PurePerformance - Beyond the Hype: Open Source, Observability, and Finding Your AI Breakthrough

Episode Date: June 8, 2026

Its rare - but it happens: A guest-free episode of PurePerformance, allowing Andi Grabner and Brian Wilson reconnect to share real-world insights from recent months in the cloud-native and observability space. From KubeCon Amsterdam experiences and the strength of open-source collaboration to emerging challenges like AI-generated contributions, they explore how the industry is evolving beyond the hype.Your co-hosts of PurePerformance discuss the changing role of observability in the AI-native era—both as a foundation for understanding complex systems and as a tool to monitor AI itself. Brian shares his personal shift from AI skepticism to practical adoption, highlighting how AI can significantly improve productivity when used thoughtfully.Hope you all enjoy this episode!

It's time for pure performance. Get your stopwatches ready. It's time for Pure Performance with Andy Grabner and Brian Wilson. Hello everybody and welcome to another episode of Pure Performance. My name is Brian Wilson. And as always I have with me, my co-host, Andy Grabner, Andy Grabner, how are you doing today? I'm doing fine. It feels like today I'm pulling out very useful information out of your nose.

Yes. And, you know, Andy, if you look at the... picture for this one. Andy's picking my nose and we're going to start off by picking Andy's brain. So I'm going to have to get my, you know, did you ever learn Andy in like great school?

Did you guys learn about the Egyptians and all that? Did you have lessons on that? Builder the pyramids and mumifying do they do that in Austria? Yeah, I believe the basics definitely. It's been a little bit too long because believe it or not, I didn't just came out of high school. Yeah.

But do you remember the whole idea of like they'd have to pull the brain out through the nose with a hook. I think so. Yeah, I do. Okay. Just one of those things that they make sure to teach kids like, yeah,

and they had to pull the brain out of their nose with a hook. It's like, okay. Okay, now let's go home and try it as ourselves. Yeah. You know, kids love that part, but it's like, is that the most relevant thing? Anyway, just with you picking the nose and me picking your brain

made me just remember something that haven't thought of in decades. Yeah. Well, I'm glad. Speaking of decades, I was going to do a nice transition here, right? Ready? Check this out.

I'm going to speak in a decade. It feels like decades since I've been on a podcast with you. Yeah, it feels like you've lost grip of your own podcast, right? Because I've been kind of running the last. Yes. True. But see how this already works? I already talk about my podcast and not long about our podcast.

So we need to change this. We need to get you back on the show. And I do hope I know for the last two times it was either. a bad timing on my end or scheduled conflicts with guests. So sorry for that. It's all right. The show must go on, as they say. Exactly.

Like Freddy used to sing, show must go on. Oh, yes. Hey, what do you want to ask me? Well, you know, so much I want to ask you, Andy. But keeping it relevant to the show, you know, you've, as our listeners may be aware, I sit here in my basement dungeon all the time while you're out.

The word of the day is gallivanting. So you're out there gallivanting around, meaning people, talking to people, either you're going to conferences or you're doing stuff with your book. And, you know, one of the trends we've had on the show lately is a lot of topics on AI, right? I think they're still very, very interesting, but I thought it might be good to go back to a little non-AI and ask you, what have you been seeing? what have you been encountering out there in the wild while you've been gallivanting? That's not AI-related. What's exciting you in those realms? Because I think what's excited us over all these years has been all the innovations

and performance things we've seen and done in the past. AI is obviously overshadowing everything. So what can we be excited about or be interested in outside of that? What are some of the real-world... experiences that little Andy Grabner has had. Yeah. I think the best thing that I experienced again and again, but just a couple of weeks ago at KubeCon in Amsterdam,

is when our global community comes together. And so KubeCon Amsterdam, right? It happens several times here, but in Amsterdam was the European version. And the best moment I had, I was volunteering as a CNCF ambassador to support the Contrip Fest. So the Contrip Fest is basically a special session format

where different CNCF projects can say they want to do a contribution fest. That means they get a room, they get a stage, they get tables and everything. And then the idea is that people come in and help contribute to that project. We did this years ago.

