3243: How Atlassian CTO Turned Developer Joy Into a Performance Metric

[00:00:03] What happens when developer joy becomes more than just a feel-good phrase and instead help shape the DNA of a company's entire engineering culture? Well, at Team 25 in Anaheim, I sat down with the CTO at Atlassian to find out more about exactly how this mindset is transforming everything from productivity metrics to platform innovation.

[00:00:30] His name's Rajiv and he leads all things engineering at Atlassian, from product platforms and security to tools and teams that ship the software. But in our conversation today, I want to unpack why developer joy isn't just some fluffy concept, but a measurable North Star and how it's helped ship AI products like Rovo in record time.

[00:00:52] And also what organisations everywhere can learn from Atlassian's approach to building responsibly with AI. And we'll also explore the engineering shifts needed to make AI a true team-mate rather than just another tool, the role of upskilling and the fine line between experimentation and accountability.

[00:01:15] So whether you're leading a global engineering team or rethinking how AI might fit into your business, this episode offers a grounded, transparent view into the future of work and a conversation from the people that are helping shape it. Can you tell everyone listening a little about who you are and what you do? Sure, yeah. I'm Rajiv Rajan and I'm the CTO of Atlassian.

[00:01:38] I manage all of the engineering at Atlassian. That includes the products we build, the platforms, the IT team, trust and security and our core engineering infrastructure. Well, thank you for stopping by today at the Team 25 event here in Anaheim. And one of the things that stood out to me over the last few days is how you've made developer joy. And that phrase is a company-wide Northern Star or North Star, should I say.

[00:02:04] So what made you prioritize emotional resonance like joy over more maybe traditional metrics like velocity or output? And how has that shifted how your engineering organization operates every day? Yeah, it's a great question. The developer joy concept was sort of natural to me. I started my career as an engineer and I just remember the joy of writing code and seeing how that works and typing up something and see how that works on the screen.

[00:02:31] And like just getting your programs to work is a sort of a joyful process, right? The where the joy stops is when you're working with complex code bases and it's not your code and you get stuck and those kinds of things take the joy away from coding. And that's what usually causes productivity to drop, right? So while we are all interested in having higher developer productivity and the Dora metrics being better and all of those different things, the ultimate price is really the joy of it, right?

[00:02:57] And if you take care of, if joyfulness is there, then the productivity has to be there because you can't have joy without productivity, right? So while we did focus on all the different things like PR cycle time and inner loop and outer loop and CICD and all of that stuff, and we have made tremendous progress in all those metrics, one of the simple things we did is, and we still do this every three months, is we send a survey out to all our engineers in Atlassian and we ask them a simple question. Are you productive in your code base?

[00:03:25] And then we ask them a few more questions in terms of what's coming the way, et cetera. But when we first did this two and a half years ago, that score was 49%, meaning less than 50% of our engineers felt they are productive and felt like developer joy, essentially. And the latest survey we did like two weeks back, we got the results. It was 75%, right? And so like we've had a big jump in more engineers in Atlassian experiencing developer joy, being productive in their code base. And that's not come by accident.

[00:03:54] We've done a lot of work towards it. We have a whole program with three pillars for developer joy. The first is awesome tools. You're as good as your tools. Engineers need great tools. And so we did a lot of work to get the tools to be awesome. The second is awesome engineering culture, which goes into modern code bases and the tech stacks you use and so on. So we did a lot of work to get the code bases to be more modernized, refactoring, et cetera. The third, which is very popular, was empower teams.

[00:04:21] We told teams, take back 10% to 20% of your developer budget, developer time, and go fix the things that bother you day to day, right? That pothole that's on the way every day in the morning, go fix it, right? And engineers really loved it because they got time back to go fix the things that they knew were bothering them every day. And you fix the top five things. Next six months, you fix the next top five things. And before you know it, you're experiencing more developer joy. And one of the reasons I wanted to highlight developer joy, because we are at this magnificent event.

[00:04:49] There's so many big announcements here this week. But, of course, none of that is possible without developers. I almost want to go into Steve Ballmer in pressure and just shout developers, developers, developers. But, I mean, one of the things we've got to highlight here is as a result of that developer joy, you've shipped Rovo, arguably Atlassian's most ambitious AI product to date, incredibly quickly. In fact, it was record time.

