Latest / Expert Insights Podcast / #54 - Empowering Productivity By Securing GenAI (Moinul Khan, Aurascape)
Transcript
- 0:09SPEAKER_01: Hello and welcome to the Expert Insights Podcast.
- 0:12SPEAKER_01: I'm Caitlin Harris,
- 0:13SPEAKER_01: Deputy Head of Content at Expert Insights and your host for this episode,
- 0:17SPEAKER_01: which is all about securing the use of AI,
- 0:19SPEAKER_01: Gen AI and AI co-pilots in the workplace.
- 0:22SPEAKER_01: I'm joined today by Moenal Khan,
- 0:25SPEAKER_01: CEO and co-founder at Oriscape,
- 0:27an AI security provider and one of this year's RSAC Innovation
- 0:31Sandbox finalists.
- 0:33SPEAKER_01: Mynell, thanks so much for joining us on the podcast today.
- 0:36SPEAKER_01: It's great to have you on the show.
- 0:38SPEAKER_01: You have over 25 years of experience in cybersecurity.
- 0:42Could you tell us a little bit about your security background
- 0:45and how that led you
- 0:46SPEAKER_01: to founding Oriscape?
- 0:48SPEAKER_00: Sure.
- 0:48SPEAKER_00: So first of all, Caitlin, good morning.
- 0:50SPEAKER_00: And thanks for having me on this podcast.
- 0:54SPEAKER_00: Super excited.
- 0:55SPEAKER_00: Yeah,
- 0:56SPEAKER_00: I have been in the security space for over 25 years,
- 0:59SPEAKER_00: worked in different domains,
- 1:01SPEAKER_00: different setups,
- 1:02SPEAKER_00: small company,
- 1:03SPEAKER_00: mid-size company,
- 1:04SPEAKER_00: large-size company.
- 1:05SPEAKER_00: I also tried a startup before.
- 1:07So before our escape,
- 1:08I was at Zscaler,
- 1:10one of the best cloud security vendor in the industry.
- 1:13SPEAKER_00: I was there for six years.
- 1:14SPEAKER_00: I was their senior vice president and GM.
- 1:17SPEAKER_00: I ran their data security business.
- 1:19SPEAKER_00: Before Zscaler,
- 1:20SPEAKER_00: I was at Palo Alto Networks for about five years,
- 1:23SPEAKER_00: worked in different Palo Alto products.
- 1:26Obviously,
- 1:27Firewall is the core product,
- 1:28but I also focused on a couple of their projects.
- 1:32SPEAKER_00: core cloud-based security products.
- 1:34SPEAKER_00: Before Palo Alto,
- 1:35SPEAKER_00: I was at Netscope when the company was at stealth mode very early on,
- 1:40SPEAKER_00: had a great time,
- 1:41SPEAKER_00: you know,
- 1:42SPEAKER_00: helping the company to create a new market called CASB,
- 1:46SPEAKER_00: Cloud Access Security Brokers.
- 1:47SPEAKER_00: Before Netscope,
- 1:49SPEAKER_00: I was at Juniper Networks,
- 1:51SPEAKER_00: not in their switching and routing department,
- 1:53SPEAKER_00: but in their security BU,
- 1:56SPEAKER_00: where my focus was remote access VPN,
- 1:58SPEAKER_00: SSL VPN,
- 1:59SPEAKER_00: NAC.
- 1:59SPEAKER_00: So,
- 2:00SPEAKER_00: yeah,
- 2:00I was fortunate enough to navigate through different domains
- 2:05within the security space.
- 2:07SPEAKER_01: Absolutely.
- 2:08And I suppose you've drawn on a lot of this experience in
- 2:12different domains in
- 2:13SPEAKER_01: creating or escape.
- 2:15SPEAKER_01: And you've drawn inspiration from the emergence of different companies,
- 2:18SPEAKER_01: for example,
- 2:19SPEAKER_01: Checkpoint during the early stages of the Internet.
