Itai Keren on why traditional EDR fails for AI agents and Bloom's runtime monitoring approach
Summary
Itai Keren, CEO of Bloom Security, argues that traditional endpoint detection and response (EDR) tools are insufficient for securing modern AI endpoints, which now include autonomous agents and plugins. He identifies the rise of 'citizen developers'—non-technical staff building AI tools—as a primary driver of new security risks, noting that AI adoption is accelerating faster than cloud computing. Keren claims Bloom Security uses a permission-based 'deny-list' approach and runtime monitoring to block risky actions without blocking entire tools, reportedly increasing tool usage by 70% among customers. He positions endpoint security as a $30 billion market and warns that organizations not adopting AI securely risk falling behind competitors.
Episode sections
- 0:02 – 1:41
Introduction and Speaker Background
Itai Keren introduces himself as CEO of Bloom Security, noting his previous companies Demisto and Dig Security were acquired by Palo Alto Networks. He states Bloom's mission is to secure enterprise AI usage.
- 1:41 – 3:48
Market Opportunity and Endpoint Evolution
Keren identifies the shift from developer-focused AI tools to business user adoption as the key market driver. He cites endpoint security as the second-largest security market after network security, valued at approximately $30 billion.
- 3:48 – 8:54
The New Endpoint Ecosystem and Risks
The speaker argues traditional EDR is insufficient because endpoints now include AI agents, plugins, and MCPs. He highlights the risk of 'citizen developers' (non-technical staff) using AI to build tools without security awareness, leading to unmonitored data access.
- 8:54 – 10:55
Adoption Velocity and Organizational Awareness
Keren claims AI adoption is accelerating faster than previous tech revolutions like cloud, with organizations often unaware of the extent of employee usage. He notes a shift from confusion to active security planning over the last 12 months.
- 10:55 – 13:49
Policy-Based Security and User Productivity
Bloom uses a 'deny-list' approach, allowing tools by default but blocking specific risky capabilities (e.g., accessing customer data). Keren claims this approach increased tool usage by at least 70% among customers by reducing friction.
- 13:49 – 15:44
Runtime Monitoring and Software Intelligence
The company employs two methods: scanning code artifacts for safety and monitoring runtime behavior against agent intent. This ensures agents do not deviate from their goals or access unauthorized infrastructure.
- 15:44 – 23:53
Contextual Risk and Team Dynamics
Risk assessment is contextual, considering user role and historical behavior. Keren highlights the advantage of Bloom's team, largely composed of former Dig Security and 81st Unit members, which enabled rapid product development and sales within three months.
- 23:53 – 28:10
Future Trends and Call to Action
Keren identifies autonomous and remote agents as the next major trend. He urges all organizations to adopt AI securely, stating that Bloom enables safe usage and that non-adoption leaves companies behind.
Claims worth verifying
Itai Keren's previous companies, Demisto and Dig Security, were acquired by Palo Alto Networks.
Unidentified speaker (speaker_1) ·0:51
How to verify: Check Palo Alto Networks' official press releases or SEC filings (Form 8-K) regarding acquisitions of Demisto and Dig Security. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.
Primary sourceEndpoint security is the second-largest security market after network security, valued at approximately $30 billion.
Unidentified speaker (speaker_1) ·3:08
How to verify: Verify against market research reports from firms like Gartner, IDC, or Forrester on cybersecurity market segmentation and size. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.
Bloom Security sold its product in less than three months after the company started.
Unidentified speaker (speaker_1) ·21:55
How to verify: Check Bloom Security's official website for product launch dates or customer case studies indicating early sales cycles. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.
Bloom customers used at least 70% more tools after implementing Bloom's security policies.
Unidentified speaker (speaker_1) ·12:15
How to verify: Request Bloom Security's internal customer success reports or case studies validating this metric. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.
Most of Bloom's team comes from Dig Security, with others from Demisto or the 81st Unit.
Unidentified speaker (speaker_1) ·21:15
How to verify: Review Bloom Security's LinkedIn company page or press releases detailing team composition and backgrounds. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.
Organizations using AI are at a competitive disadvantage if they do not leverage the technology.
