DevOps and Developer Experience with Andrew Boyagi at Atlassian Team ’24 Europe

This is Textron tv. Hi, I'm Mitch Ashley here in Barcelona at Atlassian, team 24 Europe. You know, I, it is great when we have to have these rolling conversations where from one conference to the next or one time, we, we see each other at different events.

And, uh, Andrew Boyagi, who is, is someone who I've talked to before. It's good to be having you back again, talking about DevOps, developer experience, platform engineering. You're head of, uh, DevOps Evangelism, right?

Yes. With your last, yeah. Awesome.

Tell folks a little bit about what you do. Yeah, thanks Mitch. Uh, good to see you again.

Uh, I'm, as you mentioned, I'm the head of DevOps evangelism here at Atlassian. Uh, I have a multifaceted role, so a lot of my time I spend, uh, with a lot of our Fortune 500 customers, helping them with strategy and different techniques to improve developer experience and productivity. Uh, and of course, I work really closely with our product teams on, uh, shipping features that are gonna help our developers.

It's interesting that developer experience and is changing day by day it seems like. Um, yeah, the introduction of AI platform engineering, a lot of things. But some of the announcements with, uh, Atlassian around your AI agents, um, the auto dev with AI agents.

Talk a little bit about how the, how does AI realize itself beyond just copilots and ides? Yeah, it's interesting. Um, we did a, um, a survey of developers in February this year, uh, of two over 2000 engineering leaders and developers.

And one of the questions we asked them is about is AI helping them or not? And at that time, uh, two outta three developers were saying that AI is not helping them be more productive yet. Um, and it's been amazing over those last six months or so to see how quickly, uh, companies and, and people are innovating with ai, um, because it's very different at the moment.

But if we go back and we say, why wasn't, um, AI helping developers at that time, most of the tools that were available were code gentles, right? And if there's, if you've ever spoken to a developer like I know you do all the time, you've probably never heard one say, you know what? I'm spending so much time coding, I wish someone could help me.

I wish I had that problem. Right? Exactly.

Right. And so it's not a surprise to hear that, um, cogen probably isn't improving developer experience dramatically. Um, however, the 70% of the time the engineers are spending doing other things is where most of the friction comes in their role.

And I think ai, when applied to the right scenario, um, can really help re remove some of that friction and improve developer experience for developers. Uh, you mentioned some of the things that we announced at this, um, at this conference. And one great example is actually order review and, uh, pull requests, review time, when that's very long, that's a significant contributor to a poor developer experience.

And so, uh, we found a way to apply AI to, uh, pull requests and reviewing them, uh, to improve that and speed that process up, which, uh, solves a pain point for developers and in turn improves their experience and productivity. Uh, so I think AI is fantastic. It's a great tool to improve developer experience.

Um, but it needs to be applied to the points of friction that engineers are having. Right. And to non-developers, most people think it's all about writing code.

Actually, very little of it of our time is spent writing software itself. Yeah, I mean, the role of the developer has, uh, changed so much over the last, say 10, 15 years. Like it used to be very much about writing code.

Uh, but there's been a few transformations that have changed things, right? So just something like DevOps introduced, uh, the need for developers to get involved in operations to understand testing, uh, they now need to use a whole bunch of different tools as a part of their role. Uh, then DevSecOps means developers need to understand security.

Uh, even the shift to cloud now developers need to understand cloud infrastructure and how to spin it up and use it. And so, uh, the role of the developer continues to change and new points of friction keep, uh, getting introduced into their daily role. So developer experience is super important because that's the way to remove some of those friction points in the developer's role and help them get back to delivering high quality software at speed, which is what everyone wants.

Uh, I'm gonna send you a couple reports we just issued around DevOps, future DevOps and AI and, and software development. I want your feedback. Some things that came out of it were like AI helping us with configuration errors, right?

How much, how much time do we spend setting up environments, fixing things, blue code or integration that's not working anymore for some one change kind of broke something. There's so much about the environment, even with platform engineering developers spend either maintaining, creating, setting up or building a variation of something they need just to get their work done. Yeah, absolutely.

And I think like for any role, especially for developers though, you wanna remove some of the low value repetitive tasks that they're doing. And AI is one solution to that for sure. Like it's, uh, if you look at, uh, after leading engineers for 20 years, the number one complaint from them is they don't like writing documentation.

The number two complaint is they need documentation because it's blocking them. They dunno what to do. Yeah.

So I Don't remember Zero is meetings, but you know, that whole Nother issue. Yes, that's right. Um, but I mean, AI is a really good use case for documentation.

Uh, but you mentioned platform engineering and I think there are still a lot of problems that developers are having that AI is not the right solution for, uh, yet. Uh, and so platform engineering is a way to start to remove some of those friction points and scale that out across an organization so that, uh, engineers can spend more time doing the things that they like to do and the things that they add the most value doing as well. Mm-Hmm.

