw and people plugging in API keys and giving it access to their email and oh my gosh, it's insane. And they don't even realize what they're doing. Like, so I'm using openclaw, but I have it running on a dedicated system and I'm, I have dedicated APIs that only allow certain access and I'm rolling those keys constantly. And I'm a cyber professional, so I do this for a living. Most people that are using this are not doing that, they're putting it on their lap.
7
00:05:54.000 --> 00:07:06.000
Some professionals have also gotten in trouble with this. Right. Like it's nuts. And it's an interesting thing. AI is the latest incarnation of things that can go wrong. There's no question about that. And I, and, but if you think about what an AI is from a traditionalist standpoint, it's a process and sub agent is a sub process. These are not new things. From a technology standpoint, what they can do is new, what bad acts they can perform is, is certainly new. But the way that they're operating in our systems today is very traditional. Right. The question, there's two kinds of questions around AI that have to be answered. One is should they be allowed to connect to things? That's the problem that owner ratio solves. And then if they're misbehaving, how do you stop them? Right. So owner ratio doesn't deal with detection, but it does enable detectors to signal changes in trust in a pluggable way. So that as these New governance and AI things come about, we'll just take a signal in and disconnect that thing the second it reads the wrong file. Right. And we deliver these identities not just to the work, to the endpoint or the workstation or the workload, but to every process and sub process that runs on that box. So nothing can even connect unless it's passed a series of tests. Right. And those tests are definable.
8
00:07:06.000 --> 00:07:51.000
Yeah, and that's the interesting part. And we know this. It's no different than many other technologies. You know, the way I describe AI in my layman terms is it's like the intern, like, what access are you going to give it and are you going to have? You know, because it's just a person, like, yes, it's a computer, but you're giving it access and what are you going to trust it and what access does it have and what level, what authority you going to give it to actually make changes? And then where are you putting in the approval process? It's no different than I would do if I hired a new person. I'm not going to give them full domain admin rights when they just got out of college and they've never touched a system. I may let them tinker, but I'm not. I'm going to make them have, hey, come ask me and then we'll do this together and walk them along the path. It's the same way I design an AI system or if I'm using AI in my system. Right?
9
00:07:51.000 --> 00:07:59.000
Yeah. I mean, but to complete the analogy, most people are giving the AI their own credentials. They get to act like you and they don't have your experience.
10
00:07:59.000 --> 00:08:00.000
Right.
11
00:07:59.000 --> 00:08:07.000
And so you're like, oh God, I let my intern run wild with my API keys and it can connect to anything I can connect to. And it's crazy.
12
00:08:07.000 --> 00:08:52.000
So many times I see the problems and whether it's an OT where, you know, we're connecting OT systems directly, the Internet, or a lot of the bad actions that I've seen in my experience, most of those are not because the people are A, stupid or B, you know, malicious or C, don't care. Many times it's because they have to do something and they're trying to get the job done and this is the only way they know how, how to do it, or the best way they know how to do it. They don't know the right way to do it. Nobody showed them or it's too difficult or expensive or whatever those things are. So how do people get around that? Obviously AI is everywhere. The cost of entry, the barrier of entry is really low. But there needs to be that rigor, at least a little bit of rigor around before I give it my, you know, before you give your kid the keys to the Lamborghini, you should at least teach them how to drive a stick.
13
00:08:52.000 --> 00:10:36.000
Well, for the large enterprise, I think it's really important to remember that even the best companies aren't always sure what they're running, right? I mean, open source is amazing, you can do all kinds of great things with it, but it's turtles all the way down. Like you don't know always exactly what's in each one of your systems. And the more sophisticated security organizations are doing things like SBOM analysis and signing, cryptographic signing by their developers to make sure the image you deploy is actually the image that they built. And that's to try to get at some of these supply chain problems that we're having. And that's where using those technologies to help in allowing whether or not that process or that workload should get access makes a huge difference, right? Because it's not just like I'm a container in this cluster and therefore if I'm in this cluster, therefore I'm safe, right? No, are you the right container? Are you the right image? Are you the right, you know, hash? And all those things have to be taken into account to get access. And that's where this pluggable attestation architecture is so important, is that you need to have build your own confidence. And for every enterprise that's going to be different things. They have different tooling, they have different security standards, they have different development processes. They need to be able to adapt that to their environment. Some of them are going to be in the cloud, some of them are going to be in multiple clouds, some of them are going to be using lambdas. Like all of that stuff is going to have to be factored in. And that's how you restrict this because you
don't really know what's in there all the way down. But if you are one of those more sophisticated companies, companies that like latch onto that technology and you're looking at the pieces, you can use that same evidence to ensure that nothing gets into your system at runtime that hasn't passed those gates.