Unidentified speaker (speaker_1) ·7:15
How to verify: This is a subjective claim; verify by analyzing industry reports on AI adoption rates and productivity metrics across sectors. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.
AI adoption is accelerating faster than previous technology revolutions like cloud computing.
Unidentified speaker (speaker_1) ·9:15
How to verify: Compare adoption curves from historical data (e.g., Gartner Hype Cycles) with current AI adoption statistics. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.
Bloom's approach involves monitoring runtime behavior based on agent intent and sequence of actions.
Unidentified speaker (speaker_1) ·14:49
How to verify: Review Bloom Security's technical documentation or whitepapers describing their runtime monitoring methodology. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.
Bloom reduces IT ticket volume by communicating directly with employees when actions are blocked.
Unidentified speaker (speaker_1) ·19:19
How to verify: Check Bloom Security's customer testimonials or case studies regarding IT operational efficiency improvements. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.
Autonomous and remote agents are a major trend in AI security.
Unidentified speaker (speaker_1) ·24:03
How to verify: Monitor industry news and vendor roadmaps for the emergence and adoption of autonomous agent technologies. For recent events, verify 2026-08-19 through 2026-11-19. Use another period only when the speaker explicitly names it.
Do your own research
- Verify the $30 billion endpoint security market valuation using Gartner or IDC market reports.
- Review Bloom Security's 'deny-list' technical whitepapers to assess if permission-based blocking effectively mitigates data exfiltration risks.
- Check Palo Alto Networks' acquisition filings for Demisto and Dig Security to understand the talent and IP base of Bloom Security's team.
- Analyze customer case studies for the claimed 70% increase in tool usage to determine if reduced friction correlates with actual security incidents.
- Research 'MCP' (Model Context Protocol) specifications to understand the technical scope of the new endpoint layer described.
- Look for independent security audits of Bloom Security's runtime monitoring engine to verify claims about agent intent alignment.
Full transcriptexpand
- 0:02
- Unidentified speaker (speaker_0): Welcome back for another episode of Cyber Conversations. Today I am very excited to be joined by Itai Karen, co-founder and CEO of Bloom. And as we all know, AI is changing the way that we work. And I think we're only just starting to understand what this means for cybersecurity. We're seeing people use tools, agents, and applications that give
- 0:22
- Unidentified speaker (speaker_0): them capabilities that they might never have had before. But this also creates a whole new set of challenges for the businesses trying to secure them. So today we're going to explore a lot more about how this is impacting the security landscape and maybe the way businesses need to be thinking about AI becoming a bigger part of everyday
- 0:42
- Unidentified speaker (speaker_0): working life. However, before we get into all of that, it's
- 0:45
- Unidentified speaker (UNKNOWN): such
- 0:45
- Unidentified speaker (speaker_0): a pleasure to have you here with me. Welcome to the podcast. Why don't you maybe just give us a quick introduction to yourself?
- 0:51
- Unidentified speaker (speaker_1): Sure. Katie, thank you so much for having me. Really looking forward to the conversation. So yeah, my name is Itai Karen. I'm the co-founder and CEO at Bloom Security. Quickly about my background, my previous companies were the Misto and Dig Security, both got acquired by Palo Alto Networks. Now building Bloom to make sure that any kind of AI usage
- 1:11
- Unidentified speaker (speaker_1): that we have in enterprises is being done in a secure and safe manner. And yeah, we'd love to talk more about that.
- 1:17
- Unidentified speaker (speaker_0): I think that's exactly what we're gonna do as well. So perfect.
- 1:20
- Unidentified speaker (speaker_0): Let's start with the bigger picture because obviously you've spent a lot of time in cybersecurity, you've worked with some really great businesses, and now you're building Bloom. I think Bloom, you're doing something really exciting here. So interested to know what was it about the current landscape that made you and your co co-founders founders think, hey, there's a bit of a problem here. that maybe we should tackle.
- 1:41
- Unidentified speaker (speaker_1): Sure. So
- 1:43
- Unidentified speaker (speaker_1): I'll describe it a bit broader, right? The way we ideated. We're not really looking into what needs to be secure. It was more about what's changing in the tech world, right? What impacts IT, which security always follows that, right? We know that. Now, that was about a year ago, obviously, AI.
