3249: Atlassian: Rovo, ROI, and the Rise of the AI-Powered Workforce

[00:00:04] What happens when AI becomes more than just a buzzword and starts acting as a real team member? Well, at Atlassian's Team 25 event, that question was no longer hypothetical. And in today's episode, I'm joined by Jamil Valiani, Head of AI Products at Atlassian. And he's going to share the story behind Rovo, an AI suite that's now integrated across Atlassian's core tools

[00:00:32] and offered to customers at no additional cost. But what does this actually mean for teams navigating noisy workflows, information silos, and that constant pressure to deliver more with less? Well, my guest is going to help me break this down today from demystifying the concept of agents and their potential as AI-powered teammates.

[00:00:57] And he'll also explain how Rovo is quietly saving companies time, boosting quality, and even giving teams their creative energy back. And we'll also unpack what makes this launch a pivotal moment in the evolution of work. And if you're wondering how to start your own journey with AI and AI agents, agentic AI, AI, or questioning how you can unlock the more practical ROI of this technology,

[00:01:26] rather than just adding yet another tool to your stack. Well, this conversation promises to offer you clarity, strategy, and more than a few surprises along the way. But enough from me. Time to get my guest onto the podcast now. So thank you for stopping by and joining me on the podcast today. Can you tell everyone listening a little about who you are and what you do? Sure. My name is Jamil Valiani. I'm the head of AI products for Atlassian, which includes the Rovo product,

[00:01:54] which we spoke a lot about in the last couple of days. I've been at Atlassian closing in on two years now. Before that, I spent about 20 years at Microsoft working on mostly search and AI technologies there. And you said something now about Rovo. We had a lot of conversations around it over the last few days. A lot of excitement, particularly because it's not going to cost our customers anything as well. But I don't want to lose anyone. So just to set the scene for our conversation today, for people that have not heard of Rovo, what is Rovo? And why is it such a big deal here this week?

[00:02:22] Yeah. Rovo is a core part of the Atlassian platform. It consists of three applications or three apps and a set of agents. The apps are Search, Chat, and Studio. Search is a world-class enterprise search engine that is able to index and search across more than 50 applications. And your favorite SaaS tools are all included, as well as your own data connector. So if you have a custom data source and you want that to be searched, we have an ability for you to go and search across that as well.

[00:02:52] We also have Chat. Chat is all the power you think of when you use a tool like ChatGPT in the consumer space, but grounded in your enterprise. That means, number one, it's secure. So you don't have to worry about your data being leaked to train other models or things like that, but also able to reason over the data in your company. So instead of just asking generic questions like, tell me about car A versus car B, you can actually go and say, hey, tell me about this project or this decision. Who are the key players?

[00:03:20] And it'll actually be able to go and reason over all that knowledge in your company. And then take action. You can actually tell it, hey, based on this, create a Jira ticket for each of these issues, or spin up a Confluence page, or even third-party tools like Google. Go create a Google Doc that summarizes this information. Then we have Studio, which helps you build agents. And that's a new tool that we are releasing now that really rolls up all of the learnings we've had, watching our customers build agents, weave them into automations,

[00:03:47] and really our take on how do we make that a smoother process so that everyone can partake. So we know that in this AI-powered future, we want AI and team members to work together, humans to work together. And that means that we have to make it easy for the humans to create the AI-powered tools and agents and whatnot. To help them. And so our goal with Studio is to make that as easy as tapping in natural language. So you can say, hey, I want to create an agent that does these things.

[00:04:15] And a lot of the legwork to create that agent is done for you. And then you can spend time tweaking it. You want that agent to run every day at 9 a.m.? Well, that's an automation. We can go and create the automation workflow for you so you don't have to go and think about, well, starting that blank page and trying to create the different boxes and diagrams. It just does it for you. And then you can, again, you can tweak it, customize it, then try it out, make sure it works to your needs. So those are the three apps. And then we also have a bunch of out-of-the-box agents that are included, what we call the teamwork agents. And the idea is that this helps.

[00:04:44] These are agents that we think every team could find beneficial to help you get started. So try out agents, get them working in your company, helping your team, taking off those mundane tasks. And also from some things you can learn from. So you can get to know how an agent works. You can maybe tailor it, customize it to your own needs. And that could see the confidence to go and start building your own agents. So, but that's all of that. What I just mentioned was what we announced as being now included in all of our paid subscriptions. And there's a lot of excitement around agents at the moment.

[00:05:14] I think it was Gartner last year predicted that 2025 would be the year of agentic AI. It would be the year of agents. And I think a lot of business leaders, maybe they're a little bit overwhelmed or maybe unsure on what does agents mean to us? How are we going to use it? How is it going to solve real problems for us? What's the ROI of this tech? All those questions you must be bombarded with. So you're making it easy, you're lowering the barrier for people to get involved and businesses to use these agents. But how would you describe that journey that they're on? Because you've probably spoken to quite a few here

[00:05:43] that want to jump on board too. Oh, for sure. Yeah. And we were definitely early to the party. We announced agents first a year ago and we were showing them at Team 24 in Las Vegas. And I don't think anyone was talking about agents. And we laid out a framework for what they are. They are AI-powered teammates that you can summon. They have specific knowledge. So you can say, hey, like your knowledge base maybe encompasses what's in these documents or these pages. So they have a lot of awareness. They have a set of capabilities. So you can tell the agent, hey, these are the things that you're able to go do.