[00:05:12] So what changes did you make in your own R&D or to move at that pace without compromising quality or responsibility? Well, there's several things. One is that when I started off here three years ago, we made it a goal that we want Atlassian to be a world-class engineering organization. We want our engineers to be world-class, the best in the world compared to any other company. And so we did a lot of things around that. Developer productivity was one such effort.

[00:05:38] The other was we went and raised the bar on performance and really raised the bar on talent overall. We brought in a lot of strong talent from outside. We up-leveled a lot of the people inside Atlassian. And overall, we did a big push in talent. And so we have some of the best engineers in the world on AI and search and ML and Rovo. And that's really helped us build Rovo in record time.

[00:06:02] From concept to getting Rovo up and running, less than six months, we got a lot of people hired in with great talent in the space. But we also created something called AI School, where we took Atlassian engineers and took them through a 16-week kind of course modeled on Stanford University. At the end of which, you are an AI ML engineer. And so that increased the number of people we had who had AI ML skills.

[00:06:26] And so between the people we hired and the people we upscaled, we were able to get a number of people on to Rovo super fast. And then last but not the least is just everybody realized this is an important moment in the history of technology with AI being something that everybody was excited on. And so that excitement also got us to like... And when I was doing a little research on you before we met today, I read that you said that AI should handle the clones of development so engineers can focus on the creativity.

[00:06:56] So where are you seeing the most significant productivity gains and creativity gains so far? And what are the limits of current AI tools in supporting those more profound developer thinking that we'd like to head towards? Yeah, I think I was at Berlin at the VR Developers Conference last year.

[00:07:14] And I think the thing I said was AI right now is taking care of the chores, the manual, tedious tasks, so that it frees up more time to do the more creative aspects of the developer joy parts of your life, right? And so that's really true. Like AI, if you think about something you need to fix that's a feature flag or you want to remove some dead code in a thousand different places, you can create an AI agent to go do that for you, right? Or if you have a blank page problem, you are starting to write a new feature.

[00:07:44] AI can get you over that hump and generate some code for you so that you have something to start with, right? And so those are the things where you get stuck and that's manual or tedious or like it's the part that's not joyful. That's the part where AI is giving us the most bang for the buck and helps you really get up and going and improves your productivity.

[00:08:04] Where AI is still a challenge or where it's still a challenge for AI is if it's a complex code base and you're trying to build a feature in that complex code base, it's still not at the point where you can fully understand it. And that's where you still need a lot of human creativity and understanding to go do that. But AI is improving a lot and I'm pretty sure in a short amount of time, it'll reach the point where it can help us on those things as well.

[00:08:27] Some of the demos you see with AI where it's like new website, cool new things, that sounds like magic and it's all blue sky stuff. It doesn't work as well with a complex code base like Jira, for example, where we have 20 years of history in that code base. Yeah, and I think it's that human creativity thing that I'd love to dig a little bit deeper on. So how do you approach upskilling engineers for AI without overwhelming them or creating fear around automation and being replaced? And what does responsible AI literacy look like?

[00:08:56] So like I said, one of the things we did for upskilling is we created this AI school and we encouraged a lot of engineers to go. Even if you're not working on a project that directly is on robo, you're not on the robo team. Whatever team you're on, we made it a point to say, you need to learn AI. Like this is the future. Like we made it a priority for engineers to learn this. We also have our engineers use our own robo dev agents, right?

[00:09:22] And so one of the things we've seen is that engineers who use robo dev in Atlassian have a 45% reduction in PR cycle time, which is the pull request cycle time, which is a big measure of developer productivity, right? And so as engineers see that, they use it more and that results in more knowledge of how to use AI and so on, right? So between the school, between dogfooding our own robo dev agents, and we also have an AI summit where we bring together a lot of knowledge around what's happening with AI

[00:09:50] and how it's being used within Atlassian and how our customers are using it. So those are some of the things that have helped us to sort of like upscale our engineering force to understand AI. We also have our developer productivity central team that's always looking out for the latest, coolest coding assistant thing that's out there that is having a buzz. And we're very quick to adopt it and try it out, kick the tires. And if it shows some success, then we'll make it available to our engineers to go use.

[00:10:15] So from autodev to rovo chat, AI is now embedded into the developer toolkit. But how do you keep that human in the loop in practice? Because I know it's something that you're passionate about, especially when tools are making more and more decisions and drafting code, etc. How do you maintain that balance? So this is one of the things that I'm very passionate about because I believe the key to this is not having AI be some magic black box, right? Like a magician that waves the wand and a rabbit comes out of the hat.