- 2:22SPEAKER_01: Palo Alto Networks during the application explosion,
- 2:25SPEAKER_01: and other leading cybersecurity companies born in the cloud era.
- 2:29How have these other companies influenced what you're currently
- 2:32doing in the AI space?
- 2:34SPEAKER_00: Yeah, it's an interesting question.
- 2:37SPEAKER_00: You know,
- 2:37SPEAKER_00: over the last 25 years,
- 2:39SPEAKER_00: to be honest with you,
- 2:40SPEAKER_00: I have witnessed several major transformation in the security industry.
- 2:45SPEAKER_00: I have seen how next-gen firewall replaced traditional firewall.
- 2:49SPEAKER_00: I have seen how cloud-based distributed proxies replaced legacy proxy.
- 2:55I have seen transformation on endpoint security with
- 2:58CrowdStrike and some of the
- 3:01SPEAKER_00: most innovative companies in the world.
- 3:03SPEAKER_00: I have seen MDM, Mobile Device Management.
- 3:05When iPhone and Android became very popular,
- 3:09everybody was looking for a BYOD solution.
- 3:11And I have seen transformation in public cloud,
- 3:14infrastructure,
- 3:14securities,
- 3:15and more.
- 3:16SPEAKER_00: And each time, in hindsight,
- 3:19SPEAKER_00: I found myself thinking, hmm, I could have built that.
- 3:22So this time when AI took off,
- 3:25I knew this is a space that is going to stick around.
- 3:29SPEAKER_00: First of all,
- 3:31there's going to be some transformation and it will stick
- 3:33around for at least the
- 3:35SPEAKER_00: next two decades.
- 3:36SPEAKER_00: As you know, AI is reshaping everything.
- 3:39And I felt that not building cybersecurity solution for AI at
- 3:44this early stage
- 3:45SPEAKER_00: would be a huge missed opportunity.
- 3:49At the same time,
- 3:49you know,
- 3:50I'm a strong believer of the concept of founder,
- 3:54CEO,
- 3:54market fit.
- 3:55I felt that I had accumulated enough experience in the industry
- 3:59to lead in the AI
- 4:01SPEAKER_00: security space that will make meaningful impact.
- 4:05SPEAKER_00: And that sort of like,
- 4:06SPEAKER_00: you know,
- 4:06SPEAKER_00: led me to kind of like,
- 4:08SPEAKER_00: you know,
- 4:08SPEAKER_00: focus on this particular space and then,
- 4:10SPEAKER_00: you know,
- 4:11SPEAKER_00: specialize in AI security.
- 4:13SPEAKER_01: Absolutely.
- 4:14And I definitely want to touch a little bit more on the work
- 4:17that you're actually
- 4:18SPEAKER_01: doing in this space.
- 4:19SPEAKER_01: But before we do that,
- 4:21I want to talk about some of the challenges you're solving and
- 4:24why those challenges
- 4:25SPEAKER_01: actually exist in the first place.
- 4:27Organisations,
- 4:28or at least their employees,
- 4:29are increasingly embracing Gen AI in the workplace.
- 4:32And at Orascape,
- 4:33you're not trying to block AI tools,
- 4:35but make them safer to use.
- 4:37SPEAKER_01: So what are some of the benefits of allowing your employees to use AI,
- 4:42SPEAKER_01: Gen AI and AI agents in the workplace?
- 4:45Yeah,
- 4:46we are a cybersecurity company,
- 4:47but our focus is really enablement,
- 4:50right?
- 4:50We really want to enable organizations so that they can empower
- 4:55their employees to
- 4:56SPEAKER_00: use these AI tools.
- 4:57SPEAKER_00: AI tools are really great tools.
- 5:01These tools make your employees more powerful,
- 5:04more productive,
- 5:05and they should be using it.
