Name: How Abnormal AI Launched AI In-House
Uploaded: 2025-11-05T21:00:05.203Z
Duration: 59 min 36 s
Description: How Abnormal AI Launched AI In-House

Transcript for "How Abnormal AI Launched AI In-House": Howdy, folks. Thank you so much for joining us today. I'm gonna give everybody just another minute or two to make their way into the session here. But, as long as we are filling time, go ahead and use that chat bar on your right there that you see in Goldcast. Let us know where you're calling in from. Use that same sidebar to ask questions throughout the session today. Highly encourage you to do that. Our team will try and answer everything we can get to live. We will follow-up later by email on on anything else. So you will get an answer one way or another, which at least from where I'm standing means you should feel free to ask us whatever is on your mind. We will try now. Last bit of housekeeping is that you will see a couple of polls pop in during the session. Please take those, so we can see where everyone's at. Try to drive the conversation towards the sorts of things that you would like us to talk about. So that all aside, housekeeping done. Very glad to have everyone here today for what's gonna be, I think, a really cool conversation with Mike Britton. He's the CIO of Abnormal AI, a cybersecurity company that focuses on preventing social engineering email attacks. Really fascinating company. They essentially and stop me if I'm bastardizing your product here, Mike. I'll let you explain this better than I can. But use you use AI to formulate a set of baseline expectations for how users typically behave and then when something is detected that is outside those parameters, that automatically gets filtered out of your inbox. So really cool. Hope I did a passable job of explaining that, Mike. But again, super excited to talk to you today. I'm just gonna ask you to introduce yourself. No. You did a great job. I think you, you you you nailed it pretty well what Abnormal AI does. Thank you. Just a a brief introduction of myself. So I am the CIO here at Abnormal AI. I've been with the company for four and a half years, and I've been in this role for, a little over a year now. Prior to that, I was, the our first CISO here at Abnormal AI. Helped build up our IT security and compliance and customer trust, programs here. And, just excited about, where we've been over the last four and a half years and where we're going in the future. Awesome. Well, thank you for that. And I think we can just kinda kinda jump in here. Because when we first talked, I think we touched on this sort of, maybe not quite paradox, but an interesting situation for you right now. Just the fact that you're a leader at a very prominent AI start up, and much of your mandate at the moment as CIO is actually implementing AI within your organization, much like, you know, you're trying to implement AI at other organizations through the product. So I I think it would be interesting and probably encouraging for a lot of teams out there to know that even an AI company like yours is still having to work pretty intently at implementing that technology into the way they work. Would love just to hear, like, what your experience has been like there kinda leading that initiative. Yeah. In some ways, it's a blessing and a curse. So we have very smart, AI folks and developers and engineers that build our product. I I like to joke and say that, we were AI before AI was cool. For sure. But just because your product is forward thinking and and technology advanced, doesn't necessarily mean all aspects of the business are using AI. And it doesn't mean that, a lot of your back office and and business processes aren't still running in in very manual fashion. So, part of our as we wanna be known as an AI native company on the inside and out, is how do we transform the processes inside? The things like HR, accounting, finance, business sales, anything and everything related to how we run the business. We also wanna be forward thinking there and forward leaning and leveraging AI, because the reality of it is there's a lot of manual processes. There's a lot of manual tasks. There's a lot of things that take a lot of time and effort that really would benefit from the this AI transformation on those processes as well. Yeah. I think the holistic approach there makes a lot of sense. But I think just kind of following on from that just because I I would think this adds to what is a bit of a tangled web there. Abnormal AI is a security platform. You help companies protect themselves, which again, I mean, this made me think I always go back to the who watches the watchmen question. Much of your remit is security. And I'm sure that's front of mind as you're leading all of these implementations. So how are you balancing that aspect knowing that that these tools, while very powerful, can present some unique security challenges? Yeah. One of the things we've done here, we've been very intentional and we we essentially have a moat around our product and our customer data and and how we, provide goods and services to our customers. So that's that's, you know, that's that's ground zero for us. That's the area that we have to be very careful in exactly how we process customers' data, how we use it, the tools we use, you know, acknowledging sub processors, things like that. There's a lot of, you know, the pendulum's kind of swung in in a couple different directions over the last two years where now most organizations, especially larger enterprise, I think Fortune 500, are really starting to realize, okay, we're getting a lot of AI tools in our environment. We're getting a lot of scrutiny. There's a lot of regulation. So we we