- 2:03
- Unidentified speaker (speaker_1): everybody's talking about AI, but specifically in AI, it was
- 2:08
- Unidentified speaker (speaker_1): changes around how it impacts our desktops. Okay. Now, obviously, we're just after Cursor and Cloud Code that were released. Now that mostly impacted like the tech users, right? The developers and so on. But to us, it was
- 2:28
- Unidentified speaker (speaker_1): clear that the big change is going to be for the business users, the ones that never got these cool tools, that never got the cutting edge technology and all that. And that was very exciting for us to see where's that going, and then go just before that's
- 2:48
- Unidentified speaker (speaker_1): being released to make sure that we're in the right place. to go ahead and enable and secure that usage for the business users today, their code citizen developers and all that. So that was the trend we kind of followed. And then we landed on the on the endpoint as a start. And also, when we talk about the different kinds of
- 3:08
- Unidentified speaker (speaker_1): markets we have in security, endpoint security is the second largest, right? After network, which is about 30%. billion dollars. That's huge, right? So for us, that big change with a huge market was the right place for us to build an amazing and huge company around that.
- 3:24
- Unidentified speaker (speaker_0): I think you addressed something really important there. I don't know whether everybody necessarily realizes how significant or how much of a chunk this actually looks at or covers in cyber when we look at endpoint security. Was there something in particular, you know, like how has the endpoint changed over the last few years? Which about a year ago or so when you guys were looking at this, you were like, oh, okay, this is a problem that we can
- 3:44
- Unidentified speaker (speaker_0): address now that potentially wasn't necessarily needed a couple of years ago.
- 3:48
- Unidentified speaker (speaker_1): Sure. So a hundred percent. I think even a RO there was false sense of security for organizations saying, listen, I have CrowdStrike, I have my EDR, I'm good to go. Why are you talking to me about endpoints? Right? And definitely
- 4:08
- Unidentified speaker (speaker_1): we started talking about the change, but we see how the endpoint as we know it really became an ecosystem of software, right? That we split into two layers, right? The first layer are those platforms. Platform. So IDEs, IDs, browsers, those kind of things that were with us. since forever, right?
- 4:28
- Unidentified speaker (speaker_1): But then a year ago, we start seeing those agents on the desktop. That was pretty new. But as we've never seen browser without extension or ID without plugins, now we're not seeing agents without plugins, MCPs, skills, those kind of things, right? So that's really what forms the
- 4:47
- Unidentified speaker (speaker_1): second layer of the ecosystem that I mentioned. those extensions and so on. Right. And that was a big change on the way the endpoint moved from being an IT-managed device to this kind of different layers of software that the user assembles and the agent assembles themselves when they
- 5:07
- Unidentified speaker (speaker_1): go ahead and start using the amazing technology that we we're are now having.
- 5:12
- Unidentified speaker (speaker_0): I think that's it as well, isn't it? Like there's been this evolution of not just the way that the technology operates, but also the way that enterprises have adopted the way that they're using this. There was definitely a hesitancy around it to start with, and now I feel like it's being embraced. But naturally, like you mentioned, I think there was a false sense of security that oh well, if people are letting us use
- 5:32
- Unidentified speaker (speaker_0): these tools, then of course they're going to be secure. Unless you maybe come from the industry, you don't necessarily recognize that that's not all. always on the forefront of people's minds, even though it probably should be. I think to dive into that a little bit more, one of the biggest challenges that we're seeing with AI, it's not just how it's changing how technical people work, but obviously it's giving the business the
- 5:52
- Unidentified speaker (speaker_0): ability to do things or people within those businesses where they probably couldn't have done them before. You don't necessarily now have to be a software development specialist to build applications that you can then start using in your day-to-day. But also it obviously allows people to enhance their skill set. How do you think this particular part of it
- 6:12
- Unidentified speaker (speaker_0): is changing the security landscape?
- 6:15
- Unidentified speaker (speaker_1): So I think that's the biggest trend that impacts security these days, right? Because as we've said, developers always got amazing tools. They were always cutting edge and using the top-notch tools, right? So fine, they got huge boosts because of cursor and cloud call and codex and whatnot.
