Practical AI in IT Observability with LogicMonitor CEO, Christina Kosmowski

Thank you. Thanks so much for having me today. 

Hessie Jones 

OK, so I want to. Pull out a couple of statistics, but before. I do that. I want you. To tell me a little. 

Christina Kosmowski 

Bit about your company. All right, so logic monitor, we’re a hybrid observability platform that’s powered by AI. So what does that actually mean? It means that we collect data natively from thousands of different. Environment. So whether it’s from the network to the infrastructure to the cloud. Containers through the application and then because we collect all that data, we’re able to predict anomalies before they become a problem and either bring their systems completely down or there’s performance issues across your entire IT infrastructure environment. 

Hessie Jones 

1. So here’s some things and I think we just updated some of these metrics. But from a scale perspective, you monitor over a trillion, is that right? Trillion trillion records per day? Three 3 million active devices? Or is that updated as well?

Christina Kosmowski 

Billion records? Yes, as per day. Yes, that’s that’s correct. Around 3 million active devices across 100,000. Users, you know we’re in 30 plus countries, you know, so we’re we’re really excited about our scalability, our extensibility, our depth and breadth of of coverage. 

Hessie Jones 

1. So tell me about. Given the number of metrics that you’re actually supporting, which variables do you specifically prioritize to deliver the kind of service that you do?

Christina Kosmowski 

Yeah. I mean, I think it’s super important that you first and foremost can see everything in your environment. So you need to be able to collect this data from all the different sources. So again, whether it’s your network, whether it’s your database, whether it’s your server, whether it’s a cloud container, you’ve got to be able to see all of it. Collect all of it and then that way you don’t have any blind spots, so once we see and collect it, we then have the context from being able to. Use that that ultimately we can become very predictive and find, you know, with pinpoint accuracy and anomaly before it becomes a problem. And ultimately we can we can, you know, also automate and solve that for for our customers directly as well. 

Hessie Jones 

OK, so you have different companies? Coming in from different verticals. So I would assume that there’s also different thresholds when it comes. To I don’t know, critical infrastructure problems, right? So how do you specifically adapt to each one of those scenarios? 

Christina Kosmowski 

Yeah. So we are a great use case for any type of company across any vertical in any size. So we’ve got a fortune. Healthcare company, you know, big retail name brands, we’ve got financial service institutions, we’ve got your local sports team that all use use logic monitor, but each of that data is collected within their own customer environment data. So it’s their. Data specific to their environment and then we have all the expertise around that type of infrastructure and network data that we collect. And so that data constantly gets smarter because it’s learning from our customers own data and it’s not a bolt on, it’s not a ChatGPT wrapper on top of someone else’s data, it’s actually. Our individual customers data and we don’t share that data across any other customers as well. 

Hessie Jones 

OK so. Let’s talk with some of your top company top clients. You have Coca-Cola, top golf, Airbnb. You said you also had healthcare companies. What are some of the top challenges that that they actually I don’t know try to solve on a day. Day basis. 

Christina Kosmowski 

I mean, first of all, I mean, I think we all see this even in our daily lives. But think about IT environments. They are certainly not getting less complex, they’re getting more complex and their surface areas increasing. So people are adding applications, they’re adding databases, they’re adding infrastructures, they’re moving from the cloud, back from the cloud back on. And so these environments are complicated and so more and more data is coming in as we mentioned just even. On this this podcast we’ve updated. The you know amounts of records that we’re ingesting every single day, and so that’s that’s constantly scaling. So it gets noisy and it’s hard to find the signal between the noise. So if you think about an IT operations. And they’re getting diluted with all these this data, all these alerts and they’re quickly trying to find out what alert actually matters, what one actually can really be a problem for them and make sense of all of it and kind of be able to summarize, you know, all of that together to say, where are these more systemic issues that that I can go proactively? 

Hessie Jones 

So and you mentioned anomalies versus fluctuations and so the systems that you’re using. Are they learning from the? I guess the best practices that the humans have evolved over time, or how much better are they than humans? 

Christina Kosmowski 

 Yeah. I mean, I think, you know, we’ve been using machine learning and stochastic model techniques in our platform since inception. So we were founded in 2007. So for 17 years we’ve been. We’ve been using that, but now with kind of the evolution of the generative AI, we can get even more predictive around where we find these anomalies first and foremost. And then secondly, we can use natural language to summarize this. So now we can start to say, oh, you. 