14
00:10:36.000 --> 00:11:32.000
And that, that is huge because I know OT is behind the ball. But IT enterprise, like there's so many IT architectures and enterprises, when you start thinking cloud and all of the technologies and the systems and the connections, think about how complex that is. And most of the time you have silos. I've got my, you know, this team is responsible for these technolog and applications. And these are over here, These are over here. Different people are setting up different things. Yes, there's usually an overarching cyber or compliance or some kind of overarching architecture group that's supposedly looking at those things, but nobody can be an expert to the nth degree in all of those settings. Log 4J obviously was an obvious reason for that. We didn't know what was included in those things. Right? But if you think about that, like how can I figure that out without understanding that and all of the things that I have across my entire organization and how A +B +C is going to impact my risk and connections, right?
15
00:11:32.000 --> 00:14:15.000
So that is a huge thing. And I want to give a small shout out like there's really good tech out there right now, right? There's great EDR technology for endpoints, there's great identity technology out there for user identity and two factor authentication. There's fantastic OT companies now that are fingerprinting devices and understand exactly what's connecting. When all you see is a Mac address coming into your network right at layer two, right? There's really good tech there, right? But what there aren't is a system that takes that all in and helps you formulate a policy on enforcement and then helps you enforce that. Because again, like if you have a bunch of OT devices, let's say a bunch of MRI machines and let's say you have five of them and they're all from the same manufacturer and your OT team is monitoring that because you've got a great tool for that and it suddenly sees one that starts talking to something isn't part of the normal pattern, right? There's going to be a little signal sent back and if you send that into owner ratio, we're like, hey, wait, Mac address is now suspect, right? Its trust score has dropped. And if your policy says that its trust score has dropped to a certain level, we signal back and say, hey, take that thing offline, right? And that's for ot. But we can do that with everything in your network at the process level, right? Y because the tooling that exists today is fantastic. But none of these companies, as good as they are, are in the business of solving the broader problem. They're in the business and correctly so because they have shareholders that like pay the bills building their business. And unfortunately sometimes that leaves you with silo. And if you look at what most high end SOCs do is they're aggregating all of this signaling from all of the tooling that they have into a big sim, you know, with a big set of dashboards and a bunch of custom homemade stuff. And the kind of the problem is they were working at human speed. I had this conversation earlier today with a buddy of mine. They come in and they're like, hey, there's this anomaly. And then they like, oh, that one's not important, this one's not important. But they're processing all this at human speed. What owner ratio does it takes in all that same input? It applies your predefined policies to how important all of those events are and it acts. So the protection is in place 30 minutes, 45 minutes, hours days ahead of when the SOC can reasonably act, given the volume of what they're seeing and when they're acting, they're successfully sitting there in post mortem mode because the thing's been protected milliseconds after the event crossed the threshold. So it's really important that these things be automated because the dwell times and the bad actors are what kill you. If you can nail these things in milliseconds, you've got plenty of time to figure out what happened, right? And we have obviously a log of absolutely everything. Like there was an event and it triggered this and the trust score changed and then we disconnected the device. And we can provide all that to the SOC to do that post mortem. And it makes a huge difference. It makes a huge, huge difference.
16
00:14:15.000 --> 00:14:44.000
Well, and to really simplify it down, like when I was, when I build teams and we're troubleshooting something, it doesn't matter what it is, all the way back from my desktop administration days to networking, system administration, whatever it is, all the way down to OT systems, right? I learned it a long time ago. And when I would train my teams, if we're troubleshooting something, I always would say start at the lowest layer of the OSI model first. Like I don't start troubleshooting the application if I haven't confirmed, is the network cable plugged in, is the power turned on, is heavy, have you turned it on off again, that type of thing.
17
00:14:44.000 --> 00:14:45.000
Right?
18
00:14:44.000 --> 00:14:59.000
But if you think about it like that infrastructure layer is the most important, right? That not most important. It is so incredibly important. And it crosses boundaries between technologies, between vertical, between business units, like that is the clue.
19
00:14:59.000 --> 00:15:13.000
And between companies 100%. I mean, like how does UPS know that that order that came in from Amazon, which was from a non human identity, right? It's from some microservice running in the background is a legit order. If it's an API key, if that's all it knows.
20
00:15:13.000 --> 00:15:14.000
Right.
21
00:15:13.000 -->
00:15:15.000
That's not great. That's not a great answer. Right?
