Observability 2025 with Mehdi Daoudi

Speaker A: Foreign.

Speaker B: Welcome to the Unrivited podcast where we talk about digital transformation, artificial intelligence, and people. Today, we are also sponsored by the book John and I put out, AI in a Weekend, An Executive's Guide. John. John, let's kick us off with our guest.

Speaker C: All right, thank you, Martin, as always. Uh, and thank you everyone for joining us today. Uh, so today we have a special guest, Mehdi Daoudi, uh, who is CEO and co founder of Catch Point Systems Incorporated, as well as a member of 10x CEO. Uh, so in our typical fashion, Medi, uh, I, I give you a little intro and then, uh, who's better to give an intro of yourself than yourself? So, uh, let's turn it over to you. Tell, uh, us a little bit about who you are, how you got to where you are, uh, today. Um, and again, you know, anything that, uh, kind of blends with the interests of our podcast. Digital transformation, AI and people. And then we'll go from there and keep things conversational. So, uh, Mehdi, thanks for joining Unrivited.

Speaker A: Thank you so much, John and Martin. It's a pleasure being on your platform and, uh, I'm looking forward to the conversation. So my name is Mehdi. Uh, I am the CEO of Catchpoint. I've been doing that for 16 years. Before Catchpoint, I worked at this incredible company called DoubleClick. We were the ad serving, uh, pioneers between, I started there in 97 all the way to 2008 when Google acquired DoubleClick. Uh, and before that I was at Reuters where I was doing risk management and things like that. Uh, fun fact. Uh, so the reason I am in doing observability monitoring is, uh, I had the unfortunate, uh, pleasure of turning off DoubleClick for three hours. So I literally, I made a typo in, uh, one of those typical situations where I made the blunder. I took down the system. I was done. We were done for three hours and the God of reliability decided to punish me. And they said, you're going to be in the purgatory of monitoring for the rest of your life. So here I am,

Speaker B: Mehdi. I am a victim of your mistake and thank you for joining us. I am also a self proclaimed Mr. Observability year 2020. So, uh, still holding the title not been dethroned. I appreciate you coming on and admitting your guilt. Um, so I think that leads into a question I was going to ask, which is what inspired you to start Catchpoint? I think you kind of, kind of hinted at it, but maybe what continues to help you drive your Passion for that work.

Speaker A: So what I realized when I was at DoubleClick, so I took over the monitoring team, it was a highly dysfunctional organization in the sense that the business was promising 100% availability. SLAs. We breached that. I had to pay a million and a half dollars to a company called Lycos back in the days, uh, for our breach of sla. And we said, okay, we can't do this stuff. So we transformed the monitoring which we created a team called Quality of Service. We created a culture of performance at Ah, DoubleClick. We made literally delivering ads fast and reliably a part of our business strategy. And I had great leaders and the CEO was on board. So uh, what I saw is not so much the monitoring is the fact that when companies are aligned around observability, around monitoring, and when monitoring makes a company better, then it's fantastic. Right from the business, from a business outcome perspective. So uh, in 2008 my co founders and I said, you know what, the cloud, what was becoming the cloud, which didn't exist, uh, we said this is going to change the way we operate servers, networks, all that stuff. And therefore the monitoring is need to adapt to take that into account. So, so that's what prompted us to start Catchpoint. What fuels me today is still the fact that when we do a good job, um, when our customers use our products and they can deliver things faster, deliver things more reliably, deliver things better in China or Hong Kong or whatever, then there is business outcomes. Right? You can't control the outcomes. You control the process and the effort. And I feel like we are then part of those two ingredients that, that lead to great outcomes. So I, I'm, I'm very fond of that.

Speaker B: That's, that's awesome. Um, you know, it's what's fun to uh, to think about. Around 2008 I, I did a paper on It's Cloudy with the chance of computing and uh, you know, following up, I think there was a famous uh, CEO of a company whose building is shaped sort of like a hard drive. If you know the company name, that's great. If not. And they called the cloud a fad, if you recall. It was kind of interesting, uh, that I would love to play those words over and over again as it's still a fad. It's a heavy fad that just haven't gone away yet. But in that journey you must have come across some really interesting challenges, um, as a leader and I'm sure some of them were more tenuous. Or difficult to, uh, deal with. Can you think of a time when one was, I would say, politically charged and hot? You know, it could be. Could be. Not politically as in.