I think it was in Paris or somewhere. I don't remember where. Which one? For Captain? where we basically taught people, hey, you know, this is how what Captain is, and this is how you can extend it. And then we try to help them building their own extension to Capno, using it, setting it up. Now, this time I went to the Argo CD Contrip Fest, because Argo CD, right,

it's a project that is used by many in our industry. And I thought it would be really cool to see what people are contributing. It turned out this room was completely extremely full with a lot of people came and come in. And instead of an active contribution to contribute code, people were contributing to stories, meaning the idea was that the maintainers, they said, hey, we want to hear from you on, hey, how you're using Argo right now. What are some of the things that you're doing or that you like, you dislike? And it turned into a basically an hour and a half session of people that use Argo and basically

present on lessons learned and best practices. And this was really cool. And I think this is a great way where a global community comes together and just shares and independent of who you are, which company you work for, whether you're presenting something internal to your competition, right? Because this also is sometimes a little bit hindering people to say, hey, I don't want to show how we use these things internally because maybe our competitors sits in there and they learn from it.

But what I like about this community is that it goes beyond the corporate business. founders right? There's no there's nothing holding us back and I think this

was really cool and then the other thing that happened I had a couple of things on that I wanted to

dive into or at least bring up I think the first really interesting thing there is you know

most of the time when I think of open source projects it's people in all remote places

working on them and only communicating through the repo right and I think what

fascinating idea for everybody to be in the same room, right? Because it's not like most, it's not like any of these open source projects are going to be having a conference because no one really owns it, right? So it's something like that community users. But I also want that, you know, one of the things when we were talking to several people over the past episodes about some open source projects was, you know, find a, find a project and contribute to it. Maybe it's reviewing a documentation. Maybe it's writing some documentation. Maybe something simple doesn't have to be code. did you see any of that going on in those things?

Or was it more like obviously with the stories, when it was stories, but was it all people focusing on code or were people looking for like, hey, can I, you know, help stack the chairs like after a party or just like the simple things like some of the maintenance stuff.

Yeah, no, I think, I mean, for QConn, it is how can you contribute to a project with, you know, code contributions, documentation, tutorials was a big topic. Also now more and more, I didn't want to talk about AI, but obviously AI is a big topic.

How can the documentation, how can the tutorials be updated and crafted in a way that our coding agents that people are using more and more, that they pick up those open source projects and know how to correctly use it, configure it and integrate it. I think that's a big topic. That's a new aspect. Interesting. Yeah. Yeah. And the other big challenge right now, maybe you've seen it, is there's obviously a lot of new contributions, code contributions through pull requests. Many of them are unfortunately all completely automated through AI and not necessarily have to write quality. So currently there's a lot of effort going on for the maintainers to sift through the noise that has been generated by bots, by AI agents. So that's a challenge that we need to solve as an industry.

But yeah, these are definitely topics. It goes beyond just code because not everybody is a coder, right? everybody that is using a tool you know you may know you may know how to configure it but you don't know how to how to extend it I think we also had

Diana from Romania on the show recently right she was recognized as a as a contributor to open source projects in the last KubeCon so yeah yeah it's interesting that you mentioned

the so many topics around this way But you mentioned the use of AI to do contributions, and I can really see that being a problem. You know, it's not like using AI to help you write your code and all is a bad idea. Obviously, you want to go through and clean it up, make sure it's good, make sure it fits the standards. But I see, you know, a great place for abuse in that would be I want to get notches on my belt to say I've contributed to projects. So I'm just going to spit this out and load it up to pad my resume, which then causes the problem for the whole community. so it's, you know, people, yeah, that's quite a conundrum. Anyway, we move on to that.

So if you're doing that, people please don't do that because these are real people who have to do all. But unfortunately, this is a real problem, right? We have fake accounts with fake, not fair, yeah, artificial accounts with artificial contributions, making the life of the maintainers very challenging right now. I wanted to have one more comment on the breaking down corporate walls because I also had the chance, and I think we posted it on LinkedIn. We released the book recently, the observability

in the AI native age, and I was able to give this to one of my fellow DefRELs from one of our friends from another observability vendor, right? And I think that's also, and then we share, because in the end, we have

the same kind of mission. We want to make sure that observability is a topic that everybody's aware of, and as death rel else we want to make sure that, yeah, we work together and not competing. Did you run into anybody who read your book and say, this changed my life? Did I run into any? Well, the book was just released in the week of KubeCon.