have to kinda double down on on how we look at tools and products and and vendors using our data and using our data with AI. So we we have to be very careful in exactly what types of AI, how we're processing the data. So all that being said, there's, you know, my customer trust team, works very closely with our product teams and and our engineers to make sure we're absolutely protecting customer data, we're absolutely, you know, processing it, handling it, storing it, doing whatever we say in full accordance with what we tell our customers. That's on the product side. That's on the customer data side. Now, that being said, there's all sorts of things that happen, that are far less benign and easier to kinda be a little more forward leaning and a little bit more, risk tolerant. Simple things like a a sales development rep that wants to do research before they reach out to a potential buyer. There's a lot of manual effort that could go into that of just crawling through web pages, trying to look through notes, things like that. Or we can build a agentic AI solutions to help speed them up, to help make better personalized contacts and connections. Same thing with, the HR side. There's a lot of data that goes into hiring employees, managing employees. There's a lot of ways that we can leverage AI to to make life easier to get answers to employees without having to wait for a human. All sorts of things that can be done where you can take a little bit, you know, we set up guardrails, but you can take a little bit more risk on things that have little to no impact on the customer. Right. And and you mentioned, you know, your customer trust team and and the importance of kind of that open dialogue with the customer. Because I think this is an interesting point of your capacity as CIO too. You're you're really the voice of the customer when it comes to developing security features within the product. So just like, you know, keeping things high level. Of course, you don't have to name the answer. Well and and that's one of the reasons I even found Abnormal AI in the first place. I was an early customer of abnormals, so I knew the product up close and personal. And one of the things I like about my role here is I spent a lot of time with our CTO and our product guys to to really lean in on where should we be going next from a product strategy, what features, what enhancements, what's what's the customer gonna like, what what are they gonna have a little bit of heartburn over. And so I do spend a lot of time with our product managers and and our product folks really leaning in on what are we doing today, what are we doing tomorrow, and what are we doing two years from now to to make sure that, you know, whether it's myself, whether it's my Cisco, whether it's some of these other folks that, have a strong security background. We're we're trying to be that voice of the customer on the inside because we can have those very unfiltered conversations with our engineers and product folks on on what the market expects. Right. And I mean, to drill down on that, like, what what are you finding the market expects? Like, what are customers most concerned about in your experience? Like, what are the things that you that you're coming back and voicing as that voice? Yeah. And it's probably the evolution of a lot of startups where I I really don't consider as a startup anymore. I consider as a a full fledged, mature company in in many aspects. And so, I'm my security program, my IT program, the way we handle our customer data, the way our product works and functions and and is reliable, has has to be on par with the expectations of some of the largest corporations in the in the world. And so drilling down those expectations of we're this is no longer building an MVP. This is no longer building an an early type of product that people are willing to to, take reliability or efficacy type of of issues and and kind of brush them aside, our our products gotten to the point where it's almost like a utility. You walk into your house, you expect your lights to be on, you expect it to work. Nobody thinks twice when you come into your house and you flip on the light switch and your lights come on. And so the same thing needs to be true of our product and how it functions and how it operates and its availability. And so it's building in those processes, building in those controls, building in those systems to make sure that this stuff is accounted for. Right. No. You're you're in the thick of it now. I will go back and edit out the part where I refer to you as as a start up towards the tech. No. I mean, some people still refer to it as a start up. It's I I like to I like to get us out of that mindset because if you think, hey, I'm a start up, then you can operate like a start up. If you think, hey, we're we're really an enterprise, yet we're not the biggest company by far, but we're also not, you know, five five people in a garage either. So it's it's Yeah. You know, you you have to act like where you wanna aspire to be. And if we wanna aspire to be a public company, we wanna aspire to be a company that has 5,000 employees and operates in all sectors and all parts of the globe, then it's better to start acting like that now than to realize, oh my gosh, we're there and we didn't plan or prepare for that. Right. No. Definitely definitely wise. I I wanna back up a little to more of the AI implementation stuff that we were talking about earlier, and and ask you really about prioritization. Because I think it would be easy to get stuck in this sort of we're trying to boil the ocean situation, which with all the range of tools that now exist and new ones coming in all the time. So how do you go about