22
00:15:15.000 --> 00:15:51.000
No. Yeah. And how do you, you know, if you take a big step back and you think about that as me, as a human, as a dumb human, that's a, that's a really big problem to be able to understand it. And many of the times when I go into these businesses, that's one of the problems they have is they don't really understand the end to end business process, to really be able to grasp how big that problem is. And they're trying to compartmentally solve it with I'm going to bolt this tool on or I'm going to bolt this tool on and I'm going to do this thing. And it doesn't really inherently solve the root, you know, problem at the bottom. They're whack a mole symptoms and problems that they have, they're not really fixing the true issue.
23
00:15:51.000 --> 00:16:37.000
Yeah, I've been talking about this more and more and I think the metaphor I like to use is trust is kind of like a missing layer. Right. We have this OSI model and we, the data plane and we have the control plane, but we don't have anything that's continuously monitoring whether anything and everything in the system should be trusted right now, not when it connected three hours ago, but right now. Right. Because everything that's happening on that endpoint or on that workload or on that, you know, lambda is changing all the time because other things are coming in it's behavior, it's growing memory, it's got a leak, whatever it is, it's always changing. And as those things, changes are detected by these great tooling that already exists in the world. Someone's got to act, someone's got to be bringing that together and automatically turning things off that are broken and doing that at a session level, not at a big pipe level.
24
00:16:37.000 --> 00:16:52.000
So how do you see this changing architecture, secure architecture in the future? Sdlc, product development, you know, switch design architecture? Like where do you see this being the biggest impact in making a difference in the lowering risk and changing. Changing how we approach an attack?
25
00:16:52.000 --> 00:19:39.000
If you just take the simple example of no API keys and the fact that you have to prove who you are to even, I want to be clear, to even connect. It's not like, like today if you get a token or you get a session key like you can do whatever you want with the permissions that, that Thing has, and we've seen plenty of examples with this Latest like Team TCP1 that just went through three levels of systems tearing things apart and exploding from there. If you can't like those things will not exist. Those, that whole attack vector just stops existing, right? Because it's not like you have access. You literally cannot knock on the door without strong source identity in this model. And that doesn't require development, that doesn't require changes in architecture. MTOS has been around for a long time. Like we have the technology to do this right now we just need this layer that can provide the source identity for the organization, enforce it and drive the policies that allow you to disconnect things when they misbehave and the attack vectors to just collapse, they just disappear. Now, not there's plenty of like phishing and things like that, but the same thing. If a system detects that the score changes, right? And when the score changes, if it reaches the threshold of disconnect, then that that system is gone in seconds, right? So as these events usually cascade, one bad thing happens, time goes on. Two things happen, time goes on. And that's great because trust is a threshold, right? And once you've crossed that threshold, something has to change and it has to change quickly. Quickly, right. Because at the end of the day, we're protecting the workloads with the payroll systems, the, you know, the sales customer data, you know, all of the proprietary information, the source code, all of the things that these large enterprises are trying to protect. That has to be the priority. They have to be locked up first and fast. Like they don't have time. And I hate to say this, for us humans to figure out what's going on and AI doesn't solve that either because AI still contemplates, you know, what it's doing. So an AI can assist in maybe setting up what those policies are, right? But it's not going to be in the data path, if you will, it's not going to be able to act fast enough for the kind of security that we need in these large enterprises. If you're talking, you know, banking or medical hip, you know, healthcare, oil and gas, critical infrastructure, federal government, all of these industries, they don't have time to mess around, right? They are doing their best. You know, they've, you know, like federal government is great. They've been talking about zero trust, they've been talking about policy engines for I don't know, what, five, 10 years now. And it's good tech. But the two pieces that haven't been done yet are policy engines and source identity. And without those things with this continuous trust model, they're not succeeding. You can find zero trust slapped on absolutely every security product in existence today. But it's words today. It's a very small sliver of the requirements that those standards set forth.
26
00:19:39.000 --> 00:20:36.000
Yeah, and to your point, like we have this problem no matter the vertical, no matter it, ot, enterprise, cloud, local, on prem, you know, we look at OT and a lot of times we think about, and I keep targeting OT because I think about it because it's, we're 20 years behind in theory and some people will claim it's at least 20 years behind sometimes of where we are in it and. But we're having the same problems, right? We have, you know, authentication issues. We're, we're introducing APIs and service accounts and, and all the things that we have. We have that same problem in the IT space. It's just faster. Like we don't have. Not as many
people are using AI in the OT spaces yet. So we have a little time there. But that doesn't mean that we aren't having the other problems that we still have in the IT space. And it's going to be faster, harder to do those things. And to your point, can't put it, I don't care if I have the best SOC in the world with the best tools in the world, an analyst is not going to be able to respond fast enough to stop a bad thing from happening. Because they're human.