Oh, is this the second book? A second book, yeah. Oh, I didn't know you had a second book out. We'll have to have you on as a guest. well we thanks that you mentioned this but the next

podcast recording which unfortunately will be again without you will be with Hillary and Rob the authors of our

of our latest book here what is the topic of the newest book let's go into that briefly here we'll go deeper back

the topic of the yeah the topic of the book is observability in the AI native era and how observability has changed over the years.

The power of AI and how we can harness it on the one side, when we're analyzing and making sense out of observability data, but also how we can use observability to analyze AI and optimize AI. So we cover really the journey where we started years ago with monitoring, how we moved into observability, and now how AI is again changing everything, but with observability, a very important pillar we have like with the previous book a fixtures company that we kind

of so hillary and i were writing more the uh the ideas and what we've seen from our experience in the field Hillary more on the core AI and LLM technology and also from a security perspective because that's her background i from an observability perspective and performance perspective and then we have Rob who is then bringing to life all of these ideas as he's kind of uh following the journey of a company, of an traditional enterprise company that is also kind of evolving with observability and with AI.

It's interesting that AI, well, I mean, that sounds awesome. Congrats, by the way. Is your second book ever or I know it's your second recent, but is it your... It's technically my third book. Because years ago, if you remember, in the early days of Dinah Trace,

we wrote, or at least I contributed a chapter two to the Java performance book. Oh, yes, yes, yes, yes. Yes. But this is now the second book recently, yeah. Well, congrats on that. The interesting thing I think with AI and in technology is that right away became two topics, AI observability and AI operations.

I mean, not AI. AI observability and, yeah, AI operations, right? So as you mentioned, how do you use AI in your analysis of things? But then you also have to observe the AI. And then I guess you can use AI to observe how you're using AI and start feeding, snake eating and stuff. That's just kind of interesting because I don't think we've had much of a new development. It's not like when the cloud came along,

there was as much of a focus on, oh, we're going to. going to use tools in the cloud to monitor the cloud. That wasn't as, you know, like the idea of like AI exists. We have to monitor it. We're also going to use AI at the same time to monitor what we're monitoring, right? Yeah. Just kind of interesting how that came about. It feels like a new, new area with that.

But anyway. But yeah, on that topic, so Reese Lee, she is basically a counterpart of my role at New Relic. And she was the one I gave it the copy to. So thank you so much for being such a great member of the Defville community. There's obviously many others out there from all of the different vendors, whether it's a new relic, data dog, dash zero, I'm sure I'm Grafana, they're all people in our space, and we all have the same mission that we are educating people on the best way to leverage observability and the possibility has to grow up because the way we built software has

changed over the last decade's. And also the way we need to think about observability has to change. And this is also a big part of the book, how observability is evolving, not only in the cloud native space, but especially in the AI native era. Awesome.

With regard to that, you had mentioned there's been a lot of traditional ways we look at observability. A lot of normal things we try to track. as well. Based on what you're seeing at the different conferences, you're going to, is it still

kind of all the same just getting more complicated? Or have there been any new types of things come along where you're like, oh, I haven't heard about that yet, right, in terms of maybe from an observability or types of problems, as we've seen even in AI, right? It's always the same. we always keep coming back to the same types of problems over and over again. I know I'm putting you on the spot because we didn't prep on this, but I'm curious, if you recall anything, we're like, oh, that's kind of new. I don't think there's new things.

I think what you mentioned during the beginning is, and maybe this is a self-fulfilling prophecy, that everything gets more complicated. And maybe sometimes we think that we, in the industry, we make things more complicated so that we have a better excuse to also build solutions. solutions that attain the complexity. What I do see, however, and this is why I wanted to do a shout out to both Henrik at Ina Trace, but also from dash zero, the defral. And I'm so stupid why this name currently, why I'm blanking on his name. Now I'm sure I'll figure it out in a second. He did a people in my network.