identifying, like, which workflows kind of deserve AI tools first? How do you how do you then measure, like, whether a tool is actually making people more efficient or just creating more work, whether that's for you or another team or how are you going about that that thought process? Yeah. I I think fundamentally, and this is part, I'd love to say that everything has gone smoothly and we're way far along the journey and everything is, you know, just perfect. But I think part of the problem starts in understanding what is the work to be done? What what does a whatever role it is, what is their day to day? What are the manual tasks? What is, what are the objectives and expectations of the role? And really leaning in and understanding what is the current state of work and what what are the important tasks? What are the unimportant tasks? I think part of that is the key fundamental of of any sort of AI transformation is understanding what the current state is. I will say I'm very fortunate in that we my team has a deep partnership with our HR team and our people team is also part of that AI transformation and leaning in and understanding that roles and jobs, have proper definitions, that we have a good understanding of what the work is, that we have a good understanding of what manual tasks are versus things that are already highly automated. So that all feeds into this this process in my AI transformation team. We have an AI transformation team on the technology side. We also have an AI transformation team on the engineering side and we're working very closely together. We kinda have this filter of, because there's a lot of AI tools out there. I think some are solid tools, some are pretenders. And so really, as we understand the work that is right for automation or right for AI transformation, we also have to make a decision. Do we buy? Do we build? Do we use an existing tool? And so that's all part of this formula of really deciding the most bang for your buck, the highest ROI when we talk about AI transformation. Yeah. Another question kind of about decision trees, I suppose, which is which is that of you know, innovation versus another part of your remit, which is compliance. I'm sure a lot of people would just, you know, prefer to kind of pull in the Jenna shop thing here and just go and try everything. Try every tool you have in search of an edge. I think that's a natural instinct. But but you're responsible for governance, which I think puts you in a trickier position there. How have you gone about putting together the framework by which, you know, you you kind of evaluate what passes muster and what doesn't? Sort of a different version of the same question I just asked. But Yeah. And and I think it's important to in that we we do because we service a lot of highly regulated customers, so it's important that it's just not the wild west and anybody can do whatever they want. We do have a very lightweight process. If you wanna try something new, if you wanna try a tool, if you wanna do something, it goes through my my team. We do a quick review. We make sure there's no, risks outside of our risk tolerance. And then we really we lean fast into go try it. Here's the requirements. Here's what it's going to take if you're you know, here's the restrictions. No. Don't use customer data. Don't touch our production environment. And if you wanna buy this tool, we're not buying, you know, we're not gonna buy freemium versions of things. It has to sit behind our security requirements and, our IT stack. It has to go behind single sign on. It has to, have all the governance and controls and and so, yes, go try it. But here's what you need to know if you wanna come back to me and say we're we're gonna buy it. Here's what's at stake. And so that's part of your use case. It's part of your budget conversations with finance. Just understand it's let's try some things. Let's also one of the other things that we're baking in now is, it's easy, especially when you're growing fast and you're scaling and you're trying to transform from a technology perspective. It's easy to get a lot of overlapping tools. So we're also looking at, hey, you know, marketing just asked to do another email AI tool. We've already got these other three. Let's understand what's different about this one versus the other two that you have. Is it really worth bringing in yet another tool? Or should we look and try to consolidate down to a single tool that has 99%? And and what we're also seeing too is as some of these tools pop up, really, honestly, we can build them just as fast. We we've leaned in heavily on AI for development as well. And so why go buy a tool if I can also build that feature in house for, you know, a a fraction of the time and and definitely, none of the cost associated with going out and buying the tool. Yeah. No. I I think that's all really cogent, perspective there. It sounds like you've you've done an a good job of making yourself a partner to those conversations, and I think that and it's and it's not gonna be perfect. There are gonna be times where the business wants to run faster, and that's just where we have to have the conversations. And we're not we're not trying to slow anything down, but we're also just making sure we don't create a a situation or a risk where we come back and and regret moving too fast. Right. You know, no one's risk tolerance is zero, of course. You have to make those. You gotta make those calls. But, kinda on that point, I always like to ask our guests about dealing with growth. It's pretty pretty relatable. And since you are certainly no stranger to that over the last couple years, I do wanna get your your take on, you know, you've got a remote