- 5:08SPEAKER_00: when you are thinking about allowing AI tools,
- 5:11SPEAKER_00: there is a significant amount of concern from the security,
- 5:14IT security team and CISOs and sometimes even CIOs because when
- 5:20you think about AI tools,
- 5:21SPEAKER_00: at the end of the day,
- 5:22SPEAKER_00: this is like this one additional channel in the internet,
- 5:27SPEAKER_00: right?
- 5:27SPEAKER_00: Like if you look at your employees, your users,
- 5:31They're going to the internet because they're going to
- 5:35different URLs and domains.
- 5:37SPEAKER_00: At the same time, they're using SaaS-based services, and AI is now
- 5:41just another channel which will eventually replace all your
- 5:45existing productivity tool.
- 5:47SPEAKER_00: So if I'm wearing the security hat,
- 5:50my concern is very similar as compared to what I tried to do
- 5:57over the last 30 years.
- 5:58You want to make sure that you can keep the bad guys out
- 6:02through these AI tools.
- 6:04and then you want to make sure that your sensitive data and
- 6:07intellectual property
- 6:08is not leaking out through your insider threats or through
- 6:11accidental data loss.
- 6:13SPEAKER_00: Our focus,
- 6:13SPEAKER_00: again,
- 6:14SPEAKER_00: we are specializing in AI,
- 6:16SPEAKER_00: but when we say we are doing AI security,
- 6:19SPEAKER_00: at the end of the day,
- 6:20we are monitoring what your users are doing when they're using
- 6:23these AI applications.
- 6:25We are also monitoring your application-to-application data
- 6:28flow because all your
- 6:30SPEAKER_00: modern apps today
- 6:31They're connected with some type of AI tools and LLMs and small
- 6:35language models,
- 6:35right?
- 6:36SPEAKER_00: So we are in the path.
- 6:38SPEAKER_00: We are real-time.
- 6:39SPEAKER_00: We are monitoring all these interactions.
- 6:41We are making sure that the threats are not coming in and then
- 6:46your data is not going out.
- 6:47SPEAKER_00: And by doing so,
- 6:48SPEAKER_00: we are giving a tremendous amount of confidence to IT security teams,
- 6:53SPEAKER_00: CISO and CIOs that,
- 6:55SPEAKER_00: hey,
- 6:56you know,
- 6:56empower your users,
- 6:58let them innovate,
- 6:59let them innovate fearlessly.
- 7:01SPEAKER_00: You don't have to worry about security.
- 7:03SPEAKER_00: You do not have to worry about data loss and intellectual property loss.
- 7:08SPEAKER_01: Absolutely.
- 7:08Could you talk through maybe some examples of different
- 7:12security risks you've seen?
- 7:14Some of those data leak risks,
- 7:16ethical concerns,
- 7:17security vulnerabilities,
- 7:18that kind of thing.
- 7:19SPEAKER_00: Sure.
- 7:20You know,
- 7:22I'm sure as I go through these examples,
- 7:24you will be able to resonate.
- 7:26SPEAKER_00: You know, people are going to chat GPT, right?
- 7:28SPEAKER_00: People are going to deep seek, they're going to perplexity.
- 7:31Software developers,
- 7:32they're going to different type of code development tool,
- 7:35right?
- 7:35SPEAKER_00: So when they're doing it,
- 7:37a simple example is if I come to my work this morning and then
- 7:41if I go to chat GPT
- 7:42SPEAKER_00: and if I say,
- 7:44SPEAKER_00: why did Angelina Jolie and Brad Pitt got a divorce?
- 7:48I'm going to get an answer from ChatGPT and most likely that
- 7:51answer is going to be accurate.
- 7:53SPEAKER_00: So,
- 7:53SPEAKER_00: you know,
- 7:54SPEAKER_00: my IT should not worry about like why I asked that question as long as,
- 7:59SPEAKER_00: you know,
- 7:59SPEAKER_00: I'm not violating any legal laws and things like that.
- 8:02SPEAKER_00: But I as a user, if I go to ChatGPT and I say, hey,
- 8:06I need to write a very convincing email to one of the companies
- 8:11that we are about to acquire.