Let me see first people in my network. and it is not Fabian, not Mirko. Am I so stupid now? Well, while you're looking that up, I'll give a big shout out to Sean Peterkin and Das Zero. Hey, Sean. I'm sure you're not listening, but... He did, so both Henrik and him, they did some good work on the analyzing different open source projects in the way

on how well they are observable and how well they're using, they're following open observability standards, how well they are supporting open telemetry, for instance, whether there's good documentation. I think that's Caspar. Sorry, for whatever reason, it's been a long day. Sorry, Caspar, if you listen to this, it took me so long to remember your name.

Caspar Burg Nissen is also a death rail at dash zero. He did a great talk there. Henry did a great talk on, basically analyzing how well certain tools and projects are adhering to open source, open observability standards, what are the key metrics, the key things to look into. So that was really good. And Henrik did a great job in analyzing from an AI perspective to different AI frameworks. So that was exciting and interesting.

Great. Awesome. So Brian, I think I've talked enough. But talking about AI, how does Brian Wilson use AI in your day-to-day life? Well, you know, it's kind of interesting. I'm really surprised you ask that question, Andy. You know, for a long time, I've been an AI resistor, right? I've had moral quandaries about it, right?

But I also, any time something is hyped too much, just like get this away from me, it's all hype and stupid, right? And you've heard me mention it on the shows in the past where it's like, oh, let's generate a picture, right? Or other things. But I think over the past months, as we've been talking to, you know, people in the industry who are using AI in very practical ways as opposed to novel. Like a lot of the AI is novelty that people are using for. I want to make a picture of my grandmother holding my baby, but my grandmother's dead. Or whatever.

I don't know why I went to that direction, but that's what popped in my head, right? Or I want to have it make this song for somebody I like, whatever, right? And it's just novelty, you know, and obviously we know there's a ton of power, compute, resource, and all that goes into it. So it really turned me off a lot on that. But then as we're talking to these guests and finding uses for just doing much more efficient work, right?

Taking a step back, we go back to, I believe it might have been Wilson Marr, you a while ago put together a blog of like 10 things I want to do this year in performance, right, whatever it was. And one of them was like, whatever I do that's repetitive and redundant, automate it. Like find something to automate it.

And it's, so then you have to do go ahead and write some automation code and do it. And it's like, well, that kind of stuff is a great use case for automation. I mean, yes, it still uses all the power consumption and everything, but it's not like we can deny or stop AI from going, right? So as we, so I personally am just trying to come to grips with using it, but also using it in ways that are going to be what I deem more worthy, right? That's my own judgment on it, right?

Not for these novelty things. And recently with work, it's gotten like super busy and also based upon things I've been hearing from these people that are our guests and suggestions I've been getting from my colleague. I've been slowly dabbling into using it and then suddenly the last two weeks I've just been tossing everything into it, right? Because it suddenly was like, oh my gosh, you know, number one at my job right now, I'm not going to survive if I'm not using it. So one great example, right, was, and this is more practical. is not necessarily performance related, but performance in terms of my own personal performance. I had to do this meeting and presentation to a,

to, you know, a dinosaurase demonstration to a company. And I was taking it over from somebody else, and I had notes from a bunch of different sources. Right. And at first I was thinking, okay, I have to go through all these notes, try to figure out who I'm going to be talking to, what they're all interested in, what technologies they're using,

figure out what kind of a demo flow, but I'm going to have to go through five or six different emails, try to pull the bits and pieces, organize it. I'm like, this is going to be a pain in the butt, right? I'm like, and then suddenly the light went on. I'm like, let me toss this in a co-pilot. Give it, and what I've learned from it, give it some very, very specific requests.

You know, give me a list of all the people on the call and what the notes say their interest was. Give me a list of all the technologies that are used and make sure we cover them. Give me a list of the problems they've identified, both functional problems and how that relates to business. finding that, you know, the more specific you are with your request, the better you'll get. And I popped that in.

I did the researcher mode in co-pilot, because I wanted it to be really good. I worked on something else for a few minutes, and it spit stuff back out. And I was like, wow, this just saved me so much time, right? And so it's been a lot of things like that, or even a customer wants a bunch of they want instructions on how to do certain things, right? Yeah, I can just send them a link to the documentation, but you send someone a link to the documentation.

Number one, it's a little bit lazy in terms of customer service. It's, you know, so-so. So, again, I used it to go ahead and say, give me a step-by-step guide based on Dinah Trace instruction with links to the instructions in bullet point and checkmark, whatever it is, spits it out for me again.