27
00:20:36.000 --> 00:21:41.000
Exactly, exactly. Or I mean even with computation that isn't designed to scale into the data path. Right. Like we're talking session level interruption, right. Packet by packet verification of the trust score prior to letting things continue on its way. Way. Right. So that's not something that is going to be handled by just any old software, right. It has to really be designed for that kind of performance and scale so that it can really act in the milliseconds that it needs to in order to deter these things. And OT by the way, is a great problem space because you know, they have other issues, right? Like the pedigree of their source code and you know, all of the concerns and you know, where they come from and what they do and you know, and very little identifying information. And that's where again, best in breed products that already exist in the security space are huge levers. But those things need to connect to the rest of the Internet and the rest of the things and so they need the other side of the tooling as well and they all need to come together to deliver the inputs or the trust capabilities so that if the OT sees something, the back ends can be protected. Right?
28
00:21:41.000 --> 00:21:42.000
Yep.
29
00:21:41.000 --> 00:21:48.000
That's ultimately what you're trying, you're not trying to protect your OT device, you're protecting it from going out and going nutty on everything else.
30
00:21:48.000 --> 00:22:18.000
So what, walk us through, like what does this look look like? If you're walking into an organization that doesn't have this visibility, how do you, how do you set this up? And this has always been my struggle with Zero Trust because most people, especially the way most Zero Trust products are there. You have to have a complete map of everything that talks to everything and how they talk, which protocols and when they talk. And you have to know what good looks like to be able to lock it down. And most people don't even have a clue what their environment actually does and how it works. They've bolted things on and it, it works. They roll the dice. I'm like, yeah, it seems to work, let's go.
31
00:22:18.000 --> 00:26:28.000
Yeah, well I think that you can scale this out very easily in that you, you know, you have to put in place the centralized controller, the setup, the policies, like, you know, what does a various input mean, how does it impact the trust score? And that's obviously what the software helps you do, is work through all that configuration. But ultimately you're, you know, if you think about it, I went through this example earlier today. Trust as a technology doesn't really exist as a real time thing right now. So I can't, I can't go in and say like hey, retrofit this software so that it looks at the trust score and acts accordingly because nobody's going to do that A and if they did, it would take them six to nine months to release the next version. Right. So we can do that by putting in place gateways that see every session that comes through. Right. And we can put that in an, in a reverse proxy. We can put that in the kubernetes space. We can put that on a VM on the back end. Those gateways act as the enforcement points. Right. For your system. And they require source identity to even allow anything to connect. They're protocol independent so they don't work on, you know, like it only speaks this protocol. So like SSH or RDP or HTTPs, whatever it is, we're agnostic to that. We see the connection, we verify the source credentials. If it fits within the trust domain of the environment that we're protecting, we allow it. If a trust score changes while that Session's in place, we are notified asynchronously and we kill the connection. So the first thing is to put the backend in place and then you work forward to how do you give source identity? That depends on the maturity of the organization to some extent. How well do they understand their workloads? How well. So like on the back end side where you're dealing with, you know, kubernetes, clusters and containers and things like that, they've got CICD pipelines, they've got all, all this technology to build, what it is they're deploying and they have a pretty good handle. So putting in place attestation for those environments is pretty straightforward. But it's also really cool because they can say well here's the stack of things that a workload has to pass in order for us to allow it into the system to get that source id. It's not just like did we build it? But like what's the SBoM? Does it have any CVEs, does it have this right hash on the image size? Like you can completely build this up based on your environment and then you've got your whole backend working off these source IDs. The next step is to go out to your endpoints which are, you know, a little bit harder because they're, you know, controlled environments. Usually Microsoft Windows or Mac OS are the sort of predominant things. Obviously you can add Android and iOS if you're, if you're doing mobile stuff and putting agents out there that, that can see what's happening is, is a big piece of it. The other thing that I think is important to understand is that many companies, whether it's OT or large enterprises, they have a certain amount of stuff that's just that they build and they deploy and they manage on the back end, but they also have a tremendous amount of IP and what they would consider proprietary information floating out there in the cloud in other
providers. Providers, right. So when you talking about Salesforce Data or Slack or any of these SaaS based companies, which are again great companies but almost exclusively work based off user ID and those companies don't really have any sense of what's on the laptops that they're connecting to because they're independent systems, right. And so when we start to move out to the endpoint agents, we will see all of those connections being deployed into these SaaS providers and when the trust score changes on that endpoint we can signal the sasses to disconnect the user. So completely out of band monitoring that for the first time really brings together what your internal security team knows with what your service provider SaaS companies know because right now they're completely separate. Like those two don't talk very often. Right. And so like even you know, like how much information do companies have in Salesforce? They have all their customer information. A lot of them do ordering in there. A lot of, you know, one token or one sashen key and you're off and running, just sucking that data down. And we'll know that that connection exists whether you know, whether you're an enterprise, if you're an enterprise customer. And we'll also know if your trust score changes and we'll be able to adapt accordingly and disconnect those connections. So deploying it's pretty straightforward.