- 8:13SPEAKER_00: And then I go to ChatGPT and I'm putting all this sensitive content.
- 8:18SPEAKER_00: Remember, the ChatGPT is not my corporate sanctioned application.
- 8:21SPEAKER_00: So essentially what I'm doing is I'm doing sensitive data leakage.
- 8:27SPEAKER_00: Any of that information, if it goes out,
- 8:30SPEAKER_00: you know, we are going to risk our business, right?
- 8:33SPEAKER_00: So again,
- 8:34don't go to chat GPT and then start putting your company
- 8:36sensitive information when
- 8:38SPEAKER_00: the application itself is not corporate sanctioned apps,
- 8:41SPEAKER_00: right?
- 8:41If I'm a software developer,
- 8:43it's okay for me to go to a code assistant tool and say,
- 8:48SPEAKER_00: I know C and C Sharp.
- 8:50SPEAKER_00: My manager expects me to learn Python coding.
- 8:53SPEAKER_00: What are some of the best practices for Python coding?
- 8:56SPEAKER_00: Or what are some of the similarities between C, C++, and Python?
- 9:02SPEAKER_00: This is absolutely okay.
- 9:04SPEAKER_00: This tool will train me.
- 9:05SPEAKER_00: It will make me powerful.
- 9:07SPEAKER_00: It will make me more productive.
- 9:09SPEAKER_00: But if the use case is that I go to the same code assistant tool,
- 9:13SPEAKER_00: and then I cut and paste my company source code,
- 9:16SPEAKER_00: and I say,
- 9:16SPEAKER_00: fix my source code,
- 9:17SPEAKER_00: that is going to be absolutely no-no from organization perspective,
- 9:21SPEAKER_00: because now,
- 9:23SPEAKER_00: all of a sudden,
- 9:23SPEAKER_00: you are leaking your sensitive,
- 9:25SPEAKER_00: company-sensitive source code.
- 9:27You are basically exfiltrating data or,
- 9:29you know,
- 9:30putting company at risk,
- 9:31right?
- 9:31SPEAKER_00: So again,
- 9:32SPEAKER_00: but at the same time,
- 9:33SPEAKER_00: if I'm a CISO,
- 9:34SPEAKER_00: and if I say,
- 9:35SPEAKER_00: you know what,
- 9:35SPEAKER_00: let's be sledgehammer,
- 9:37SPEAKER_00: Why worry about these problems?
- 9:39SPEAKER_00: Let's just block all AI tools.
- 9:41SPEAKER_00: Guess what?
- 9:41SPEAKER_00: It's not gonna work, right?
- 9:43SPEAKER_00: User will always find a way to use the AI tools,
- 9:47SPEAKER_00: they use AI application that makes them powerful,
- 9:51SPEAKER_00: right?
- 9:51SPEAKER_00: So again, the question is, how do you allow them
- 9:54but you allow them in such a way where your data,
- 9:57your intellectual property is always secure.
- 10:00You allow them such a way where you don't allow any malicious
- 10:04content to come to
- 10:06SPEAKER_00: your network and then eventually laterally move,
- 10:09SPEAKER_00: right?
- 10:09SPEAKER_00: So again, those are some of the things.
- 10:12SPEAKER_00: Interestingly, these are not made up use cases.
- 10:14Every single time when we are going and talking to an organization,
- 10:18we are talking to
- 10:19SPEAKER_00: CISO and security people, they're doing this.
- 10:22SPEAKER_00: That means, you know, they're resonating with the problem statements.
- 10:25And they also realize that to some extent,
- 10:29they also have false sense of security today.
- 10:31SPEAKER_01: Absolutely.
- 10:32So could you tell me a little bit about how AuraScape helps
- 10:36organizations to
- 10:37address these risks without disrupting employees' productivity
- 10:41so they still feel
- 10:42SPEAKER_01: empowered to be able to use these kinds of tools?