That's something that might have taken me an hour to put together. Five minutes. It's out there, right? So it's really about making my job efficient in those ways. And also just my own personal stress, right? When you have too much work going on, right? And this could be the same for a developer. You have a huge backlog of stuff you have to get done no matter what the work is. It could really help you getting at least 80% of the way there where you just have to like go through it with a fine tooth comb, make sure it's all good again. You know, even go back to your open, source contributions. Not like there's anything wrong with having AI help you write the code, but don't just copy and paste it, submit it, go through it, look at it, make sure it's efficient, do the checks and balances. So it's been quite a bit of an eye-opener for me, and I definitely see the trend.

We always hear the idea of, you know, do more with less. And hopefully, companies will give their employees time to figure out how to use AI and this stuff before they say, oh, you're going to use AI and do more with less, right? because we know we can make it more efficient on that side. The other really cool side of it, though, has been within tooling, right? So there's two halves of it. How do you interact with your tool sets and how do you do things more efficiently at work? Right?

Another quick efficiency one we had on the podcast several episodes ago. There was the discussion with chaos engineering, right, and how they feed all the data into the AI, and the AI is coming up with chaos tests for them, like different ideas, right? So instead of everyone, you know, so that's more about like doing things more efficiently. But, and again, this is not a plug for Dinah Trace, right? This is, you should be able to do this with anything that has either built-in agents or connected MCP. But using agents within the tool has been really wonderful and impressive.

So you can just go in natural language. You know, I've been away from the tool for a while being a manager and now I'm getting back into the tool. So I'm getting deeper into a lot of the new crazy things that we could do in Dinah Trace. And part of it is like, well, tell me, give me this information. And it spits it out for me. Then I'm like, well, now give me the query that I would have needed to do that on my own. And it's going to give me the query. Right.

So I think the future of interacting with tools just with natural language is really, really fascinating. And we're seeing a lot of different tools starting on that journey, right? It's still kind of in the early phases. but it's, as we've seen with a lot of AI, the early phases are moving very fast with AI stuff, right? As opposed to other projects that would take years to crawl along. This is like month to month, there's improvements, and month to month there's more people expecting it there.

We've talked to customers who ask us about our agentic approach of the tool. Like, they're expecting it already, right? So for anyone else, for any tool you're using. But then it was also efficiencies, like with MCP, right? most of our guests in the past have been talking about using it for code, right? And in our case, it was putting a prompt in to do some research on a company we're going to talk to, and it's generating a dashboard for us, and it's even generating a JavaScript to dummy example, to not dummy, but to generate example data in this dashboard.

That's based on nothing set up, right? but it's building these proofs of concepts. And it's literally just sending in a prompt to do this. So again, I know I sound like a nob doing this. People have been using this prior like, yeah, duh, where have you been? But again, I've been one of these resistors. And, you know, you have a sign behind you that says, join us, right? I feel like I'm joining the cult.

Becoming Borg. But there are just so many ways it can be used for efficiency and really taking the drudgery. out of that. Of course, people would probably have fun learning how to write that code, right? Anybody who's done any developing, you know, it's fun writing some code and running it and seeing it work. But then after 20 times doing that, after a thousand times writing code and debugging, it's like the novelty's gone. You know, and I think it just opens up so much more that we can do outside of that and take that drudgery work away.

but still double check it. So, yeah, I've been diving into it, and I really come to understand that if you're not using it in that way, not from the concept of you'll be fired if you don't. And I don't mean that from Dinah Trace. A lot of companies are like, you've got to use AI, you've got to use AI. But using it now, I'm like, oh, yeah, I'm starting to see why you got it because it's just complexity of everything is getting crazy.

And it just takes it away and your stress level goes down. Now, hopefully, soon enough, we'll figure out ways to do it more efficient, or we'll need some sort of technological breakthroughs. But, yeah, it's been a very interesting last two weeks, two and a half, three weeks, really. But I think you said you feel like you are kind of like a late adopter. I don't think so. I think there's still a lot of people that have not yet played around and have not gone to this breakthrough moment.