Speaker A: Yeah, yeah, no, I hear you. Yeah. I'd love to hear your stories. I think there were three that if I look back, um, one of them was Covid and how to adapt to Covid. Because, you know, there is no book that says, hey, or open a book on Amazon. How to run Catch point and page 65, chapter three, how to deal with COVID Right. I think that literally challenged our resiliency and uh. And I think it was an awesome exercise, to be honest. I mean, I think companies that came out of that, uh, alive. Uh, it's a great experience. I mean, I, I don't want to. I don't want to minimize the death and, and all that stuff, but. And uh, I think my grandfather went through the. The first flu, uh, the Spanish flu. Right. Uh, my great grandfather. But the fact that we went through something like this, uh, and experience it and tell stories about it and build companies and resiliency, even with our own children, I think is a fantastic, uh, it's a great learning experience. So that's one. The one that I was the least prepared for is I had the chief Revenue officer that passed away from cancer, uh, two years ago. And that was also completely, uh, uh, like nobody was prepared for that. Right? Not. I don't think we. They prepare our CEOs, VPs, anybody, right. How to deal with the death in. In your leadership team and uh, and then the replacement and, and dealing with the family and dealing with the. The morning inside the company was really, really, uh, interesting. Uh, and we haven't recovered from it, to be honest. I, I haven't even had time to. To deal with it. But that was a very, very, uh, interesting experience.

Speaker B: Interesting. Oh, that's all right.

Speaker A: I didn't mean to be like, uh. So, uh, we'll lift us up.

Speaker B: I'm going to lead us into a good place for forward and onward with. With John. After I, I come out with this one, I will be ready to jump in in a heartbeat. So specific.

Speaker C: We appreciate it.

Speaker B: Absolutely. So specifically, let's go off to a couple things. One, first of all, congratulations. Be pointing in that upper. See if my arrow goes correct.

Speaker A: Yes.

Speaker B: Upper right hand quadrant in the, uh, Gartner.

Speaker A: Uh, thank you.

Speaker B: Uh, congratulations. That's a big honor.

Speaker A: You're.

Speaker B: You're amongst the. The titans and the dinosaurs that are up there. I should say dinosaurs, because some of them don't exist at public companies as they once existed. But, um, know they, they say when companies get acquired, they sometimes go off to retire. But, uh, you're, you're, you're rising. That's awesome. Um, in specific to digital experience and leading into actionable insights. We're going to get there. John, how is Catch Point? How is Catchpoint potentially leveraging AI and machine learning to stay ahead of the competitive landscape and observability?

Speaker A: It's a great question. So I'm not going to lie and say, oh, Catchpoint has an AI thing or whatever. I try not to be, uh, to ever lie about what we can and cannot do. But whether it's called AI or anything else, we're in the data business. At the end of the day, we generate, consume data and present data to customers because data becomes information becomes knowledge. And that's literally the way I look at what we do. And uh, we've been doing this for, for a while. Even at Doubleclick, uh, I used to spend a lot of time looking at data. So what AI is doing and what our mission is, how can we connect the dots faster? At the end of the day, it's like even if you think of 9, 11, the dots were there, but nobody connected the dots, right? Our job in observability is, hey, we're seeing a pattern. Is there something, is there a story? Is there something to investigate? And so what we are doing and what, uh, we'll keep doing is how can we leverage AI for either actionability, uh, information, uh, uh, escalating to the right people, et cetera. So we have an AI strategy that is really around making life of our customers easy. And the way I measure the life of our customers is in very simple terms is meantime to repair. At the end of the day, I look around our customers and say, hey, did our tool help you catch a problem, help you catch a problem faster? Then did it help you identify the problem? Because most of our customers use our tool like, uh, in an emergency room situation, right? In triage, right? Is this happening or not? So if we can do our job faster, then we've earned the fees that we collect from our customers. So AI is really here to help us connect those dots even faster and broader. The only thing I will tell you that we've been able to leverage AI is we've been able to look at all the data we collect across the entire network. So we collect about a few billion events a day, roughly 50, 60 billion events a day. And we say, okay, uh, can we see certain Patterns that are invisible if you just looked at them independently. Right. So are we seeing a fiber cut in Africa that could affect, uh, Meta, for example, or is Adobe having a problem that is impacting 20,000 website? So that's how we're leveraging AI is like, how can we consume as much data as fast as possible? Because that's really what the power of some of these things are. It's just like consumption of data at massive scale that we, we've never been able to see before. Yep.

Speaker C: And Martin was quite right because that just spurred my, my question, uh, my question, uh, break.

Speaker A: Are we still going up?