You know, we talk about knowledge sharing as a general term in the developer's world. It's what framework, what framework are we using? What version of it?

Where do I go? What are the APIs that we have designed that I don't have to recreate or I need to use? Um, it might be part of a third party tool or framework.

It might be something we create. There's a lot of information you need just to know how to do the work, how to, whether it's testing or development and that knowledge sharing. It seems like some of the things around the agent capabilities, the enhanced search that you're doing now through your agent, how does that help with those kinds of issues?

Yeah, I mean, uh, there's some lots of stats around how much time developers actually spend looking for information. Uh, I think the stat was, uh, 70% of developers spend eight hours, uh, at least eight hours a week looking For Yeah, yeah. Looking for information.

Now, I, for non-engineering people, I like to, I give them this analogy, uh, imagine every time you have to leave the house, it was mandatory for you to look for your keys for one hour, right? Imagine how frustrating that is. Imagine you get everywhere late, um, and then your boss is ringing you saying, where are you?

Or the kids are late for school. Um, and that's what it's like just spending your time or wasting time looking for information. Uh, and so I, things like AI search that you mentioned, even, uh, something that we use internally is our, um, our developer experience platform Compass.

That really answers a lot of questions for us in terms of, uh, who owns this piece of software? What does it do? What events have happened lately?

Uh, where's the documentation for this? Which team owns it? Uh, and it really removes a lot of that friction in terms of looking for information.

Uh, so Compass is great for that. But then of course we have AI search and VO that help us, uh, find that documentation fast as well. I'm sure I was really impressed with Compass when I saw it, not the last team, but I think 2023.

Say some more about what Compass is and how that helps give you a, a window into everything that's happening within the kind of the software, both creation, but operations world too. Yeah. So, uh, compass is a developer experience platform.

It's aimed at removing the points of friction, uh, for developers. And it does this a few different ways. Um, we, it aims to enable self-service, uh, and really the, the core feature of it is the component catalog.

Um, so, you know, for every piece of software, there's a whole bunch of metadata that lives around that. And in large organizations, teams document that in very different ways, which means for every team you engage, you need to remember where do they put their stuff? How do I, who is it?

How do I contact them? Uh, and Compass really solves that problem. So, uh, in Atlassian we have one URL where everyone knows where to go to find what they need, and you can very quickly find, you know, what a service is, who owns it, what activities happen, deployments, incidents, uh, you can see complete dependency tree upstream, downstream.

Um, all that information in one place is super valuable for us. Um, so it's something that, uh, Atlassian developed, I think eight years ago now. It was an internal, uh, platform and since then a lot of customers started telling us that they were having those problems.

They were asking us how we solved it internally. And so, uh, we changed it into a product that our customers could also use. Nice talk about.

So part of what must kind of fuel that is the teamwork graph, because it isn't just saying, well, who checked that in? Let me go talk to that person. It might be long gone.

Someone else is doing work on that part of the code, or whatever it is. Talk about the role that the teamwork graph plays in, whether it's Compass or AI or the development experience. Yeah.

Teamwork Graph is really the foundation of, um, of all of our products plus all the products that, um, that developers or, or companies might be using. So it brings information and it can link information from things like source code repositories to projects in Jira to, uh, incidents that happen in production, uh, to Google Docs, um, that someone might be working on. And it's really that rich set of data with all the connections between that really help us paint a picture of, uh, what teams are working on, what their goals are, um, what issues are they facing.

It's a really powerful thing when you, when you connect all those things together into a single platform. Um, and that enables us to do things like with vo. Um, there's so many awesome things that you can do with it.

If you were working at Atlassian, you could say, what's Andrew Boyagi you been working on? And it'll bring up a whole bunch of documents and confluence pages and, uh, maybe PowerPoint slides and things that I've been working on. Um, and it'll just show you at a glance, you know, with all those things, what are my goals?

Uh, it brings all that information together in one place. So, uh, the one thing that AI needs to be good is a rich data source. Mm-Hmm.

Uh, and the Atlassian ecosystem is one of the greatest and richest, uh, platform, uh, information sources that, uh, is out there in many companies. And With information context is everything, right? That tells you who, what, what's happening with it.

Uh, so you don't have to work just within the Atlassian ecosystem. A lot of people work with GitHub. Um, you have a connector into GitHub.

I think the search now is working, the AI search works across GitHub, is that right? Yeah, Absolutely. So, um, you know, our, um, our motto is really open by design.

So we recognize that, um, teams should pick the best in breed tools that they want to use, and they're free to pick any tool they want. They can connect it back into that, um, Atlassian ecosystem. And it becomes a part of, uh, the teamwork graph.