I think this breakthrough moment is the critical thing that also happened to me, when you all of a sudden realize how this can really help you day in and day out, even if it's just a couple of minutes here, a couple of minutes there where it saves you, but it saves you in the end, time that you can either spend on something else

or go home earlier, whatever, whatever it is. Yeah, and it's interesting too because I was explaining to a friend about it because one of my friends uses Siri a lot, talks into his phone, or we like to call AI Al, so he'll ask Al about something

and he was asking me and another friend who's much more advanced on A than me about how do I use this better and my explanation was it's really just programming right you're giving a computer a set of instructions right like I firmly believe

there's no intelligence it's just prediction right but what I found in myself and again this is not groundbreaking is that the more specific you are with your input the better you're going to get something back. If I were to ask a developer, create a web page to sell books, right? I know other instructions. I have no idea what I'm going to get back. It's going to be up to the developer. But if I say, give me a web page, I want recommendations on the left side with yellow background,

blah, blah, blah, you know, giving as many specifics. That was a quick early lesson is that the more specific you are in your instruction, the much better of an output you're going to get. And you have to realize it's not like I'm asking a human being to do all this insane work. for me. If it was a person, I might be, well, I don't want to bother you with it. It's like, yeah, give it. And it's only going to give you as good out as what you put in. So the more specific you can be, that was a big lesson for me, is that it just increases

the quality of what you get out from it. So, yeah. Yeah. Cool. Yeah. So maybe a shout-out to everyone that is listening that has not yet experienced that breakthrough moment. Do it like Brian.

I also want to shout out to all of our guests who've been coming and talking about it, right? Because it's big inspiration from our guests and also big inspiration from my colleague Greg Speckhardt, who's been a lot earlier of an adopter on using it in his work. In fact, when he was telling me about it, he was like, I would have never gotten through Q4 because he was so busy. He was like, I would have never survived Q4 if I hadn't quickly gone into this. Because again, he just used it to take away all that work. And then he started giving me all these tips. and between that and the stories

we're hearing from our guests, I'm like, you know, and it was funny. It was like the trickle effect. It was like, I used it maybe once or twice one week, the next week three or four times, and then last week and this week, it's just been opened on my desktop. So it's like, bam, babe, we're all in, you know, like just.

But yeah, check it out, explore it. Ask your friends, you know. I kind of felt dumb asking because I figure, oh, everyone's supposed to know how to use this just because we see it everywhere. But the truth is, most people don't. So just like the regular IT community, everyone's happy to help you. Everyone's happy to share what they're doing, what's working.

Ask your friends, ask your neighbors, ask your colleagues, what's working, what's not working. How are you using it? Because everyone's got different ideas and different novel approaches that you might be like, oh, that's exactly what I'm looking for. I wouldn't even thought of using it for that. But yeah, it's there and take advantage of it. even just for your own sanity and health. It's definitely here to stay.

It's another tool. And some of us we need to adapt our way of working, but in the end, I think there's, I don't have to fear that it will take away our jobs, but all jobs will potentially, most likely change, obviously. It'll become more interesting, I'd say. If you're not spending that time coding, you can be designing, you know?

Exactly. Yeah. Yeah. All right. Good stuff. Well, as I think our guests can tell, we didn't have a guest today. Not our guests, our listeners can tell. We didn't have a guest today.

But we wanted to catch up since I've been absent for a while. And really hoping. Do we call this a ketchup episode? Chep episode. Yeah, we can call it a ketchup episode. Yeah. We can call it. It's kind of like Alexa and Siri talking to us.

or what's the Google one? I don't know if you remember the old videos where they took the two different bots and had them talking to each other and it was fascinating, but we're real human beings. Talking about human beings, I need to cut this short now because I'm here in Vienna. And I'm going to meet with a human being. One of our partners also, quick shout out to Roman Feastel from Triscon,

our friends from Triscone. We also had him on a podcast once. He's the one who studied, astrophysics, if you remember that. I'm not sure if I completely recall it correctly, but he was into astro, and then he's now doing low testing and performance testing, and I'll catch up with him now. I'll tell him, I say hi.

And thank you to all of our listeners. I hope you get something from this. I know I did, even just talking to you. As always. Anyway, Andy, thank you so much. Thank you to everybody for listening. be back with our regular programming in the next episode. And take care, everyone. Bye-bye. Hi.