- 10:45SPEAKER_00: Yeah.
- 10:46SPEAKER_00: So the first,
- 10:46SPEAKER_00: you know,
- 10:48SPEAKER_00: we are in line and we are monitoring and enforcing policy real time,
- 10:52SPEAKER_00: which is very different than,
- 10:53SPEAKER_00: you know,
- 10:53SPEAKER_00: a lot of other tools you're going to find in the market,
- 10:56SPEAKER_00: which is all API based and after the fact,
- 10:58SPEAKER_00: right?
- 10:58SPEAKER_00: So we are in line.
- 10:59SPEAKER_00: Essentially what we are doing is we are sort of like a proxy.
- 11:03SPEAKER_00: We sit between the user and all these AI tools in the internet, right?
- 11:07SPEAKER_00: So
- 11:07SPEAKER_00: Everything is going through us.
- 11:09SPEAKER_00: We are inspecting the content on the wire, right?
- 11:13SPEAKER_00: And when we are looking at the content,
- 11:15SPEAKER_00: the purpose is to make it more contextual,
- 11:17SPEAKER_00: to make it very,
- 11:18SPEAKER_00: very powerful,
- 11:19SPEAKER_00: to make it absolutely 100% contextual so that IT team can actually make
- 11:24SPEAKER_00: policy decision, right?
- 11:25SPEAKER_00: So again,
- 11:26some of the examples that I gave you just a few minutes ago is
- 11:30if a software
- 11:31SPEAKER_00: developer is going to a code assistant tool and saying,
- 11:34SPEAKER_00: I want to learn Python,
- 11:35SPEAKER_00: we will look at that interactions and we will say,
- 11:38SPEAKER_00: no risk to the company,
- 11:40SPEAKER_00: right?
- 11:41SPEAKER_00: But if the same user is putting company source code,
- 11:44SPEAKER_00: then A,
- 11:45SPEAKER_00: we are going to be able to alert admin users.
- 11:48SPEAKER_00: If they want to take an action automatically,
- 11:51then they can create an automated policy where that particular
- 11:55transaction within
- 11:56SPEAKER_00: that application will be blocked.
- 11:58SPEAKER_00: So that's sort of like how we do it.
- 12:01SPEAKER_00: We are looking at both outbound and inbound direction.
- 12:04SPEAKER_00: Outbound, the primary focus is data loss and intellectual property loss.
- 12:09SPEAKER_00: inbound at the primary focus is threat prevention.
- 12:11SPEAKER_00: So that's how we do it.
- 12:12SPEAKER_00: Now,
- 12:13SPEAKER_00: as we were developing these services,
- 12:15SPEAKER_00: one of the things that we strive for is the end user experience,
- 12:20SPEAKER_00: okay?
- 12:21SPEAKER_00: If you want to be a sledgehammer, user don't like it,
- 12:25SPEAKER_00: It impacts their user experience.
- 12:27If your solution stack is not efficient enough to really reduce
- 12:33the latency,
- 12:34SPEAKER_00: then nobody is going to use the product,
- 12:36SPEAKER_00: right?
- 12:36SPEAKER_00: So A,
- 12:37SPEAKER_00: while the users are using these applications,
- 12:40how do you do this without introducing absolutely zero latency
- 12:45so that they don't
- 12:45SPEAKER_00: feel that their internet experience is slow?
- 12:48SPEAKER_00: That's number one.
- 12:49SPEAKER_00: And B,
- 12:50if you are trying to block a transaction or even you are
- 12:54running in monitoring mode,
- 12:56what kind of services that you have built so that that properly
- 13:00notifies the user,
- 13:01SPEAKER_00: coach the user,
- 13:03SPEAKER_00: what are the right things to do,
- 13:04SPEAKER_00: what are the wrong things to do,
- 13:05SPEAKER_00: right?
- 13:06So if someone is going to their personal co-pilot account,
- 13:11in line we see that and we will say,
- 13:14SPEAKER_00: would you consider using your corporate co-pilot account, right?