[00:10:45] Like that is not helpful to developers. Developers need to understand how the code works. And so the way autodev or now we call it like coding agent works is it's an interactive way of writing code. It's not like you say something and it goes off and comes back with some magic code. There are like many steps to it. And at each step, the agent tells you what it's doing. So it's very clear like what the agent, I'm generating this code, but I'm doing it because I did X, Y, and Z. Or I'm looking at these source code files.

[00:11:14] Or I'm looking at this document in Confluence. And if you don't like what it's doing or where it's looking and getting its knowledge, you can change that. You can go in and say, no, no, no, don't do that. Like go look at this code or look at that document. Or you're missing this step. And it will adjust. And so by the time you get the code at the end of like five or six steps, you have worked with it at every step. You understand what it's doing. It's not like a magic black box to you. And that, I believe, is the key to this. Because then the code that gets produced at the end of the day is something the developer understands.

[00:11:42] Because it's a developer who's going to submit the code to the repository. It's not like the AI that's going to do it. And so your name is behind it. And so you want to know like this code is good, right? And so that's where the human angle comes in. Like it's a human being who commits the code, not the AI agent. At least that's how it is today. And so you really want humans to take accountability for it. The analogy I would give is like when Airbus came out with a fly-by-wire plane, right? If you ask pilots, they're like, they want to know what the plane's doing, right?

[00:12:11] You don't want to be flying the plane and like some black box is doing things for you. And if you look at whenever there's problems, it's because like the pilot didn't understand like what the plane's doing exactly, right? And so that's the kind of analogy you want, right? Code is written by human beings for human beings. And you want the human element to be there. And the key to that is really the transparency of what AI is doing. And one of the things I try and do on this podcast every single day, shine a light on areas around the world that are doing so well with technology.

[00:12:40] I mean, I'm from the UK and you're here in the US. If we look beyond that, one of the things I was reading about you is you've described India as a global force in AI adoption. What do you think the rest of the world can learn from the mindset and momentum of India's developer community right now? Yeah, I think India has just positioned itself as a natural talent powerhouse for software engineers. Maybe it's due to the fact that a lot of people in India had success.

[00:13:09] Programming or math is valued a lot. So like I think like that leads to a lot of people doing it. But one way or the other, there's no denying the fact that like the talent available in India for software engineers is pretty awesome, right? And in fact, India is one of our fastest growing geos for Atlassian in terms of number of engineers, right? So I think it's to me, it's a great asset for the world. And I'd say any company that wants to do anything should be looking at India for talent.

[00:13:38] We also think the way we are doing it in India is a little different than other companies. We don't give like our sort of backwater project to India. We actually have our most advanced projects coming out of India. We have global teams in India. So it's not like it's a small team that's an offshoot of a team in the U.S. or the U.K. or Australia. Like sometimes the teams in India have engineers in the U.K. or other places reporting into them, right? So really treating India as a first class place where you can build products from scratch

[00:14:06] is the change, is the difference in Atlassian's approach. And I think that's going to be a trend going forward. And according to Atlassian's State of Developer Experience report, which I think surveyed something like 2,100 external developers right across the globe, one of the things that stood out to me was that only 44% of developers are aware of the issues that contribute to a poor developer experience. Did that surprise you? Yeah, look, I think we are, like I said, we are on this journey of world-class engineering.

[00:14:36] And to me, like a big part of world-class engineering is having that deep connection down from engineers all the way up through leadership. And we've made a tremendous amount of progress in that sort of direction. But one of the things we are doing, for example, is we are doing Ship It, which is a tradition in Atlassian where it's our hackathon, essentially. We really want everybody to participate. I participate myself, and I make it a point to try to write some code. And that's an activity that we encourage a lot of our leaders to do,

[00:15:05] because no matter where you are in leadership, you want to have your hands in the code and understanding how things work for everybody. The other area where we have made a lot of progress is in incident management. Whenever there's an incident and something's not working, we make sure that we have everyone who is involved, from engineers to engineering leaders, understand what's going on, and try to go and then address it in a post-incident report and so on. But yeah, this is something that's important to me, and it's very important for leaders to be hands-on

[00:15:34] and connected to what's happening with the engineers on the ground. And the data I was reading in the research, it also points to somewhat of a mismatch in some organizations on AI, where leaders see it as a productivity silver bullet, where some developers just aren't seeing the tangible gains yet. So I always try and give people listening a valuable takeaway. So how should a company listening or a business leader listening reframe their AI strategy to better align with the developer reality and avoid chasing the wrong improvement? Right.