32
00:26:28.000 --> 00:26:51.000
Yeah. So the other side of the coin and I can hear and the devil's advocate folks, but what happens when it breaks and I don't know what's going on and how do I troubleshoot it and how do I know where it broke in the stack and what caused it to break and how do I push that thing through? It was a really important process that to happen and I don't know how to, you know, how do we make sure that all the teams know where to look when they're troubleshooting? Because now I added another complexity along the way.
33
00:26:51.000 --> 00:27:53.000
Yeah. So the system itself is completely self logging in that we have complete audit records of every single event that occurs in the system. We know exactly what tests you pass for attestation. If you put your S BOM in there as part of that process, we have a capture of the source BOM for the images that are running. If an event occurs that changes your trust score, we have that in our logs. We cryptographically sign that so they're immutable. If we do an enforcement, we know. So the exact history of everything that we've done as a system is constantly available to anybody who wants, you know, with permissions to look at. The system itself will, when it's deployed will be either a SaaS itself for companies that don't want to have to manage the infrastructure, but it will also be self hostable either, you know, by the preferences of a lot of large security organizations. Financial services aren't going to put this kind of technology in a public cloud probably they'll probably put in a data center. Perfectly okay with us. And for the record, we eat our own dog food. So everything in our system doesn't talk without source identity and destination identity and trust scoring continuously. And anything that can happen in our system. Happens to us as well through the same process.
34
00:27:53.000 --> 00:28:44.000
Yep. That's the key to this. Right. Is we have to have that visibility and understanding. And I see imagine that I have an understanding of all the applications and everything they're talking and not just the application but the process end to end which we don't many times have the visibility. Yeah. To your point, the application team understands their stack and they understand their kubernetes environment and they understand that but they don't ne like there's a transition to the networking team that's then a transition to the client facing team. There's really very few people that really understand end to end how that process works and where it breaks and many times there's gray areas every time there's a transition that. Well, they don't really. It's the Spider man thing where they don't really understand who owns it, who supports it, how do I troubleshoot it. Like why is it broken? No, you did it. It's always rather routing, it's always a network team, it's always a firewall issue. Right.
35
00:28:44.000 --> 00:28:45.000
It's always DNS.
36
00:28:45.000 --> 00:29:02.000
Exactly. So this would give the a lot more visibility into what's going on even beyond just the blocking and tackling of the bad things that are happening. I can see it also being useful in a understanding of what's going on at that layer end to end because you're seeing that full thing as it
37
00:29:02.000 --> 00:29:21.000
goes out and comes back to the session. We will probably over time do more and more with metrics around around the session information. Not our primary objective is to secure the systems. Not to necessarily measure every step, but you will natively see every step in the connection from source to destination through our system because it will pass through various checkpoints along.
38
00:29:21.000 --> 00:29:44.000
Well in theory you have to. Right. To be able to score it. You have to really understand that full process. So that's the value add to really, really know and that you know, those are the things that you start thinking about if I can dynamically. Hey Aaron. Usually Aaron's good and Aaron's laptop has all the right things and it's got the CVS and the patches and he connected to a secure environment and he's on VPN and all those types of things. But Aaron just popped up in China.
39
00:29:44.000 --> 00:29:45.000
Yeah.
40
00:29:44.000 --> 00:29:46.000
So that changes his risk score.
41
00:29:46.000 --> 00:31:48.000
Yeah, absolutely. Yeah. And being able to do that like a lot of companies worry about things. That is a great example about where things are in the world geographically is a huge issue. You can have policies that adapt to that.
