AI-Driven Observability and the Future of Self-Healing Systems with Alois Reitbauer – KubeCon Europe 2025

This is Textron tv. Hey everyone, it's Alan Shimmel. We're back at CubeCon.

I'm really happy, you know, we've done remote before, but never in person. Once In person, where do we CubeCon? Paris.

Paris or Chicago? One of the two. Okay.

Well it was long enough ago where I'm entitled to not remember it. Yeah, That's fine. Yeah, you talk to a lot of people, it's totally fine.

My friend Rebar Alice is chief Technology strategist. Yes, exactly. At Dynatrace, and believe it or not, my friend Alo is here, has been at Dynatrace 18 years.

The company's 20 years old. 20 years, 20 year anniversary. You've been there for 18, you're legend.

Yeah, Because I started at seven years old, obviously, because I am 25. Right? Yeah.

Something like that. But, uh, seriously, give people a little bit of an idea of over the 18 years Yes. Different roles you've played.

Yeah. The different roles that I played in the company, it was always very customer focused, but obviously it was employee number 30 or something like that. Very small company startup, um, world.

And I, I try to relate it to what observability, a PM, whatever we used to call it day was like back then the biggest challenge was people were not sure, do we actually need observability? Um, do we need monitoring? A lot of it was, we have a problem, we need a solution right now.

It was more like this bandaid character. Biggest challenge back then was getting all the data, collecting it in real time without bringing systems down. So that's where I started in the company doing pretty much everything.

Then I moved more into what you would call today, derel solution engineering. This was when we transitioned to, okay, we need to provide more out of the box analytics. Like we used to have pure press, or this would be the phrases, which was the more the founding idea behind Dynatrace.

That's why the company dynamic tracing. Right. Got it.

Um, Then when we started to build very special visualizations, like we built like one of the first ever color visualizations of the trace, like what are the most important parts of the data, like finding the needle in the haystack. Um, so that, that's the next step. But then we moved more and more into analytics.

That's where I enter product management working. Our second generation product where we first introduced AI 12 years ago, so that's how long it have been actually working in ai. It was predictive causal ai, right?

Automated root cause analysis. So this was the first shift we saw in the industry from trying to collect all the data. So, okay, now we have the data, now we need to make sense out of it.

Um, get to the root cause, analyze it. And most recently I've been working a lot around ai, so building more AI into Dynatrace, generative ai, how we can use it. Also monitoring AI applications.

I think now we are entering like the third stage of, uh, the observability space as we call it today. Where to your point now, it's about taking action. The tool no longer being passive, showing you the data.

No, Everything is actionable if it's not actionable. And that's behind this whole agenda, AI and everything else. And, and this for us actually has three pillars.

It is self-healing, so automatically helping to resolve or guiding your solving, uh, self-protecting, doing the same thing on the security side. If you use security posture management, you detect vulnerabilities, proposing how to resolve it, and the last pillar, self optimizing, continuously watching your system, optimizing it for performance forecast, but also for like business process flow type of, uh, topics. So that's the three big pillars that I see, and also where agent it fits in for us.

Excellent. I want to, um, I wanna focus in, look, people watching this, or we're gonna be doing an article out of this interview as well. They're not here.

They're not. I'm, yeah. I'm sorry for them.

I'm sorry for them too. It's a great conference, but being they're not here, oh, tell people what's, what's dynatrace's loose here? What are you guys seeing here?

What do you think comes out of this? So what, what we really start to see here, obviously the big topic here, as everybody can imagine, AI is ai, but there's also a lot of observability here. We just talked about it beforehand.

A lot of company who are in the, in the observability space. Yeah. Or it's a tangential to observability by providing storage.

So we really see this move across the entire industry, um, right now that on the one hand people start to adopt ai, think how you can use agent making exactly that change of mindset we had before, like using ai, like how can AI help me to do my job better? I remember the very first time I was having a talk on how AI can help you do your job. Over 10 years ago, I was almost rejected from that conference.

You can't tell people that AI is going to do their job for now. Now we're starting two years ago. It's actually the opposite.

How can AI start to do my job? So we see a lot, obviously, um, how AI plans into observability. Observability overall, I think has not reached its maturity state where everybody knows you.

They're gonna need it as, as I mentioned before, when I started in my career, there was a lot of evangelizing why you need to look into it. Is it really worth the money? I think people have come across this and I think the two also have blend together, observing AI workloads, managing AI workloads, especially as we move to ag agentic.

Yeah, I see ag agentic as the biggest paradigm shift. I think it's bigger than monolith to microservice. Monolith to microservice was huge, but now we talk about application that more or less assemble the way they work based on the LLM at one time.

Yeah. I, I don't disagree. Look, monolith to microservice is something you and I get excited about.

We're geeks, we're geek. Gee, that's big agentic AI is civilization changes. Yes.