first workforce that's multiplied many times over now, and your time at the company, I'm sure, is still growing quite quickly, while you're also leading at the same time this very involved overhaul of your technology writ large, you know, the way you're going about work. In the course of all that scaling, any has anything been a particular problem? Is there anything you thought would scale better than it did? Any kind of advice for those who may be in shoes like yours, maybe a couple years earlier, managing both an increase in headcount and trying to wrap their heads around all this all this new tooling that's out there. Any any advice for for anyone else? You know, scaling in general has gone much smoother than I would have expected. I I think we probably had a 125 employees when I started four and a half years ago. We're probably pushing fourteen, fifteen hundred in in the the not so distant future. So that's that's a lot of growth. Where it's been I wouldn't say painful, but where we've had some growing pains at times is when we've moved into other locations. I think we were largely US, Canada, Singapore when I started. Now I'm spanning all over the place. I've got, employees in Brazil. I've got employees in, UAE. I've got employees in Japan. And and where it gets difficult is it gets difficult around deploying hardware, things like that. And I I have a variety of suppliers, and I would love a a single supplier that covers every jurisdiction in the timelines I want, but, that's just not something that really exists at the the price point that that I wanna pay. So it's it's a, you know, it's a a conglomerate of different suppliers and processes to to help feed in the stuff. Because part of it too is when someone wants to hire someone, you know, they don't wanna wait six months. They don't wanna wait four weeks for a laptop, things like that. Where where we've also had some growing pains is just getting ahead of the business and getting the business to come to us, a quarter or six months ahead of wanting to hire in a new location, because while IT is one function and factor of of what needs to be accounted for, there's also things like just the legal entity structure, pay, equity, all of those aspects behind getting into a a new location. So, you know, some of that has caused scramble over time and it's just going back to the the business leaders and reiterating, as fast as we do move, some of these things, we just need some better planning and and upfront coordination and communication on. Yeah. No. Of course. I think that's perfectly understandable. Logistics can can be quite daunting. I'm sure. But, I wanna close with more of a, let's call it, borderline philosophical question that I suspect you already spend a good amount of your time thinking about. Because right now, present state, you are implementing AI tools at the moment. And provided that all, you know, goes well, which it it sounds like so far it has, the logical progression of that motion is gonna take us to some really interesting and, I think, newfangled places within the next few years. I think it's, you know, it's it's near horizon stuff. So just picking a, a date out of thin air, let's say, like, five years from now. This could apply to your to your role specifically or the CIO role more broadly in the industry? Like, what does this all look like, you think, crystal ball time, when AI is that much more ingrained into the way we're all working? What's your kind of theory of the case there as to where all of this is headed for for IT and security in particular? I think honestly, if if we do this right, whether it's a year or five years from now, a measure of success would be looking around the room and looking at the CFO, looking at the head of HR, looking at the CIO, and really being able to answer the question from that layer on down, are you the best AI enabled CFO out there? Are you the best AI enabled CIO out there? Are you the best AI enabled HR, business partner out there? And really being able to answer that question because I can go build a bunch of stuff, but what I really need is I need everyone in the company being comfortable with the tools, embracing the change, embracing a new way to work, and and really not having this mindset, and I think we're doing a good job here, is not having this mindset of, oh, well, I don't know how to code so I can't build AI or, that's, you know, too scary for me to to worry about or I don't wanna embrace it because it might eliminate my job. I think the reality and and I've heard this this is not anything unique that I'm saying, but I think we're getting to this point of AI is not gonna replace jobs, but AI is going you know, what we're gonna do is look at people that are leaning in on AI. Those are gonna be the people that replace people that don't wanna get on board with AI. And so it's really we're at this kind of pivotal moment of get on board or there will be someone that's on board that will eventually take over your job. You know, there will probably be roles and parts of roles that will be eliminated as as part of the full AI transformation. I feel like we're still in this state of more AI enablement, AI assistance versus full automation, full autonomy. I think we're still a ways away from that, but, you know, there will be certain things that I can fully automate, certain things that will, you know, mundane tasks that will go away for a human. But I do think the human will still be there from a creativity standpoint, from a human in the loop, control standpoint, from a a guardrails and ethics standpoint. So, you know, I think that's where we're going. Yeah. No. I I think a lot of people would would agree with you. So, that is as good a