- 6:35
- Unidentified speaker (speaker_1): But really, the fact to your point, KD, that today finance teams, right? HR admins, right? Everybody, marketing Team can build all of a sudden things that they never could actually build, right? They've always needed developers to build those tools, but now marketing teams
- 6:54
- Unidentified speaker (speaker_1): building campaign portals, right? Or dashboards and those kinds of things, right? Finance building agentic workflows to basically take all their manual work and process that automatically, right? So that's a huge change, and that's a big boost to the businesses, right? Today, an organization that does not leverage that technology is
- 7:15
- Unidentified speaker (speaker_1): behind. So for them, it's amazing. But those guys never got any kind of security training or any kind of awareness for security. They needed that, right? So that's where the risk starts, right? They're running things, they're just putting in prompts, they get an amazing dashboard out of that. But a lot of
- 7:34
- Unidentified speaker (speaker_1): append happened between that prompt, they the input, and the result they got. That agent probably tried to download code packages to build that UI, right? That agent probably accessed databases or the CRM or Jira or those kind of things to basically get the data. So a lot has happened in
- 7:54
- Unidentified speaker (speaker_1): between, and they didn't really know. And they honestly couldn't They care less, right? All they need is the their result. So I think that's where the risk starts and you know. it goes much beyond that.
- 8:05
- Unidentified speaker (speaker_0): Yeah, I think it's sometimes just this lack of awareness. And it's often really easy to forget that when you work within cybersecurity, how obvious these things seem. But when you're not part of the industry, when it's not a conversation that's frequently brought up, I think there's an element of aligned trust, like I mentioned, that gets put into these companies and these products. Ultimately, as well, there's probably an element
- 8:25
- Unidentified speaker (speaker_0): of that's it's easy to bury your head in the sand because it's also not my fault or my responsibility a lot of the time, which I suppose is also another question of where the responsibility falls when we use these technologies. Do you think businesses are fully aware of what's happening with their employees and their devices and they are actively trying to address it? Do you think they're burying their head in the sand, or do
- 8:45
- Unidentified speaker (speaker_0): you think businesses just don't quite realise realize the extent that maybe employees are utilising utilizing these? kind of technologies without the security measures in place?
- 8:54
- Unidentified speaker (speaker_1): I think just like the previous revolutions we had with internet and then cloud and now obviously AI, definitely the adoption comes even before the organization really understands it happens. But I think with AI it's like on steroids, right? Because today it's much easier to
- 9:15
- Unidentified speaker (speaker_1): basically use an AI tool than it was easier to basically spin up a cloud resource, right? So I think today anyone could just use any kind of skill. At the end of the day, skill is just text that you could copy-paste, and then there you go. We have an agentic on our endpoints. So I think there's been a huge
- 9:34
- Unidentified speaker (speaker_1): progress in the way organizations were looking at that risk. 12 months ago to where they're looking at that today. They are much more aware the fact that it happens. But we're just seeing the tip of the iceberg in the both from the adoption perspective, but then definitely from the security perspective of that. So
- 9:55
- Unidentified speaker (speaker_1): I think organizations definitely are have more appetite to allow employees to use AI, have more appetite to basically not live with that risk and make sure that they're securing the usage. So definitely I think we're seeing a massive trend. And for us, this is like great news.
- 10:13
- Unidentified speaker (speaker_0): Yeah. And I think that's kind of it as well, isn't it? Because there's a real balancing act here for businesses when companies are embracing this kind of technology, which arguably is then almost giving the company more for their money from what these people are now able. to do, which is a great benefit. So you don't want to stop them using tools that make them more productive or let them automate a
- 10:33
- Unidentified speaker (speaker_0): lot of their work and therefore really lean into the part where we can't replace the human. But you of course also can't just let everyone connect whatever they want to company systems, and you can't just use whatever random tools without being assessed first. How do you get that balance right where you're not capping the sea? ceiling, but you're not going, we just have no way to
- 10:53
- Unidentified speaker (speaker_0): measure the risk here?