Right. So I mentioned that like there is no trust layer right now. Right. But once this trust layer is in place, because we're doing it with gateways to make sure that it's deployable on day one, we're also delivering SDKs that can allow for grappling. Graduated. Right, access. So if you want to modify your apps over time to be more granular in how you deal with trust scores, that's a reasonable thing. And more sophisticated organizations will probably take that path so that you could actually make your policies adapt to that very scenario that says, hey, when, when Aaron's in China, read only access, baby. There's no writing to the database. Right. Or however you want to enforce it for what you're doing. Right. So you can adapt those policies in real time. I just wanted to drop a thing because it's a weird side effect of this. But if you're doing things for like mtls from the endpoints all the way through the back end, some of the current things this just fall away that we're doing right now. Like do you still need a VPN if everything is cryptographically secured all the way? That's just a big expense for enterprises is how do they pay for VM. Yeah. If you're using source BOMs for, you know, attestation of workloads and the new CVE comes in and you as the, you know, SEC DevOps or whatever the title is in your organization, have to figure out out the blast radius of that cve. How do you do that? You've got CICD systems, you've got Bill systems, you've got this, you've got. We can tell you in seconds what's running in your system right now because we took the source BOM in before it even got access. And these are just things that fall out of the architecture that bring additional value to enterprises that are real problems. When these new, you know, you mentioned the log 4J, but when that thing came out, I mean the company I was at at the time, we spent weeks trying to figure out if it was in impact, what was impacted, what we'd sent to people, et cetera. I can tell you, you know, this system will tell you exactly what's running, which is different than what's in your development and your repositories, but what's running in your system right now and whether or not it has that issue.
42
00:31:48.000 --> 00:33:32.000
Yeah. And the thing with that, right, is as, as we know, it's so dynamic in what we're doing and everything from, you know, we're doing the best we can. Like you mentioned the VPN stuff, right? We use the technology and capabilities that we have because they're the best solutions that we have have. And we're, we sometimes struggle to change because change is scary. I've got my process around these things. So changing the way I do something is bigger than installing technology. It's installing technology and changing processes and procedures and training my people. And now I've got to train my support staff on troubleshooting. Like all of those things, they're not problems, but at the same time they are, right? It's that learning curve of anytime anything new that comes out, I mean, you see it with AI right now, everybody's scrambling, how do I use this? How do I do it it safely? Can I do some organizations just block everything AI others are building environments safely in their space so their data is not exfiltrating out and people aren't trying to use chat GPT there. Anytime you have something like this that is fundamentally and architecturally changing the way that we do things, even if it's not like having to add on to my stuff, it's still changing things as I am. To your point, right? As I implement a technology capability like, like this, it's going to break things because I don't know how they work. And then we're going to have to troubleshoot and we have to understand that. So how do I update my documentation and all of those types of things? These are just things that people. Which makes people hesitant. And I understand that. Which is why some people are very slow on the adoption of new technologies, capabilities, et cetera. And there's others that are more on the bleeding edge. They're comfortable making those things and being that bleeding edge, at least in a certain space. So where are you seeing that kind of resistance as you talk about things with a new capability and kind of focus on how people are approaching it?
43
00:33:32.000 --> 00:36:03.000
I think it's important to sort of prove your mettle at the right level. Right. So my goal for owner ratio is to really be starting in the large enterprise space where security is clearly a prime directive, right. Because they have both the scale and the dynamic capability to really put this product through it, through its tests. Right. And with that I think you get a sort of, you earn your credibility, right, by saying, like, look, we solve this problem at scale, right? We're definitely going to be able to solve this problem at a lower level because there's, you know, how many companies even have a sock? Like some don't. Majority don't Right. They have a. Maybe they have a CIO and three guys that run the network. But, you know, the enterprises have teams of people who. This is their job, right? And this should make their job a lot easier because it switches them to, you know, something bad is happening. He's all falling on me while I figure it out, out to something happened. The assets are protected. Now I can calmly go through my job and figure out what happened. It shifts the burden a little bit because of the way it instantaneously operates. So you take the knowledge that those practitioners have, you work with them to put it into the policy engine so that it. It acts correctly for their environment. And then it takes a lot of the heat off because almost everything they're doing is with the confidence that their backends are already secure and not under the, you know, watchful eye of the executive committee while they're, you know, trying figure out what's going on and having sleepless nights and blah, blah, blah,
because they're on the call all the time. So I think, though, I expect there to be a pretty warm welcome for this kind of technology in the very. In the places where this is the real challenge, which is at scale in the large enterprises where security matters. And it really is a priority. And with that credibility, I think people will see very quickly that this solves a tremendous amount of problems. This is not world hunger. This is, you know, there's. This is not like a panacea that solves everything, but getting rid of the spurious access is a huge, huge issue. And then that ability to dynamically handle the trust just changes the game tremendously for the protection. And it's weird because the AI is exploding right now. We just saw this mythos anthropic announcement with all the new stuff being found, like, guess what? Most of those attacks require you to get into the network. And if you can't connect a lot of that stuff, you're a lot safer without being able to connect. Right? Like, so it. It's like, again, this isn't something we haven't thought about as a community for many, many, many years. Years. No one has stepped forward and said, like, here's how you solve the problem. And I, you know, we can build it, and it's really quite straightforward.