Yeah. Right? And that, that, that's an order of magnitude different, right?

A agentic ai monolith to microservices changes how the average developer or ops person right, is doing things or the observability person, agent AI here that may be the new observability person or the right. And I, I think that's really, I think people gotta realize that now as we sit here today, you know, raise your hand if AI took away your job. It hasn't taken anyone's Job.

At least not from the 14,000 people who are here. No, Not these people. Yeah.

So, you know, so it's not, it's not necessarily taking your job, but how is companies like Dynatrace gonna make them more productive, going to allow them to do more? So the way I see it is look at what, what you're doing every day. There is the exciting part of your jobs.

There is a not so exciting part of the job quality toil. And that's the even more exciting part that you don't get to, right. And I think of how we have to shift is getting a less exciting part of the job done and getting it done more efficiently.

There's like, like roughly 80% of like operational tasks you have to do running software where AI can take a lot of the burden. This is not value creation for the company. This is keeping the lights on, like painting your fence every year.

Nobody's gonna get excited what a beautiful fence we have. But if you don't do it like your house will rod, your value will, will go down. Agreed.

And really focusing on what moves your company forward. Uh, sometimes when I have this conversation with customers, I bring up nobody ever got promoted for keeping the lights on, although people maybe should that that's, and I think as we get there, even a more competitive market right now, we are still in a very tense market situation. Overall.

Companies have to figure out how to best use the, the human capital they have in the company. Brilliant engineers they've hired, how can they create the most value to market? And especially right now as companies also have to adopt AI to enhance their own product experience, that build AI native applications.

You don't want to be stuck with the past by just running things that you could easily automate. And that's exactly the shift that we are helping our customers to be on. And we now have conversations we, we didn't have four years ago, zero incident policy, auto remediation of all standard incidents.

So even the C level now sorts of think that way the first time to think, Hey, we have to fundamentally change. We, we think about running and building software because there's like this big opportunity or threat depending on how, how you work on it ahead of us. And we have to get prepared, uh, to be ready for this.

Let me ask you a hard question. Uh, yeah. When, when will we see agent AI from Dynatrace and observability?

Uh, we, we do have some agent components already in the product right now, like our root cause analyst being the most prominent one. It's gold driven, it's autonomous, it figures out how to work. Um, I think we have to talk almost like about general NTVI and very tough specific ones.

And this is the route that we are going. So we invested in root cause analysis and what's we have been doing for 10 years now. We went in what we call preventive operations.

Basically taking those operational tasks out of people's, uh, lives that they need to do. You don't know exactly when to do it, how to do it exactly that. The pieces you couldn't automate was like professional efforts, tools, and then extending and guiding users in the right direction by proposing, okay, this is how you would have to modify something over here.

This is maybe how you need to tune the application over here. And we are right now also working to extending and opening the ecosystem because we won't do everything. There's no reason why we would build an agenda code editor.

So we see as integrating with other tools or we can't recommend, um, The best EC2 instance to run your workloads on. But we very well understand what the workload looks like. So you would interact with an Amazon Q.

So we have already bits and pieces of agent, um, but it's still kind of like in, in a very specific domain, target focused. We do it for remediation purposes, we do it for security purposes. And this is how we're readily building out the use case use cases.

Uh, but it's not like the catch all type of agent d ai. So we are very clearly building out, uh, the use cases and also extending the levels of freedom The AI has. AI works great if it's 3, 4, 5 steps that the has to do the thinking on, bring out sessions.

Once you do more, it starts to get hard. So the short answer would've been, we do part of it today and we will extend it on a use case by use case basis. And you can, for people who are, again, we're in the cucu, but wanna find out more about what's going on with this, where do they go?

Uh, what's going on with us? Yes, it's very easy. Go to our website, go to our YouTube channel.

com. It's dynatrace com Lot there though. Uh, we have a very big YouTube channel that covers lots of topics too.

You can reach us on social media, just engage with us. I really recommend the YouTube channel. There's a lot of content there.

Also on the observability, how to use the Ion Dynatrace, getting an idea of what's possible today. How what we're Working people like to do video And there, along with the videos, there's actually a lot of content out on GitHub, just getting, getting started to try it yourself. So some of those we talked about, about, It's great for this crowd too.

It's uh, like on preventive operations, you have all the materials. Like we share all of it out in the open. People just can try to play around with it, see how it fits in there.

And even if not becoming a Dynatrace customer, it gives you an idea of the art of the possible and what you can implement in your daily life. I love it. My friends good seeing you in person.

I remember that. I saw you here in London. We are here at Q Con, but to check out Dynatrace, there's a lot going on there.

It's a lot going on in observability here at Q Con. There's a lot of observability going on in Dynatrace and there's a reason for it. This, a shimel will be back in a moment.