- 10:55
- Unidentified speaker (speaker_1): Yeah, sure. That's a great point. And and that's the challenge really, Katie, right? I mean, because everybody wants to adopt AI and the market is in this anywhere between confusion and panic mode into all right, what AI tools should I use? How do I secure them? I must use them. Like the board is pushing me, the CEO is pushing me. So
- 11:15
- Unidentified speaker (speaker_1): definitely seeing a lot of confusion around that. And then, of course, they don't want to slow down the business. They must not slow down the business. So definitely they're looking on ways to allow that. And the way we approach it is that we are looking at tell me what kind of capabilities you
- 11:35
- Unidentified speaker (speaker_1): want not to allow, right? Because you definitely want to allow more than you want to block, that's for sure, right? So, for example, if you want your tools not to interact with DeepSec whatsoever, if you want your tools not to process customer data, if you want your tools to be able to not be able to do those
- 11:55
- Unidentified speaker (speaker_1): specific things, that's great. I can work with that, I can make sure that every tool basically is allowed. but is not allowed to do those specific actions, right? So that's the way we're seeing talk to me in policies, talk to me in capabilities, right? I will enforce those are not allowed, right? And
- 12:15
- Unidentified speaker (speaker_1): eventually, what we found out in our customer base is that they ended up using 70% at least more tools than they were using before, right? And much more, much faster, right. So I think when we make that kind of change in the paradigm, the way IT and security teams are controlling
- 12:35
- Unidentified speaker (speaker_1): their desktops, then you see a massive change in the benefit of the organization and the business.
- 12:41
- Unidentified speaker (speaker_0): Yeah, and I think as well, that kind of goes into that trust element, doesn't it? If you can trust that the tools that you're using are secure and you can just use them to your full advantage, like you'll go, Great, like I'll go all in here. And naturally it'll also start to become part of your I think you're like everybody now is really starting to recognise how much we maybe lean on this technology as
- 13:01
- Unidentified speaker (speaker_0): well. Like it is becoming part of day to day life of how we operate in the workplace, which absolutely is a great benefit, but it does then mean that we need to trust that those tools are secure or at least connecting to what they're allowed to connect to and using. the information that they're allowed to. Because I think when we've traditionally thought about endpoint security in terms of detecting
- 13:21
- Unidentified speaker (speaker_0): malware or stopping something malicious, that's been the main way I suppose we've looked or categorized endpoint. But I think with AI, the risk isn't necessarily that it's just the tool itself that's bad. It could be like a perfectly legitimate tool that has access to some sensitive company information. When it comes to determining whether something is actually risky in a
- 13:41
- Unidentified speaker (speaker_0): particular environment, is it a case that you're looking at the tool itself, how it's being used, or what it's connecting to? Is it a bit of a combination?
- 13:49
- Unidentified speaker (speaker_1): So we have two kinds of ways we tackle that specific point, right? First, we build our software intelligence, right? That software intelligence basically leverage amazing technology that we build. to basically scan and sandbox all those code artifacts, the add-ons, the plugins, the MCPs,
- 14:09
- Unidentified speaker (speaker_1): the skills, and making sure that we have this intelligent data set that can tell us anything we need to know from the security and compliance perspective of any kind of software that lives in the world. right? Once we have that, it's very easy for us to say anything that goes. beyond our gate is
- 14:29
- Unidentified speaker (speaker_1): known, is vetted, and is safe and secure to use within the organization. Now, once it's in, the next thing we need to do, and this is the second thing we're doing, is to look how it's configured and how it's being used. Now, for that, we're actually monitoring in runtime based on the goal of the agent session and the sequence
- 14:49
- Unidentified speaker (speaker_1): of actions to follow that. including the critical infrastructure it connects, the sensitive data it access, and so on, to make sure that the agent does not misalign with its intent. All right. So that's how we're basically monitoring anything that it's doing, anything that it's connecting to, anything that it touches, and make sure that it doesn't go wrong from its original
- 15:09
- Unidentified speaker (speaker_1): intent, and making sure that we are enforcing in runtime that the agents are doing what they're supposed to do. So that's software intelligence and enforcing what's installed. Next is the intent-based analysis and enforcing in runtime what they're actually doing.
