Building Autonomous IT With Agentic AI

Hey everyone. Welcome back here to Techstrong TV. My next guest is Garth Fort.

Garth is the CPO, chief product officer, over at LogicMonitor. Garth, welcome. It's great to have you on.

How's everything? Everything's great. Good.

Good to hear. Hey, Garth, I always like to let our audience know who they're listening to, who they're watching here. Give a sense, I mentioned you're the chief product officer at LogicMonitor, but you've done other things in your life, I'm sure.

Share with our audience a little bit of your story. Sure. I was a biochemistry major who entered into Microsoft in the mid-'90s.

Those of you who want to go really deep back into the Google archives will find that I was the first person to introduce the world to Clippy. That was Microsoft's first- Oh, God. Yeah.

Hey, an early AI. Yep. It was the first attempt of Microsoft to introduce...

It was machine learning at the time, but anyway. Yeah. So I've been kind of involved in that for a while.

I spent 20-plus years at Microsoft. Went over to AWS to learn how to build and run cloud services at scale, and that was a great thing. Then I went over to Splunk, and then as Splunk was getting acquired by Cisco, I took a bit of a hiatus that eventually brought me over to LogicMonitor, and I was excited to be here.

Excellent. That's a great history of some really monster successful companies and projects. I actually just interviewed, who was the CEO at Splunk?

Well, there were two. When they sold it. Gary was the most recent CEO, or are you talking about Doug?

Doug Merritt. Yeah. Doug Merritt.

Yeah. He's now at- That's who it is. I had a brain freeze on his name.

Yeah. He's at Aviatrix right now. Yes, he is.

I just interviewed him, I guess it was about two weeks ago. Yeah. Because it was funny, I was at Cisco Live last week, and I interviewed the GM for the Splunk division there at Cisco Live.

But the week before, I actually interviewed Doug. Well, he's at Aviatrix now, as you said. Yeah.

So, small world. But it is quite a celebrated list there, Garth. Congratulations.

" Some may know, a lot aren't going to know. How would you describe LogicMonitor to the audience? Well, when I was meeting the management team, LogicMonitor started in 2007 as really a network performance monitoring company.

So it was really deep in that. And then over time, they've expanded to this hybrid estate. People just didn't need to understand their own network.

They wanted to understand how their applications were performing across a hybrid estate. And so I think LogicMonitor evolved quite a bit, before I even got there. But LogicMonitor right now really just helps customers understand the full performance of their estate.

We did this acquisition of Catchpoint, that we announced in December of last year. Mm-hmm. But really, our goal is to help customers understand the performance of every packet that goes in and out of every nick on any device, whether that's running in your own data center or in the cloud.

And then being able to understand what's the end-user customer experience. And so we can understand how every pixel is rendered on the Chrome browser in Croatia, et cetera, and that's through the Catchpoint acquisition. And so, we want to help IT professionals understand literally the performance of every packet going in and out of the nick of every device, and then how that translates into how pixels are rendering on the end-user device that people care about.

I love it. I think it's a great way of describing nuts and bolts what you do. Yeah.

And of course, LogicMonitor's been around a while itself. It recently announced a kind of milestone in terms of ARR, right? Mm-hmm.

I think it was $400 million a year now you've surpassed? Yep. And now, of course, that's serious money, right?

That's the kind of money where you're talking a million here, a million there. Before you know it, it adds up to real money, right? And $100 million here and $100 million there, it's really real money.

Yeah. Big stuff. Is there a question there?

No, I think we just wanted to state the fact. As I told you, I think in the green room, look, our audience watching this, they're not the kind of people who O&R over ARR. But that being said, $400 million is a beyond respectable amount, showing some level of success, right?

Yeah. But more importantly, Garth, it allows you to bring certain functionality to the market, such as the Catchpoint acquisition that you spoke about and what they're doing. But you guys recently launched a new autonomous IT innovation program, and I think this is something our audience really would like to dive into.

So if you wouldn't mind, let's take a dive. Yeah. Certainly, the ARR is an interesting kind of metric.

The thing that I spend the most time on with my teams, so every Wednesday, it's kind of like, we call it church. We look at how do customers engage with the product? What does the telemetry tell us about that, et cetera?

I've got John and a bunch of other really good people that focus on revenue and everything else. But from a product perspective, what we're most interested in is how are customers actually engaging with and using the product? And so the telemetry is, in some cases, I would just think that's the most important thing that I focus on- Mm-hmm ...

as a product person. The fact that we've got X hundreds of millions of dollars, that's useful from an investment perspective, but ultimately, I'm interested in understanding how customers are engaging with and using the product. And so that- Spoken like a true CPO, right?