place to close as any. Mike, I so appreciate you taking the time to talk with me today, and thank you to our audience, of course, for coming out today to join us. But, Mike, I will I will let you go. Thank you once again. Thanks. Thanks for having me. Yeah. You got it. Alright. Everybody, thank you thank you very much to everyone who came out today to watch and engage. We love seeing everybody kind of participating in the chat, taking the polls. All that is so great to see and kinda just makes this worth doing. So, I'm just gonna share some slides here real quick. And I also I know that we have a couple, outstanding questions for Mike. So thank you for asking those. He did have to jump, but we will try and get those answered and follow-up by email. So don't worry about those. We will be in touch with respect to those questions. But, those aside for the moment, while we've got everyone, would love to preview a couple more sessions that we have coming up in the next few weeks. So first, we have this one that you see on your screen there now, which is a really exciting, webinar coming up a week from today. That's gonna cover all things MDM. We tend to run into a lot of people and therefore companies that just don't fully understand what MDM software does and, more importantly, what it does not do. So whether you're in more of an HR ops or chief of staff role or you're managing IT yourself and you're just trying to build a case to leadership as to why you either need a net new or a better MDM solution, this will be a really great chance to hear from our product lead here at Rippling who handles MDM and here we're up to with respect to that product and and what's all been going on in the market right now as it's been quite a couple weeks in that respect too. So that's gonna be on November 12, same time as today. One week after that, same time, the nineteenth, for kind of similar session in terms of the product focus. We're gonna be talking device life cycle management altogether. So here at Rippling, that really means a combination of MDM and what we would call inventory management. And we're also gonna be talking about a really exciting new addition to that product suite, which I have been sworn to secrecy about for another few days, but we'll be discussing that live in that session. So I'd really encourage anyone curious to come through, and we will talk about what we're building there. So suspect that would be of great interest to a lot of you and hope hope I'm right, and then I'll see you there. Same time, same place. That one again is on the nineteenth. With our remaining few minutes here, I will quickly deliver kind of the Rippling IT pitch and give you the sense of how we see the market, and what we can help you do as we're, you know, undergoing one element of the sort of technological transformation that we were just talking about with Mike. So to start with and you can and should check out our IT maturity matrix, which I've linked in the related content today. It's in that docs tab, on your side panel there if you wanna see a more detailed version of this. This is a pretty basic summation of how we have observed IT operations to typically scale. You start manual, and you start looping in some point solutions to take on some of that work and take it out of a manual stage and and start automating bits and pieces. But we believe quite firmly that there is a ton of power in end to end automation specifically, and that is, like, not automation in one tool that then must be linked, again, often manually, to an automation in another tool and so on and so forth in this cascading mess. We see a lot of this in the IT and security market because as many of you are no doubt aware, we have a lot of point solutions in this space. This slide is a very far cry from capturing everything, but just the acronyms alone here, I often like to joke, are quite daunting. There's a lot to keep track of. There's a lot to manage with these tools. There's a lot to patch together if you're using a wide range, of these point solutions. So the thesis behind Rippling IT from the beginning has been bringing as much of that as we can into a singular system, which is built around a singular unified data architecture that allows for end to end automation, which is something you will hear me talk about a lot, because simply put it just empowers IT teams to do more. That is our firm belief here, across identity and access, device management, and inventory management, which are really the three pillars of what that what our platform can do. And that does include warehousing in that third category, by the way. There's a lot that we can help you with. So if any of that sounds of interest to you, please take us up on what I think is a a pretty I need to launch the survey. There it goes. If any of that sounds of interest to you, please take us up on what I think is a pretty fun offer. If you are one of the first five people there it goes. There's my slide. If you're one of the first five people to book and attend a demo for Rippling IT, we will happily gift you a pair of Sennheiser Accentum Accentrum. I'm gonna go Accentum Pro headphones for your time. Those are good. I may not know how to pronounce it, but I have used them. So please take advantage of that if that's of interest. And with that, I believe I will leave it there. So thank you again to everyone who who stuck with us for the whole half hour, and please do let us know if there's anything we can help you with. Again, we will follow-up on any outstanding questions over email. So keep an eye out on your inbox for those. And, yeah. Again, thank you so much for coming, and we hope to see you again soon. So till then, take care, everybody.