44
00:36:03.000 --> 00:36:10.000
So what's the most exciting thing for you to see with this new kind of approach at targeting a different problem, or at least a problem a different way?
45
00:36:10.000 --> 00:37:24.000
I mean, for me, it's just the, you know, I've been sitting in the security space for a while Now I was, I've seen great companies build great products and it's just been, I don't know, disappointing a little bit because again, they just created, create their own little walled garden of tech. And I get it, but there really has to be something that brings them all together, right? Whether it's the crowdstrikes and the Microsoft defenders. Like, they're not. They're competing like, but. But they also are both great products and they do things well and they need to be leveraged. Companies have massive investments in these tools, but there's a huge inefficiency in deploying them, which is why they all funnel all that data into dashboards and socks is because they're trying to bring it all together together. This platform brings it all together in one place and it acts on that data instead of just putting up red, green and yellow bars for people to click through and this and that. I'm super excited to see that in action because that, like I said, dwell time is everything in this game, right? When you know something is wrong, you need to lock it down now. And when I say lock it down, I mean at the session level, not at the big pipe level. Right. The second you see something go wrong, it loses access this and then can't do anything. Nothing happens at that point.
46
00:37:24.000 --> 00:37:33.000
Nothing spreads. So one thing gets in, then it's isolated to that space and it's not spreading across my environment. Like, you know, so many attacks we see that happen.
47
00:37:33.000 --> 00:38:01.000
And I, I think I said it, but I just want to make sure everyone understands this. This company that I'm building is not in the detection business. Right. I don't want to compete with anybody, nor could I hope to. CrowdStrike will forever be better at me at endpoint detection. And they're great at it. You know, we want to leverage that existing technology in a unified platform form to be able to continuously monitor, trust and deliver the enforcement aspect as well as the strong attestation to make sure this thing works end to end.
48
00:38:01.000 --> 00:38:40.000
And the benefit of that, right, is many times you're like, oh, well, I have to do this, but I've already spent all this money on these other tools. And now there's the Venn diagram of overlap of, you know, I've got which one of these tools do I use this capability? Which one is better at what? And, and I'm going to use this one over there and this one over there. And I don't have full complete coverage. And I've deployed it in 80% of my environments. Like this is a Problem as old as time. But when you have something that sits in the middle and can really interact with, receive data from and alerts on and take actions upon, that's where you start really getting a multiplier. You're making the tools that you already have more valuable. It wants you to more invest into Cloud Strike and the tools that you already have.
49
00:38:40.000 --> 00:38:41.000
Absolutely.
50
00:38:40.000 --> 00:38:44.000
Because the more things they see, the more things that you can do and take actions on.
51
00:38:44.000 --> 00:39:09.000
And you know, to your background, the OT tools are fantastic. Right. I mean, they have incredible capabilities and guess what? They have signals and they see things that other tooling don't see and they're fingerprinting and doing all that cool stuff and you want to integrate that with your EDRs and your, your network detection and your identity and all of this great stuff. I mean, from my perspective, the answer is P. All of the above. Like I want to use them all.
52
00:39:09.000 --> 00:39:10.000
Yes.
53
00:39:09.000 --> 00:39:10.000
Right. Because they're all great.
54
00:39:10.000 --> 00:39:11.000
It's not either or.
55
00:39:11.000 --> 00:39:20.000
It's yes, all and overlap doesn't scare me at all. You know, I don't want customers buy overlapping products, but if they do, I'll, I'll use both and I'll get the, I'll get the best of both worlds.
56
00:39:20.000 --> 00:39:35.000
Yeah. That's awesome. So what's the. I ask everyone this question. What's the, you know, one thing, and I probably already know the answer to this, but I'm going to ask it anyways. What's the one thing that kind of keeps you up at night that, that you're concerned about coming up over the horizon and maybe one thing that you're excited about?