- 13:19SPEAKER_00: Because that way our company sensitive data will be always secure.
- 13:22SPEAKER_00: And then when user sees that message, they feel that
- 13:26SPEAKER_00: I am part of the decision-making process, right?
- 13:28SPEAKER_00: It's not a big guy who is trying to block me, right?
- 13:32SPEAKER_00: So they are more receptive to comply with all your company policies.
- 13:36SPEAKER_00: So again, a huge focus on end-user experience.
- 13:40SPEAKER_00: We have a lot of focus on how do you real-time,
- 13:44SPEAKER_00: how do you coach your users,
- 13:45SPEAKER_00: notify your users as and when you are seeing these violations.
- 13:50SPEAKER_01: Absolutely.
- 13:50I think that's a really good point on making sure that the end
- 13:53user feels part of the decision.
- 13:55SPEAKER_01: Of course,
- 13:55the in the moment training is great for getting them to stop
- 13:58and think about the decision,
- 14:00but having them feel like they're contributing to the security
- 14:03of that organization
- 14:04SPEAKER_01: is also a really big,
- 14:05SPEAKER_01: really big piece there.
- 14:07So we've discussed what AuraScape is currently doing to address
- 14:10some of the
- 14:11SPEAKER_01: challenges in this market.
- 14:12SPEAKER_01: And now I'd like to turn our attention to the future for a little bit.
- 14:16You recently secured $50 million in funding and you're one of
- 14:19this year's RSAC
- 14:21SPEAKER_01: Innovation Sandbox finalists.
- 14:23SPEAKER_01: What plans do you have to continue to innovate in the AI security space?
- 14:28SPEAKER_00: Well, yeah, first of all, we want to become a big cybersecurity company.
- 14:34SPEAKER_00: As AI is becoming mainstream,
- 14:36SPEAKER_00: we feel that we are going to be more relevant,
- 14:39SPEAKER_00: more relevant than everything else that people are using today.
- 14:44SPEAKER_00: So again, we want to be a big cybersecurity company.
- 14:47SPEAKER_00: From innovation perspective, this is what startup is all about, right?
- 14:52SPEAKER_00: You know, the
- 14:53SPEAKER_00: The startups, they should be thinking about innovating, right?
- 14:56SPEAKER_00: It's not about, of course, everybody needs business.
- 14:59SPEAKER_00: We need business.
- 15:00SPEAKER_00: We want our customers to believe it.
- 15:03SPEAKER_00: We want to see ARR growth and all of that stuff.
- 15:05SPEAKER_00: As a CEO,
- 15:06SPEAKER_00: I will let our sales team worry about the sales,
- 15:10SPEAKER_00: but my primary focus is going to be innovation.
- 15:13SPEAKER_00: I'm a product guy, and the innovation really motivates me.
- 15:19SPEAKER_00: This is not about making dead fish and selling them as sushi.
- 15:25SPEAKER_00: So we will continue to innovate within AI space.
- 15:31SPEAKER_00: I think our biggest opportunity is really how we built our platform,
- 15:39SPEAKER_00: how we built our foundational stack.
- 15:41SPEAKER_00: Because when it comes to building features and services,
- 15:46SPEAKER_00: Many people can do it.
- 15:50It's not a difficult task to build features and services when
- 15:53it comes to threat
- 15:54SPEAKER_00: prevention and data security and DLP.
- 15:56But the foundational stack is our differentiator and we will
- 16:01continue to differentiate with our
- 16:03SPEAKER_00: our platform continue to differentiate our foundational stack.
- 16:07In fact,
- 16:08over the last nine months,
- 16:09we filed 14 patents,
- 16:10and these are not about the features.
- 16:12This is about how do you build your platform so that you can
- 16:16quickly build out all
- 16:18SPEAKER_00: these services that organizations need when they think about AI security.
- 16:22SPEAKER_01: Absolutely.