We had 12,000 alerts. Well now we’re going to bring that down to 100 and be able to say these are the. Specific areas that. Those alerts are happening and now an IT person can actually ask questions. You know, in a in a normal language and say, alright, tell me more about what’s happening here and then we can actually give recommendations on what that root cause analysis is and what they should do. And as they get. Comfortable with that and we can then even start to automate those those recommendations. 

Hessie Jones 

OK, so for your client. Help me understand what’s important for them in order to trust that your system is doing the job that it’s supposed to do. 

Speaker 2 

Doing, yeah, I mean, so we just were working, we just were working with a customer recently and they had an issue that they didn’t know about. Three years and within one hour of giving logic monitor in their environment, they were able to find that. Issue. Yeah, I mentioned the number of alerts that are coming at them. They have 12,500 alerts coming at them. We reduced that by 75% immediately. So I think first and foremost is we’re able to get up and running quickly to show this value early and then they’re seeing these real time results about you know being able to reduce the noise, being able to pinpoint the issue. 

It is impossible to pinpoint and then that’s ultimately making them more efficient so that they can go scale and they can go do more strategic innovative projects instead of being buried in the noise, trying to kind of. Find these these issues. 

Hessie Jones 

Issues. OK, thank you so. So how does your? I guess your generative AI, how is it used? First of all, and how is it helping to adapt and and monitor some of these specific use cases when it comes to the different verticals, but because I’m assuming there’s going to be, there’s going to be. These differences among the different verticals, yeah. 

Christina Kosmowski 

I mean, I think I keep going back to we’re not a bolt on, we don’t have a ChatGPT wrapper, we are actually at the first and foremost, we’re a hybrid observability platform. So we collect this data natively. Therefore we use a combination of rag techniques and kind of small language models that get trained on specific observability data, right? So the large language model such as like open AI or even. You know the open source models, like they’re not trained specifically on observability data and we are. We’re trained on that. That’s our bread and butter. That’s what we do. And so we’re in our own customers environments. We’re already that trusted partner in their environment. We’re getting that data natively and so the models are continually getting smarter. You can approve that budget. 

As the customer’s data is continuing  to kind of grow and evolve with logic monitor. 

Hessie Jones 

So you mentioned Reg. Because I just learned about RAG, probably with everyone else like two months ago. What the heck is that? It’s another not another cleaner. It’s actually a technique called. OK. Remind me what it stands for again, because I just had it. 

Christina Kosmowski 

Runtime. Performance and so this is really where. We we’re we’re not reliant on like a single a single large language model. We’re able to go kind of kind of left and get this data kind of real time and and learn on those models instead of being these dependency on kind of these these larger larger models. 

Hessie jones 

So I knew. I don’t know what the R is, but I know it’s oh runtime augmented processing. Is that what it is? OK. 

So I guess from that perspective, because my next question was going to be how you ensure that it doesn’t inadvertently expose sensitive information from one client to the next by it by training it on only observable data, then using Reg, then you’re actually only ensuring that the information that. As is, is what the. 

Christina Kosmowski 

Customers environment, their environment, their own environment and I think that’s super, super important. And you know we take security very seriously. We’ve been you know, we used across thousands of customers. We’ve been in business for 17 years now. So it’s it’s something that’s very important to. 

Hessie Jones 

So tell me about. What makes you unique? Because you’re considered a hybrid? Nature observability. So tell me about that a little. 

Christina Kosmowski 

Yeah, I mean, nobody else can do that. You know, I think, you know back in the day, you had kind of on premise observability tools that were born in kind of the late 90s, early 2000s, then in, you know, the 2010 with the rise of the hyperscalers, you had folks kind of rush to to monitor the cloud. 

There was nobody that was bridging those two things together and the world is hybrid. So over 86% of companies are hybrid and expect to remain hybrid for the foreseeable future. So you had that plus the fact that you know IT proliferation is continuing to happen at a pace. That we’ve never seen before. We’re really in this unique position where we can see the step and breadth of information in a single unified view that nobody else can. 

Hessie Jones 

1. So one last question, Christina, so we know that it we I think we already ran into an AI height about 1.5 years ago. And now we’re generative. AI is a different it’s a different beach. So how do you ensure that you are developing or you’re delivering tangible value to your clients and not just another round? Of hype that. They have to latch on.

Christina Kosmowski 

Too, yeah, definitely. I mean, we use the term practical AI quite a bit and you know everything we do is we innovate with our customers. So this is something that our customers have been asking us for and we’ve been able to Co innovate with them and get you know, again using their real data, solving real business problems that they have and showing those results. In a quick time to value and that’s really why our customers really love working with us. 

Hessie Jones 

Perfect. Thank you so much. Well, thank you and yes. We will be back.