57
00:39:35.000 --> 00:41:24.000
What am I concerned about
on the, from a company standpoint, look, we're, we're just getting started. We're seed stage. We're, we're getting ready to start raising money. That's all always daunting. We're definitely, you know, I've talked to you about this earlier. We're looking for design partners, companies that want to get in early work with us. I'm not asking for money from design partners, just attention, roadmap advice, making sure that we build this to fit the large enterprise environment, but definitely looking for people to reach out on that because I think that they can have a huge influence on, you know, if it's OT focused or if it's large enterprise or the SaaS aspect. They're all great, you know, standalone pieces and the design partners are going to help shape that. And obviously we'll cut them deals when they become customers later. So I do you Know, as, as a solo founder, I definitely worry on that front just in terms of how to get this thing going faster, but, man, am I excited. Holy cow. I have not had a conversation yet where anyone's like, I don't really get what you're doing. Like, everyone's like, really? You can do that? Like that? I, I, it gets this, it's exciting. Like, you never know how you're going to be received with these kinds of things. I, like I said, I've been in this space for an enormous amount of time. I'm pretty confident in what I'm doing here. And I'm also confident because I'm not, this isn't rocket science. I'm deploying technologies that exist today in a new and novel way because they should have been done a long time ago and nobody got around to doing it. And it solves so many problems and ultimately will save money. It will save billions of dollars in cyber risk, maybe even save your insurance bills. You know, I just think this is a great story and there's going to be some really cool features here because this really gets out a lot of just structural problems we have both between, between the way companies organize themselves and go after their customers, as well as how bad guys try to try to get at this technology. But the era of API keys and tokens and all that, that's gotta go. It's gotta go.
58
00:41:24.000 --> 00:41:34.000
Yeah. I think AI has broken that model for sure. It really, it was always a problem, but it's really exacerbated the problem now with it's just making it that much faster.
59
00:41:34.000 --> 00:42:05.000
Yeah. And it has to operate at all those levels. It can't just be like, we have workload identity and user identity. But man, do we need process identity and sub process identity and all the layers. Because 98% of, you know, AI users, even the guys coding in the big enterprise companies probably have. I mean, it's not like it's advertising what it's doing. When you say go, right, it's like spawning off a bunch of sub agents and running around and changing files and accessing things you didn't know it looked at and sending them back to the server. It's crazy. We got to get this under control 100%.
60
00:42:05.000 --> 00:42:12.000
No, I love it, man. So any other call to action you want folks to know, reach out to, how do they find out information, get a demo, all the things.
61
00:42:12.000 --> 00:42:36.000
Yeah, reach out. Love to talk to people. I'm taking meetings. Design partners are key right now. I'm really excited about that. Be nice to Sit down with some good companies and talk through how this can help them and see if I can get them interested in participating again. I want to get real clear. I'm not asking companies for money. I'm going to go out and raise that money. I just want them along for the ride because they see what this can do for them and want to have early influence on how we build it out.
62
00:42:36.000 --> 00:43:30.000
Yeah, absolutely. Awesome. Well, all the details are in the show notes. Everybody check it out. Frank's left his information in there, how to reach out, all that kind of stuff in there. Definitely reach out. It sounds like an amazing thing. I'd love to. Can't wait to see it in the space and solving some of these complex problems that we have and all the different verticals, from OT to IT cloud and everything in between, obviously, with AI and all the things that that's going to bring, I see it as a great opportunity to dig in and really just that, you know, really secure the entire stack around the way. And then it's more about, you know, what signals can I get, who do I trust, what do I want to trust, what does good look like? And then I'm not as much worried about that layer. I can do more of the things that I want to do. I'm not reacting. I'm more proactive in the space and that that's where we want to get to as, as cyber professionals and honestly, as business owners. Right. Is I just want to make the decisions I want to make not because I'm having to, you know, spend all this money or time or effort because I'm. I'm being forced to do so.
63
00:43:30.000 --> 00:43:31.000
Absolutely.
64
00:43:31.000 --> 00:43:41.000
Well, thank you for your time, man. I appreciate the dive. Deep dive. And it's exciting, man. I'm on your side and, and excited to see grow in all of our products and spaces that we have.
65
00:43:41.000 --> 00:43:55.000
Well, thank you, Aaron. I, I love talking about this. As you can tell. Yammer, yammer, yammer. I just get going, but I'm super passionate. I'm really excited about this and I can't wait to have more conversations. And thank you so much for having me on. It's been a pleasure and I really appreciate it.
66
00:43:55.000 --> 00:43:56.000
All right, man, thanks a lot.
67
00:43:56.000 --> 00:43:57.000
Take care.
68
00:43:57.000 --> 00:44:22.000
Thanks for joining us on PrOTect IT All, where we explore the crossroads of IT and OT cyber security. Remember to subscribe, wherever you, wherever you get your podcasts to stay ahead in this ever evolving field. Until next time.