- 15:24
- Unidentified speaker (speaker_0): And so when we look at it from that as well, then so obviously we've assessed, okay, like the agent is working how it should. Does it ever have the impacts where Well, it depends on the employee that's using it? So for example, like the same tool can be completely fine. and approved for this particular employee working in this role, but could it then create a risk for
- 15:44
- Unidentified speaker (speaker_0): somebody else depending on their role, what they can access? How do you maybe approach that? Okay, it's not necessarily company wide approved, it's like almost like this person approved, or is it a case that's actually irrelevant and once the product has been assessed, it's okay, it's got those guardrails that stop it doing what it shouldn't?
- 16:03
- Unidentified speaker (speaker_1): Yeah, 100%. I think today the desktop change on every click, every download, every installation, anything, right? So just looking at a tool like disconnected from the endpoint it lives on is irrelevant. You can give it any kind of security score, but then do you have sensitive data on the
- 16:22
- Unidentified speaker (speaker_1): endpoint? Can that tool access sensitive data from my in my CRM and those kind of things? What other softwares are living alongside? Is there like a VPN that intercepts any kind of network traffic this specific tool sends? And more generic, like we need to profilize the different users and
- 16:42
- Unidentified speaker (speaker_1): understand: oh, that user never really ran any kind of PowerShell command. Why all of a sudden its agents start doing that? That user never really connected or touched customer data, right? Why does it do it all of a sudden, right? So definitely you need to look at the context of that specific
- 17:02
- Unidentified speaker (speaker_1): desktop, that specific user. What does it usually do? What team that user belongs to? What kind of industry that organization lives with, and what's that specific industry at the moment? So I think. with the velocity of AI with so many parameters, if you don't take into equation all those different
- 17:21
- Unidentified speaker (speaker_1): various parameters, you will just bring noise which is far worse than not even securing things.
- 17:28
- Unidentified speaker (speaker_0): I think for any security leaders that are tuning in, they could probably sit there and be like, that's quite overwhelming because suddenly there's all these different things I've got to assess. And this is obviously where
- 17:38
- Unidentified speaker (UNKNOWN): I want
- 17:38
- Unidentified speaker (speaker_0): to bring it back to bloom a little bit because of course this is where you guys then step in and really help simplify this. Talk to us about from a security leader's perspective, how is it a case that you can go from there's actually all these really complex factors that need to be not just addressed from the tool, but from who's using it, where they're using it, how they're using it. How do you guys support
- 17:58
- Unidentified speaker (speaker_0): with managing that?
- 17:59
- Unidentified speaker (speaker_1): So beyond the security perspective and governance side. of things, so we talked about how we're autonomously making sure that what's installed is safe, remediating existing problems, analyzing the intent of the actions, but then beyond that, there is the operationalization of things, right? So at the end of the day, behind that
- 18:19
- Unidentified speaker (speaker_1): desktop, there's a user that needs to do his work, right? So just imagine, Katie, that someone would take your favorite tool and just remove. it. That's frustrating. Someone would take your agent and block anything you try to do. That's annoying, right? So that's why we are doubling down on the communication with the end user.
- 18:39
- Unidentified speaker (speaker_1): So in anything that we ever do, we will always communicate with the employee. We'll will say, or its agents, by the way, right? So it could be, listen, you tried to block. to install that dangerous skill, all right? We block that. Here's why, right? And then get them alternative, allow them to ask for more
- 18:59
- Unidentified speaker (speaker_1): detail, ask for exception, anything they need, right? And then what happens is really two things, okay. One, the employee got someone to talk to, all right. And from the other end, security teams and IT teams, the only thing they really hate more than tickets, like they're more than hours, is tickets, right? So
- 19:19
- Unidentified speaker (speaker_1): now no more tickets, right? We're seeing a huge reduction in the number of tickets they need to handle because we ended up throwing them seamlessly. Okay,
- 19:27
- Unidentified speaker (UNKNOWN): so
- 19:27
- Unidentified speaker (speaker_1): that kind of operationalization on top of their amazing policies that we enforce in runtime are done by this kind of communication and a seamless interaction between the security and the employees.