Speaker C: We're still going up. We're still going up and we're still going in the direction of AI. So, uh, the first question is really, you know, you kind of mentioned, um, that solving, you know, problems using AI tools, uh, or finding patterns that lead to, you know, potential events that could occur within your clients, uh, digital ecosystems. For a lack of a better way of generalizing it, do you think that as people become more and more familiar with AI, uh, that the expectations for clients is going to be for faster and faster resolution of their problems, uh, whether or not AI, as we call it today, which essentially AI today are large language models, um, do you see more and more people coming to, um, you know, Catchpoint or other, you know, companies that operate in the same space and saying, hey, you know, it's 2020 X or whatever, and you guys have all these AI tools. I want my problems resolved in, you know, nanoseconds rather than like.

Speaker A: So let me paint the picture of the future for you, John. So you're the CIO of a bank. Two years from now, a year from now, you're going to wake up, you're going to go to the bathroom, you're going to make yourself a cup of tea or cup of coffee, whatever, whatever your thing is. Or maybe a shot of whiskey if it was a rough night. And, uh, you're going to read your email and something is going to tell you. While you were sleeping last night, uh, we detected, uh, BGP route leak. We automatically took action of it. There was a DDoS attack. And, uh, we engaged with one of our CDN, Akamai Cloudflare, that took care of that. And, uh, there was an API problem in GCP and we rerouted the traffic to AWS for that API service. Uh, there was 1.5 seconds of downtime. Uh, we impacted maybe three customer orders. Everything is great. And then we are writing the postmortem to how to make Things better. That's literally where we're going a year from now. Right now that's not going to be Catchpoint or any of the other monitoring tools. I don't think any of these monitoring tools have the ability to do what I just described. But whether it's Catchpoint, Datadog, Dynatrace, you name it, any of these companies will basically feed data, uh, to other systems that have the ability to connect those dots and also take action. Because monitoring tools like ourselves and others do not have the ability to take action. Right. Only the companies themselves have the automation capabilities. But the monitoring data will drive the automation. But literally what I described to you is what resonates with everybody I talk to. It's like automation is the key. We cannot find people, we cannot hire people to look at dashboards at 2 o' clock in the morning. Uh, the systems are getting so complex. So the automation is going to happen and uh, AI is going to play a big role in that. And then the humans are going to basically the verification, validation and making sure that the automation, uh, plumbing is done correctly.

Speaker C: That led right into the other question that I was going to ask was, um, because Martin and I have, I mean, when we've worked together before, we've, you know, I think we've had the same discussion as, you know, we can't have the expectation of people, uh, staring at a dashboard, waiting for something to break so that they can go and take action. And I'll be interested to hear, I think you started kind of alluding to it. Is that in the monitoring and observability space, uh, for like, you know, SRE type of work, you know, it seems that, you know, when we came from like data analytics to machine learning, Data analytics and machine learning are essentially trying to cover some of the same, uh, space. Right. We're trying to make predictions about the future. And in my opinion, the big differentiator between those two is data analytics is kind of like where the monitoring observability space is like, we crunch all the data, you know, we can pull out the signal, but it still requires the human to take action to perform something to. Yeah, make sure your website doesn't crash. If we think about it in the context of ML or AI, if you will, is saying, like, how do we take that data, identify that pattern and then automatically, you know, salt resolve the problem. And it sounds like that's kind of where, you know, the industry is moving. So, you know, building on that topic a little bit more, I'm, um, interested to hear, you know, not only maybe what, what you and your company are doing, but if anything that you've seen in the industry on moving from that, you know, human oriented action taker, uh, as to someone who, you know, I don't have to wake up at 2 in the morning, uh, because something's happening. Because I know that my system is reliably taking care of whatever the event may be, uh, for you.