- 16:23We're hearing a lot about the concept of an autonomous SOC
- 16:27analyst or an autonomous SOC team.
- 16:29SPEAKER_01: And although we're not quite there yet,
- 16:31it is something that quite a lot of companies are focusing on
- 16:34developing at the moment.
- 16:36Is that a space that you can see yourselves expanding to cover
- 16:41in terms of securing
- 16:42SPEAKER_01: these analyst tools?
- 16:43SPEAKER_01: Absolutely.
- 16:44SPEAKER_00: Yeah, eventually.
- 16:45SPEAKER_00: Now, our current solution has some part of that already.
- 16:49SPEAKER_00: I talked about our foundational stack.
- 16:51SPEAKER_00: At the end of the day,
- 16:53the easiest way for you to understand is we sort of like build
- 16:56three robots in our
- 16:58SPEAKER_00: foundational stack.
- 16:59SPEAKER_00: Those are basically...
- 17:00SPEAKER_00: you know,
- 17:01SPEAKER_00: AI agents and,
- 17:02SPEAKER_00: you know,
- 17:04that sort of like establishes when I said you really have to
- 17:07leverage AI to fight
- 17:08SPEAKER_00: against AI.
- 17:09SPEAKER_00: One of these robots, what we call is NeuroOps.
- 17:13And what it does is it automates the whole incident management
- 17:17workflow for your
- 17:18SPEAKER_00: admin users and sort of like,
- 17:20SPEAKER_00: you know,
- 17:20SPEAKER_00: automates the interactions between end user and admin users,
- 17:24SPEAKER_00: right?
- 17:25Simple example,
- 17:26I went to ChatGPT and I put some sensitive company information
- 17:30SPEAKER_00: our neuro ops,
- 17:32SPEAKER_00: that robot,
- 17:33SPEAKER_00: will be able to detect it and say,
- 17:35SPEAKER_00: in one hand,
- 17:36SPEAKER_00: it's going to start interacting with the user and say,
- 17:39SPEAKER_00: I have just seen that you're doing that.
- 17:41SPEAKER_00: Do you have a good business justification?
- 17:45And then user then interacts with the robot saying that this is
- 17:48why I absolutely have to do it.
- 17:50SPEAKER_00: That request then goes to admin.
- 17:53SPEAKER_00: And then without admin really doing anything manually,
- 17:56SPEAKER_00: manually changing policy,
- 17:57SPEAKER_00: that AI agent automatically creates an exception for the user.
- 18:02SPEAKER_00: And what we call is LTE, limited time exceptions.
- 18:05SPEAKER_00: So we are going to allow user to do it for next four weeks.
- 18:08SPEAKER_00: four hours,
- 18:09SPEAKER_00: and after four hours,
- 18:11the policy will be automatically changed so that they can no
- 18:14longer do it after
- 18:15SPEAKER_00: four hours,
- 18:16SPEAKER_00: right?
- 18:16So this whole notion of automating the workflow between end
- 18:20user and admin user,
- 18:22SPEAKER_00: you know,
- 18:22SPEAKER_00: it makes our product,
- 18:24SPEAKER_00: you know,
- 18:25SPEAKER_00: much more deployable,
- 18:27SPEAKER_00: right?
- 18:27SPEAKER_00: Like, you know, nobody, the SOC team no longer has time to deal with
- 18:3119,000 alerts that coming from 70 different security tools that
- 18:37they have in place,
- 18:38right?
- 18:39SPEAKER_00: So we sort of like did it,
- 18:40SPEAKER_00: but again,
- 18:41SPEAKER_00: this is just one component of what you said,
- 18:44SPEAKER_00: which is autonomous SOC.
- 18:46Eventually,
- 18:47as we expand and as we provide more security services,
- 18:52we will get there.
- 18:53SPEAKER_01: Yeah, absolutely.
- 18:54It's really exciting seeing what you're doing at the moment and
- 18:57also what you've got planned.