That's what you should be engaged in. Well, that's what we focus on. Every Wednesday- Yep ...

at 9:00, every Wednesday, we go through and we rotate through how are customers using the product from a cloud and a core cloud AI, the Edwin AI perspective, et cetera. We just rotate through this six-week cycle where we just really drill down deeply into the telemetry and what we're learning about how customers are engaging with the product itself. And for me, that's the most important thing.

How many hundreds or thousands of customers do we have that are using this particular feature, et cetera? The ARR will follow- Sure ... naturally.

If customers love your product, they're using your product, they're engaged in it, how many monthly active users on any given feature, we've got Pendo all over the place, we've got all these wonderful telemetry signals that tell us how customers are actually using the product to solve their problems, and that's the most important thing that I think we're focused on. So Garth, talk to me, though, about this IT innovation program you guys recently launched. What is that about?

So we have come to the conclusion, and this is not rocket science, but we've decided that AI is going to be the future of our product, not just an optional feature. Mm-hmm. But how we want to engage with customers on how they want AI to be infused into the experience of how we use the technology, that's kind of this new program that we've announced.

We're looking for 15 to 20 customers to help us really go deep in understanding how the future of AIOps should work. I'll give you a couple of examples. I think I'm super curious for customers, just collecting telemetry, we do a great job of that.

We've got 3,000 out-of-the-box integrations. We can take- Mm ... telemetry from every different source of your on-premise, your cloud.

We support monitoring applications that run in Azure and AWS and GCP and OCI, et cetera, and we bring all of that data in. Now, just collecting the telemetry data is interesting, but it doesn't really help you ensure that you're building applications and you're ensuring resilience for what customers really need to do. And so that's kind of this idea, this new thing that we've announced is sort of like, how do we think about how can we apply AI to helping you run AI operations in a much more efficient way?

We bought this company called Dexta back in 2021. We can collect all the telemetry data that you have across your estate, and we can throw alerts when things go out of normal operating boundaries. But throwing out 1,800 alerts per day doesn't help you really understand the health of your systems.

And so with Dexta and Edwin AI capabilities. We're able to find all of this noise in the system and really do root cause analysis and do the alert correlation that takes 1,800 alerts per day, and we turn it into 18 that you really need to really think about. And I think when we think about applying AI to all of the stuff that we do to help customers understand the health of their overall IT estate, AI is really going to be the technology that we're using to help you understand the signal from the noise, if that makes sense.

Absolutely. Right. So we're collecting all of this noise, all of this telemetry data, but how do you understand what it means to have a healthy system?

And this is part of why we got super excited about the Catchpoint acquisition, was Catchpoint does this wonderful thing where there's all this noise that exists in the world, Catchpoint has this beautiful system where they can run synthetic tests in 3,100 plus different points of presence around the world. And so in November of last year, if you remember, AWS had a little bit of an outage with their DNS service in US- A little bit. A little bit of an outage.

Yeah, a little bit But if you looked at what happened with the outage that AWS had, there were zero Catchpoint customers on the list of people that were impacted by that. " But all the Catchpoint customers had 20 to 30 minutes heads up, and they were able to figure out how to reroute traffic in a way that transient performance issues never became an incident for the Catchpoint customers. And so for us, it's bringing in not just the real-time telemetry data that we can bring in on everything that you're running in your own data center or in any third-party cloud, but being able to understand the weather patterns that are cascading around the internet, and then bring that into a real-time performance dashboard that allows you to understand, like here's how things are happening.

Sorry, my dogs are barking over here. But we can actually help you understand how these patterns are evolving, and then give you a heads-up so that you can prevent these transient performance issues that happen on the internet every day, every hour, every minute. But get ahead of them and prevent them from becoming major incidents that impact your customers or your core business.

I've been around this business a long time as you have, right? It's interesting how many people our age didn't have computer science backgrounds but got into tech and have done pretty good for themselves. First, the bottlenecks have moved.

First, it was we can't collect enough data, right? And then you look at what LogicMonitor does now with all of these integrations. You look at Splunk and some of the tools.

Probably, I'm going to say 10 years ago, maybe, that stopped becoming a problem. The problem was we collect so much data, how do we economically store it, analyze it? How do we abstract signal from noise?

Yeah. Then AI came about, and we said, what a great thing, even before it's a generative or agentic AI, just machine learning and pattern-matching AI. What a great thing to help us abstract signal from noise.

Now with generative AI and now agentic AI, we have an increased capability of getting our arms wrapped around huge data sets and finding actionable intelligence, things that really we weren't able. This stuff you're talking about from Catchpoint is amazing. With today's AI, we have abilities to do things with this data that we just never had before.