Speaker A: Yeah. So I first, it starts with uh, companies being comfortable with automation. Not everybody's comfortable with automation. I remember this is a true story. In uh, 2000 at DoubleClick, we built the shit system. So the shit system was literally self healing it. That was the acronym of this thing and doubleclick. We had this problem where because the volume of ads will grow during the day, we will run out of disk space. Because we were logging all these ads, right? And uh, we literally had, we had 2, 300 people in operations in our ops team that just doing this cleanup. And he said this is insane, this is not scalable. And we had like 15, 20,000 AD servers at the time. It's like this doesn't scale. So we had the guy that built this automation framework and he called it the SHIT system. Self healing it, which is like, okay, I can recognize these patterns, right? I will get an alert from the monitoring team that will say, okay, disk space law, I know how to clean this. I can delete files that are older than three days. Boom, problem solved. So it starts with first engaging with companies that want to automate. And today what I'm seeing already is there are fantastic companies that are on the forefront of this. LinkedIn, eBay and so many others that have taken the first step, which is automating the cdn, the content distribution delivery system. So they want resilience, they want reliability, so they don't stick to one CDN. They use multiple CDNs. And in this case they use our product and other products to basically collect enough intelligence, enough signals. And they have both synthetic signals, which is one of the things we do. And they use our real user monitoring signals. So now they have two visions, uh, two signals coming in and, and they can uh, securely and surely say, okay, the performance of CDN, uh1 is not good in Europe, let's switch to CDN2. Performance of CDN3 in Asia is not great. Let's switch to CDN4 and that is automated at this point. Like people have built enough knowledge and I've seen that over the last five, six, seven years. So, so we're already starting to see that now it's moving to the cloud where I don't know any of our big customers using only one single vendor from a cloud perspective. So they have their stuff in gcp, Azure, uh, aws, Alicloud, et cetera. And now they have services that says, okay, I have this API microservice running on each one of those cloud vendors. And they can literally move the workload like, hey, this is the API checkout. Move it to AWS if it's not working on gcp and no blip recorded. So we're starting to see that. And some companies are moving at that, uh, moving towards automation at the uh, like a French railroad, uh, train. Like it's super fast because they've, they've, they're embracing it. And some companies don't want to embrace automation. But it starts with like, do we want to automate or not? Do we want to free the SRE from doing BS work day in, day out and have them do more interesting work? Because you have an SRE team to build a reliability team. Right. The reliability automation. Not for them to keep doing the same task over and over and over again every day. That's a waste of a very expensive resource.

Speaker B: Wow. So let's unpack a little bit here. We got into um, self healing it shit. Talking about this on shit day, which sure, happy it's Tuesday, which also happens on Thursday. Well, we definitely have an shit squared opportunity here, John. Um, and it's fun to uh, you know, um, you know these acronyms and whether they're homegrown or universal, they have a place in this world and without them the technical world would not operate.

Speaker C: Yeah, I think that uh, should be the title for the podcast for sure. That'll get some eyes on it. Uh, and if not, we'll have to create a podcast. I'm surprised I've never heard that um, acronym before.

Speaker B: So I'm ready to, I'm ready to do a squared, uh, on the title. Yeah. So, because we are recording on a Tuesday. So, um, this is really cool and, and thank you for, for coming on here. If you had to step back and think about what you've accomplished in your career and you, and you had to give yourself advice as a young Mahdi, you know, young Mahdi in his twenties, what do you think you could tell yourself to short circuit some of the gains and get there faster?

Speaker A: Uh, I think uh, I would spend more time on the hiring. Uh, like really hiring is so important. We tend to minimize that and therefore you tend to make mistakes and hire over and over again for the same thing. I would definitely give myself, uh, advice on rereading and being a stoic philosopher. Not, uh, jump the gun. Uh, be resilient from a stoic perspective. Accept failure, deal with it. You can't control what you can control, so stop worrying about it. Uh, that would be, uh, the best advice to a young me is, uh, just focus on the effort and the process. Don't, you know, don't over analyze what can go wrong because you can't control it. And I've spent way, I've lost my hair worrying about things that I can't control. And maybe I would be looking like you, Martin, by now if I, if I would have given myself that advice a long time ago.

Speaker B: I, I always attribute it to the men's hair club. I don't know. I don't know. I, I, I, you know, I can't control what's up here. It's just there, it has a mind of its own. I'm sorry, Maddie.

Speaker A: No worries, no worries. No offense taken.

Speaker B: Well, I want to thank you for being on riveted podcasts. I know, John, we're wrapping up here and I know we both appreciate you coming on. Our listeners will love to, uh, learn more about Catch Point and we'll put the notes into the podcast write up and, uh, we'll make a plug for you. We look forward to catching up with you again, maybe for 2025 and talking about observability. 2025, sometime, if that sounds good, will

Speaker A: be reading customers email saying that's what they've experienced when they woke up. AI took care of stuff while they were asleep.

Speaker B: That's 100%. AI fixed my problems when I was sleeping.

Speaker A: There we go.

Speaker C: Or maybe it will be, maybe we'll be replaced by AI and AI will be reading about AI replacing it, and we'll just be AI replacing itself.

Speaker A: I think people should embrace AI. I think running away from it is, uh, is a huge mistake. I think, uh, we should all use it because it's, it's really super helpful. I mean, I can't work without it on a daily basis. So. Yeah.

Speaker B: Well, thank you for being on the podcast.

Speaker A: Thank you.

Speaker C: All right, thanks, buddy.