- 19:40
- Unidentified speaker (speaker_0): And there's something there that I think is definitely worth highlighting is the fact that you give kind of that feedback, that context of not just saying, Oh, you can't do this, and this is why we've stopped this particular action. It's a case of explaining why. And I think a lot of the time that it people really underestimate the value of context of this is actually the negative
- 20:00
- Unidentified speaker (speaker_0): implications of what you could be doing. You probably didn't know. this. You don't know what you don't know, but now you do know. And also here's some solutions to help with that. And I think that can be something that's quite easily dismissed because it's almost like an extra add-on of we don't have to do that. But actually it really brings the long term value from that employee's perspective of then not wanting to just try and
- 20:20
- Unidentified speaker (speaker_0): oh well you've put hurdle in my way I'm now just going gonna to try and make my way around it. I'll actually listen to what you're saying. There's another thing I want to mention as well about Bloom, though, because I think you guys have a really unique way of that you've built the business in the sense that the team behind Bloom, right? Like you guys have got a lot of people who have worked together for years,
- 20:40
- Unidentified speaker (speaker_0): particularly through your time at Dig. How important do you think that shared experience was, and how do you think maybe it's impacted the way that you've been able to build something new? Because you guys obviously you've been able to do quite a lot of exciting stuff in arguably quite a short space of time.
- 20:55
- Unidentified speaker (speaker_1): So that's a great point because I'm very, very proud of our team. And yeah, as you said, most of the team is coming from our network in the past. So most of the company today comes from dig security, right? Post Post-acquisition acquisition of Palo. The rest are coming from the Misto or from the 81 unit.
- 21:15
- Unidentified speaker (speaker_1): of one of my partners' network. And when you have in day one of the company, like 20-something people that already know each other, right? They've been building together for about five years at this point. It's like an organic team, and they just keep building, right? There is no overhead of like, oh, we need to be
- 21:35
- Unidentified speaker (speaker_1): polite to each other, we need to get to know each other. You can start. going very quickly and when you have AI on your end, you understand that honestly, from the engineering perspective, there are no limitations, right? You can build anything in no time. And we saw the impact. We were able to sell the product in less than three months than we started the company,
- 21:55
- Unidentified speaker (speaker_1): which is great. That's something that's unheard Heard of a few years ago, right? So I think the people behind Bloom is
- 22:03
- Unidentified speaker (speaker_1): the most important thing for me. And beyond that, it's the team that's actually been building like secure enterprise security solution. And that's very important because you know what they're actually looking for. So sure, you need AI usage control, you need supply chain risk prevention. But what about the enterprise readiness? Right?
- 22:23
- Unidentified speaker (speaker_1): That's very important since day one, since our ICPR ICP used. are huge enterprises, right? You have to have that audit log, you need to have R back, you need to have all those things. And if you don't have them in day one, you're in a problem. You won't be able to sell to those organizations. So that experience and obviously our reputation goes with us because when people
- 22:43
- Unidentified speaker (speaker_1): hear that we build D, we build the misto and those kind of things, they'll We'll
- 22:46
- Unidentified speaker (UNKNOWN): say,
- 22:47
- Unidentified speaker (speaker_1): Oh, I know those. That's
- 22:49
- Unidentified speaker (speaker_1): I trust that team. So that experience for us gives us amazing value.
- 22:54
- Unidentified speaker (speaker_0): You can see how quickly you guys were securing customers. It is pretty much unheard of within the industry. But also I think it always circles back to this trust element, whether it's trusting the technology, trusting your team. I suppose that's just what allows people to build and scale, no matter what aspect you're looking at it from. And I
- 23:12
- Unidentified speaker (UNKNOWN): think it's
- 23:12
- Unidentified speaker (speaker_0): another thing that's testament to how quickly you guys are going to continue to be able to evolve the challenge you have when you're working against solutions that are specifically against AI technologies, the fact of how quickly it evolves. Therefore you have to keep pace, which isn't easy, but when you mention you've got a team like that that trusts each other
- 23:32
- Unidentified speaker (speaker_0): and can just run with that innovation is great. But obviously the market is still moving incredibly quickly at the moment. And I think with the amount of investments, the new companies that are emerging around AI, how do you see this space going to develop specifically around what you guys are doing, but also how we see enterprises utilizing
- 23:52
- Unidentified speaker (speaker_0): AI technology?