Yeah. Not just in IT, in medical, biopharm, everything. Yeah.

But at the end of the day, the mission hasn't changed, has it? It's still about abstracting and taking that signal out, separating it from the noise, and then acting on it. Yeah.

A few months ago, I was trying to explain this to a lot of our customer-facing teams. We're collecting data, two trillion metrics per day. Crazy.

So hold on. Yeah. If you were to go to Claude or Google or ChatGPT, name your favorite thing.

What happens two trillion times per day? Alan, what do you think happens 2 trillion times a- 2 trillion times a day. First of all, I think it's hard for a human brain to wrap their heads around 2 trillion times.

2 trillion. That's a big number. So, the funny thing about it is, in your noggin, Alan, your brain will fire neurons about 2 trillion times a day.

Really? Yep. I didn't know that.

That's a great- So I was trying to explain this to our own people. We're collecting this crazy amount of data. So, literally, the neurons in your noggin are going to fire about 2 trillion times a day.

19 billion transactions, credit card transactions per day. 219, yeah, 219 million. Million or billion?

19 billion? Yeah. 19 billion.

219 billion transactions per day of credit cards. So your brain's firing all of this data. We are collecting 10 times more data in our systems today than there are total credit card transactions- Or transactions in the world ...

that'll happen on June 11th, period, at this point. This order of magnitude. Yeah.

So, we're already collecting that. That's a crazy amount of data. How do you make sense of that?

And then what's the signal-to-noise? And for me, the other interesting thing, and I will credit Mary Meeker and a bunch of others in the industry that have gone after it, the available supply of raw computational capacity that's available for AI, the cost per token or per millions tokens, there's a lot of different ways of looking at it. We've actually reached a point where I believe we can actually apply in a cost-efficient way, how do you make sense out of 2 trillion metrics per day, but focus the humans that are running these systems in a very efficient way on just the signals that make the most sense?

And so for us, there's the expanding... We've got a great partnership with OpenAI. We've got a great partnership with AWS and Azure and all these other third-party companies.

And so for us, it's sort of like we now believe we have the ability to figure out that signal-to-noise ratio in a way that is much more compelling. We can find the needle in the haystack in a way that is incredibly more efficient than what we could do five or 10 years ago. Oh, no doubt.

Yeah. I think what would be real interesting, though, is if you could, and if you can't tell us, don't tell us, but can you tell me what the token cost of that is? How much does finding that needle cost me in tokens?

Because is it worth it? Well, the lovely thing about... Sorry, this is just me being a geek.

We are constantly working on making that more efficient. There are things that I can do today that I couldn't do a year ago. But I will tell you, in terms of LLM and the foundational models that we're using, things are 10 times more efficient than they were a year ago.

Wow. For us to be able to do basic inference. Now, in, I would say two to three years ago, we would be relying on LLMs.

Humans would come in, and they would type, and they said, "How do I want to interrogate my data? " Or blah blah blah. So we were very token inefficient two years ago.

Now we're becoming a lot more efficient, and then the foundational models are also becoming a lot more efficient and cost-effective. And so, our vision over time is, I don't think I need to use LLMs necessarily for people to interrogate the data directly. Let me just say that.

You'll still be able to, all the data that we collect on your behalf, you'll have access to that. But my vision is, I don't think human operators need to be concerned with that. " Right.

And so I can start to proactively give that to the operators that are responsible for the health of their overall IT systems. And I can do that in a way that's like Crazy cost efficient, and doesn't require that they need to understand how many... The fact that they- Right.

They don't need all the noise. No. That's the signal they're looking for.

I'm collecting two trillion metrics per day. I get it, Garth. Tens of millions of devices around the world, running in all these different environments.

And my goal isn't to just pass that noise through. It's to really get to, here's the key signals that you need to be aware of. And that's where the Catchpoint thing was so exciting for us is, if you have an issue with customers interacting with your e-commerce experience in Croatia, on the Chrome browser device, et cetera, I can tell you why that's an issue, and how that's impacting your business ultimately.

And then I can give you all the traces and the data that you need to help you understand how do we fix that. But it's sort of like, how do you detect that signal from the noise and do that in an efficient way? And I think I'm very excited about what we can do in the next five years.

I am, too. Yeah. I can't look out beyond that.

But Garth, we're about out of time. I got to wrap up. So I enjoyed, you were talking, you said geeking out.

We're your people. Geek, right? We geek with us.

But I got to end this. We got to go to another one. com?

Yep. That's where we'll send them. All right, Alan.

Very much- Garth, I want to thank you for coming on. All right. " We'll be back in just a minute.