So whether it's Figma or, uh, any other tool, you mentioned GitHub, it could be GitHub. Um, any tool that a team is working, once they connect that into the Atlassian ecosystem, it becomes part of the teamwork graph. And we are able to link that information like goals to teams and the work that they're doing, uh, making it super powerful.

And Mike, CEO of Atlassian talks about, um, in another session, talks about the architecture is one of the strengths. The investment you've made in it isn't just by happenstance, AI pops up here, right? And we didn't just bolt on ai, we've gotta have the teamwork, uh, or having the teamwork graph, having the data sources to build that on.

As a software engineer, put that hat on, you know, how do you look at the architecture of Atlassian and what they've, what you've created? Uh, it's amazing. So, you know, as Mike would've, uh, mentioned, we've invested heavily in the platform, uh, which is why you see things like Rvo working across all Atlassian products seamlessly.

It's not something that was built into Jira or built into Confluence. It was built into the Atlassian platform, and it reaches into all of our products and all of the products that customers connect into that ecosystem and makes that, uh, as you said, search really powerful with context. Um, and it helps really link work to goals, to teams, um, making it powerful.

You know, it's not like, uh, we see a lot of products where, uh, they've added an ai, you know, bot to their tool, and it's great. It's still a lot better than what we had before. Uh, however, the context is really limited to the information that's in one specific product that they're using.

Uh, whereas, you know, Atlassian and the platform that we have enables our rvo to go across everything, find the context, and really deliver, uh, something valuable to the user who's trying to find something Very good. What, what do you predict the future is for DevOps? How is it gonna evolve and change?

It's, you know, some people said platform engineering, DevOps is dead. A lot of us said, oh, I think there's a lot of work to do on both. Uh, what, what do you think?

Where are we going to next as you kind of look ahead in software engineering? Yeah, I mean, DevOps is not dead. Uh, strongly disagree with, um, some of the, that that comes out.

Um, I think like DevOps and the principles, a lot of teams are still trying to adopt DevOps principles and they're still moving towards, uh, really nice maturity in terms of DevOps. Uh, in terms of platform engineering, I think that's a great way to enable DevOps. Um, if you look at DevOps and how, uh, companies have adopted it and some of the challenges they've face, it's too much to accept, uh, to expect a software team to also look after their dev tool chain and to do everything associated with that.

And so, uh, platform teams and platform engineering, uh, really an enabler to helping teams move to a sustainable DevOps model. Um, so if you look at, um, you know, some of the challenges and the friction points that engineers face, you can't expect each software team to solve their own friction points at scale, especially when everyone's having the same problem. So, uh, in terms of platform engineering, I see that as a super important in terms of helping teams adopt DevOps.

And that's something that's happening now. Uh, in terms of what's happening in the future, um, obviously AI is gonna play a huge role. Um, every month I'm seeing so much innovation in terms of AI being applied to a new use case.

And I, I think we're still, uh, AI's still quite immature, and I think as we see that mature, we're just gonna see better and better results in terms of helping developers get back to delivering high quality software. Fasts. I'm, I'm guessing you agree with me.

Really, we hit the beginning a phases of how much AI is gonna play a role in, do you believe all the predictions, we won't need developers anymore in six months or a year? No, um, not even in a longer timeframe. I think we'll always need developers.

And the reason for that is, I was talking before about how the developer role has evolved. Their role is not all about coding now. It's really about understanding a business challenge and coming up with a technology solution for that.

And coding is a part of that, uh, Systems problem. It's not a right code problem. Exactly.

Uh, I do see that developers will be working with AI and it'll be helping them. Uh, so the role of the developer will definitely change. Yeah.

And it's, it's been changing. So I see a lot of developers who are worried about AI and what's it going to do? And when I speak to them, I just remind them that your role has already been changing for the last 15 years.

This is another change. It might be a bigger change, but there's nothing to worry about. It's something that we should be embracing.

Wasn't that long ago, a developer might say, I'm a C, or I'm a CC plus plus. That's what I know, that's my specialty. Now it's, well, it's Kubernetes and it's, you know, whatever cloud platform, or they're a lot more into the infrastructure part of the application or underpinning the app.

Yeah, I mean, one another way to look at it is look at what more you can do now with ai. Uh, it can help you learn a new language. It can help you do things that, uh, maybe you didn't have the time to learn before.

So, uh, I, the way I use AI is I work with it. Uh, it helps me and I learn, uh, as a result of using ai. And sometimes I stop using it for something once I'm, uh, competent in the thing that I need to do.

So there's new things that we can do. I think we just need to open up our minds a little bit and, uh, see how we can best leverage it. See what you get when you tune in and listen to, to Atlassian team.

Andrew, it's a pleasure to talking with you. Always is. Thanks for our next chat.

So thanks for listening. Follow this guy. He knows what he's talking about.

We'll be back with another great, uh, conversation here. Atlassian, team 24, Europe.