[00:16:04] So I'm in the camp of saying that we are seeing some productivity gains, but it's not in the crazy territory of what some folks are saying. Maybe we'll get there where we get tremendous gains, but right now we are seeing perhaps a 10% to 15% boost in productivity. Many engineers save seven to eight hours a week, and that's more realistic situation. So we are being very realistic in terms of where we are today. And we're also seeing that most of the gains are in sort of eliminating these manual repetitive tasks,

[00:16:34] the chores, if you will, of programming. And that's where we are focusing our AI efforts on, is to find those set of things, removing dead code, adding in feature flags, or accessibility improvements that you can repeat across the code base. So we are directing our engineers to look at those activities as the places to focus on AI and not like some magical dream where AI writes the whole thing for you, right? We might get there someday, but it's important to be realistic. It's important to crawl, walk, run.

[00:17:02] And so that's been our approach in a class. And for any CTOs listening that are feeling the pressure right now to have that solid AI strategy, any tips on how they should balance the need for experimentation with accountability, especially when ROI in some of these tech projects is still murky, especially for the AI projects? Yeah. So I'd say a few things. One is like obviously upskill your engineers because AI is here to stay, and it's something we all need to learn. So you want to be looking and doing that.

[00:17:31] You don't want to suddenly put like a thousand engineers into AI just because like you want to do that. That's not going to help you like build real things. Like even we have done a lot of stuff on robo, but we haven't gone crazy. Like we've put the number of engineers we need to build robo, not like twice the number of engineers, right? So be cognizant of how many people you actually want to put in AI. And most importantly, look at the value that your AI that you're building is giving to your customers, right?

[00:17:57] If you build a bunch of AI stuff that no one's using, that's not a good use of your time or your customer's time, right? And so that's what we have done with robo is we have made robo essentially available to all our Atlassian customers for no extra charge. And the reason we have done that is we want to see what actually works. And we want to add value to our customers. We're not trying to like sell a bunch of AI stuff without knowing what the value is, right? So if you are sort of tuned in to what value you're adding to your customers,

[00:18:24] you will be more deliberate about how many people you put on AI and what you actually go build, right? The idea is not to go and try like every single sort of cool thing that you think, but be sort of informed by what is working for your customers. And here at Team 25, there are so many different sessions, keynotes. You're doing back-to-back interviews and conversations with your customers. What's the key message that's being delivered here at the event for people that can't attend? And what are you going to be taking away from the event too? Yeah, a couple of things.

[00:18:54] I think this is a turning point for Atlassian, this particular team event. People have known us as the Jira and Confluence Company, which we are. But we are really making a big change over here towards being the system of work for all employees in all companies, not just for engineers and IT folks. Anyone in any department can use Jira. Jira is available for everybody as part of the teamwork collection. And it's powered by Rovo, which is our AI product that we believe is awesome and one of the best in the world in terms of offering world-class search,

[00:19:24] chat, and agents as part of AI Agent Studio. So to us, that's a big statement we are making here is that we are here to help you do work for your whole company, any company, any size, powered by AI. Well, I know you're incredibly busy. So just a big thank you for taking the time to sit down with me today and share the story and the importance of developer joy. Thank you. Thank you. It was a pleasure talking to you. So a huge thank you to today's guests for joining me in Anaheim at Team 25

[00:19:53] and for giving such an open, honest, and thoughtful look inside Atlassian's engineering engine. And what stood out for me most is how their commitment to developer joy isn't just philosophical, it's deeply practical. It's about removing friction, increasing creativity, and building trust between people and the systems that they use. Whether that be building AI agents that actually explain themselves

[00:20:20] or giving engineers the space to fix their own pain points. Atlassian is demonstrating that if you want velocity, you need joy. And if you want meaningful AI, you need the human firmly in the loop. So if this conversation sparks any ideas on how you can reimagine developer experience in your own organization, yep, you guessed it. I'd love to hear more about it. So techblogwriteroutlook.com, LinkedIn, X, Instagram, just at Neil C. Hughes.

[00:20:49] Nice and easy to find there. And before I go, thank you to everyone at the Atlassian team for hosting such a powerful event, giving me access to so many great people, and for shining a spotlight on not just the people, but the principles behind the code too. And an even bigger thank you to each and every one of you listening. And until next time, stay curious, stay creative, keep building, and we'll do it all again tomorrow. Speak with you all then. Bye for now.