- 18:58I know that myself and the rest of the team at Expert Insights
- 19:01are really looking
- 19:02SPEAKER_01: forward to what's on the horizon for you.
- 19:04SPEAKER_00: Awesome.
- 19:05SPEAKER_00: Thank you.
- 19:06SPEAKER_01: So I've just got one final question for you today.
- 19:09What would be your final words of advice to security teams who
- 19:12are looking to
- 19:13SPEAKER_01: safeguard the use of AI apps and co-pilots within their organization?
- 19:18SPEAKER_00: Yeah,
- 19:18SPEAKER_00: look,
- 19:19SPEAKER_00: you know,
- 19:19SPEAKER_00: as and when we are talking to all enterprise customers,
- 19:22SPEAKER_00: everybody,
- 19:23SPEAKER_00: the enterprise customers today,
- 19:24they have either an internet-facing firewall or they have a
- 19:29cloud-based proxy or
- 19:30SPEAKER_00: some type of proxy for their web traffic.
- 19:32SPEAKER_00: They're using SSC, Secure Service Edge.
- 19:35SPEAKER_00: These are great products, right?
- 19:37SPEAKER_00: These are the products that have no secure enterprise environments.
- 19:41SPEAKER_00: Now, when it comes to AI, AI is very different.
- 19:44It is no longer someone going to a website or someone using a
- 19:48SaaS-based service.
- 19:49SPEAKER_00: And to secure AI, you really, really have to leverage AI, right?
- 19:53SPEAKER_00: You know, it's not about a marketing tagline.
- 19:56SPEAKER_00: My fear is that many,
- 19:59SPEAKER_00: many,
- 19:59SPEAKER_00: many security professionals out there,
- 20:02SPEAKER_00: today they have a false sense of security.
- 20:05They think that I can just use a URL filtering engine to block
- 20:10an AI application,
- 20:11right?
- 20:13SPEAKER_00: I have an HTTP proxy that can monitor all internet traffic.
- 20:17SPEAKER_00: These are all wrong assumptions.
- 20:19SPEAKER_00: And my advice is going to be make sure that you have
- 20:24full,
- 20:24complete visibility for all the AI tools that your users are using.
- 20:28The visibility cannot be just the name of the applications that
- 20:31they are using,
- 20:32but you need to have deep decoders so that you can understand
- 20:36what kind of prompts
- 20:37SPEAKER_00: are going out,
- 20:38SPEAKER_00: what type of responses are coming back.
- 20:40And by doing that,
- 20:41you will have the full context of all of this AI usage.
- 20:44Once you have that full context,
- 20:46then your policy is going to be much more meaningful,
- 20:49right?
- 20:49SPEAKER_00: So again, don't just settle on what you have today.
- 20:54SPEAKER_00: I've seen many inertia problems in the industry, right?
- 20:56SPEAKER_00: It just takes time for people to really admit that they have a problem.
- 21:00SPEAKER_00: So again,
- 21:01SPEAKER_00: pay close attention to AI applications,
- 21:04SPEAKER_00: AI usage,
- 21:05make sure that you have a platform that can truly protect your
- 21:09organization from
- 21:10SPEAKER_00: adversaries and from insider threats.
- 21:12SPEAKER_01: Brilliant.
- 21:13SPEAKER_01: Thank you.
- 21:13I think that's some really actionable advice that our audience
- 21:16are going to find
- 21:17SPEAKER_01: really helpful,
- 21:18SPEAKER_01: especially those that are sort of just starting out on this journey,
- 21:21SPEAKER_01: not really sure where to begin.
- 21:23SPEAKER_01: So thank you very much.
- 21:24SPEAKER_01: And thank you also for joining me on the podcast today.
- 21:27SPEAKER_01: It's been wonderful speaking to you.
- 21:29SPEAKER_00: Absolutely.
- 21:30SPEAKER_00: It's a pleasure.
- 21:31SPEAKER_00: Thank you.