- 23:53
- Unidentified speaker (speaker_1): So
- 23:55
- Unidentified speaker (speaker_1): what I'm saying right now would probably be relevant only for the next 10 minutes with how
- 23:59
- Unidentified speaker (UNKNOWN): things are moving with
- 24:01
- Unidentified speaker (speaker_1): AI. But definitely we're seeing
- 24:03
- Unidentified speaker (speaker_1): the trend of autonomous agents and remote agents, like, for example, managed the Anthropics Cloud and those kind of things. We know about Grokbot. We see that as a big trend. Organizations are starting or thinking of adopting it. Some of them, like the more forward thinking, are already trying that.
- 24:23
- Unidentified speaker (speaker_1): So we're seeing another type of agents that the end user is interacting with. And for us, it was very natural to go ahead and secure those because at the end of the day, those are agents running on endpoint just in other environments. So for us, there wasn't a big change needed to be done. So
- 24:43
- Unidentified speaker (speaker_1): that's where we're seeing AI is It's going to, like the cloud remote ones, the SaaS agents, also very important, running in all kinds. kind of platform that enterprises use. And this is where we see the future.
- 24:55
- Unidentified speaker (speaker_0): When we see all this different technology in the future that's coming forward, it's exciting, right? As much as sometimes I look at it and I think it's daunting. It is exciting. How do you think in a few years, how different is the way that we're working going to be? Do you think the way that enterprises operate is going to shift or do you think this technology's technology is
- 25:15
- Unidentified speaker (speaker_0): ideally just about enhancing what people can already do? And do you think maybe this means that security teams have got even more responsibility for keeping everything safe? I suppose this is a hard one to predict because like you say, who knows, a prediction that's accurate today could be totally wrong by tomorrow. But interested to hear your thoughts.
- 25:34
- Unidentified speaker (speaker_1): I think just like we're we had in the cloud, there was a lot of mess. and chaos, and like who's in charge of that, who's handling that, who's the owner of that? When you ask today 10 different organizations who owns AI security, you will get 10 different or 11, maybe different answers, right? So I think it will take some
- 25:54
- Unidentified speaker (speaker_1): time for that thing to stabilize and understand who's in charge of that. I think eventually we'll get like stable platforms that connect to their existing. workflow and processes of the organization. And I think we'll get to a point where this is controlled and maintained, and it won't be really more over for the security team until the
- 26:13
- Unidentified speaker (speaker_1): next revolution comes and they will get like swamped again all over together.
- 26:18
- Unidentified speaker (speaker_0): I think a part of that is really exciting, and it's interesting because I think, like we say, it's evolving so quickly, it can be really hard to predict. I think there's certain things that we know are going to happen, and there's certain things like it could go this way, but arguably it could become even more advanced than that, and we just don't know it yet. But I thank you so much for being here today and joining me on this episode because I think it's been
- 26:38
- Unidentified speaker (speaker_0): great to learn a bit more about Bloom, but equally deep dive into a huge part of the industry that realistically I think it deserves the attention that it gets because it's a huge problem, like you mentioned, second behind network. networks. It's definitely something that I think people really need to take into consideration. Just to close out the episode, if people were
- 26:58
- Unidentified speaker (speaker_0): to, in your opinion, anything that they could take away from today's episode, anything that you would say, look, even if there was just one thing that you took away from this particular episode when you've tuned in, this arguably would be the most important thing.
- 27:10
- Unidentified speaker (speaker_1): I'm calling to all organizations today, all businesses to use AI. This is very important because when I hear about an organization that tells me that they don't use AI, I'm like, so what are you doing? Right? So I'm calling them all to use that. I do understand the concerns, I do understand the risk very well. But if you have Bloom while you're
- 27:30
- Unidentified speaker (speaker_1): adopting AI, then you're good to go. So use AI, use it safely, use it securely, and use it with Bloom.
- 27:37
- Unidentified speaker (speaker_0): I absolutely love it. Ultimately, why would you want to be scared of a technology that can offer so much opportunity, right? Honestly, this has been a great episode. Thank you so much for joining me. To everybody tuning in, that is it for today's episode. Make sure to share, like, subscribe, share this episode with your security peers, and make sure to follow along with the team over at Bloom.
- 27:57
- Unidentified speaker (speaker_0): I'm sure they Tay will be more than happy to help out if you've got any further questions on how Bloom may be able to. help. That is it for today. So thank you all and I will see you all next time for more conversations that are shaping cyber.