[00:06:13] You can edit tickets. You can update a page. You can go and render these sorts of decisions. They tend to be trained with specific instructions. So that helps them become experts at certain things. That means that they can, you can say, hey, this is an agent that I'm going to really teach it what it means to triage a bug, right? Like, you know, I have these standards. You know, I want you to make these types of decisions. Something else may be trained to, you know, make sure that you write documents in a correct tone of voice, right? Like it helps you make those things. So you give those instructions. And finally, you might give it some character, right?

[00:06:42] Like, hey, like be friendly, you know, like, you know, be succinct, right? These are all things that they sound small and maybe, you know, comical sometimes. And they can be. But they're also important. Like if you have an agent that you're relying on to create an executive summary, but it's long-winded, that's not really helpful. And so when we say, hey, like, no, this is the personality of the character we want. It's really to help you adapt to the culture of your team so they can feel like they're part of the team. So that's really like, you know, the key components of an agent. And really when you put that all together,

[00:07:11] it winds up just being a virtual team member. And I think the real exciting things happen when we see mass adoption and all these technologies converge. But securing adoption is tricky. But I think here at Team 25, you may have solved that problem by offering it as free as part of the existing package. And there is a lot of subscription fatigue out there. So how has this been greeted? I would imagine it's really helped increase adoption and excitement in that. Is that what you're hearing? Yeah.

[00:07:38] So I think that certainly one key part to adoption is removing these upfront barriers. And we think that especially with something that's so new and greenfield as incorporating agents as these AI team members, we really think that getting people to, you know, try them out, like start, you know, seeing what they can find value in when they bring it to their team. What are those first sets of problems? We want to get people to that level of discovery as quickly as possible. And I think that our move this week and all the feedback I'm hearing on this front,

[00:08:08] I think we succeeded in that goal. Like all the conversations I'm having is really in, hey, like, how do we get started? Like, let's, what are the first things we build? What are the first types of problems we can go solve? What are your recommendations on best practices? These are questions of people who are now like feeling empowered to go try things out. And that was a pretty substantial goal for us this week was to really give people to that level because we know that that will accelerate this sort of infusion of AI and humans working together as a team. And I would imagine when you build these agents

[00:08:37] or when you build the entire system, you probably have certain use cases in mind and things you want to go after. When you speak with all your different customers, they're probably going to come at you with many, many other different kind of use cases that you would never consider. Are there any bizarre or unusual or things that you would never have seen coming that people are asking you about? A lot of the different use cases we've seen often are starting with like the drudgery, you know, like it's like, I do really cool stuff at work. Our company is doing something really inspiring, but gosh, I spend half my time

[00:09:07] just like managing the status reports, managing the sort of incoming tickets, triaging the bugs. And really our goal is to say, hey, the agents will take away the drudgery, right? And let the humans work on the really critical, gnarly problems that really make their company, their team special. And so from that perspective, actually like I get excited when people say, oh, can you go and like help me with this like seemingly boring use case, right? Like doing the status report every Monday at 9 a.m. I was like, no, no, please, I want that. I never want you to have to worry about that again, right? You know, and if we can take that off your plate, great.

[00:09:36] And I think what we're learning a lot about are the rituals that people have, right? Like, for example, something that sounds kind of complicated, like, gosh, every, you know, every time we get to a customer interview, we record it in Loom. And gosh, there's like a bunch of work after that to go from that recording to actionable things that a team member should work on. And it's things like, hey, transcribe it, figure out what the actual nuggets of information are, file them as JPD items, like log things in HubSpot for the sales team to follow up on. You know, like these are all like,

[00:10:06] like send a summary in Slack so the rest of the team can benefit from it. All important, right? But man, like that's all like a play, like a half a day, a day is days of work before that product manager or those engineers can actually start actioning on the customer feedback. And so we hear lots of cases like that where people are saying, gosh, like, yes, I feel like I can eliminate these sorts of steps or reduce them dramatically and then get to the thing that's truly amazing. You know, I had Ronnie Katzenberger who from Procore, who was one of our early Rovo customers on stage with me yesterday.

[00:10:33] And I heard a ton of great use cases from him. Like they really leaned into this. They are using agents now to help them, you know, enforce quality guidelines, make sure that every time they enter an issue, it's up to a reasonable standard. And that's helping their team in all sorts of ways, communicate faster, move faster. And that's maybe the next level I see is like, you know, folks really using agents, not just to take away the tedious work, but to help them up their game a bit, right? And that's the kind of direction I'm seeing this go as well, which is exciting to hear. And another thing I think is exciting is as this technology matures,

[00:11:03] we're moving away from the hype, solving real world problems and getting back to good old fashioned ROI, you know, it's solving those real problems. And before you came on the podcast today, I was doing a little research on you and I was looking at a few of your posts that you've shared in recent months. I think a few stats here, customers experience two times ROI on their AI investments. They save around 105 minutes a day, a full workday each week. And 85% say it's improving overall work quality.

[00:11:32] Can you tell me more about these stats and the real world tangible benefits and measurable impact that you're seeing? Yeah, for sure. I think across the apps I mentioned, you know, like they each sort of self-complimentary challenges and AI powered team. I think with search, what we're seeing more and more of is your search quality is just getting better and better. And as we add more and more of these connectors, people are finding what they need. They're finding it faster. We think they're saving at this point up to one full. We know that people are spending one full day a week

[00:12:02] just searching for information. And we think that with a lot of what we've done and the enhancements we've provided, we can cut that time at least by half right off the bat, right? Like with things like the AI enhanced answers that help you save yourself from hunting and pecking with better relevance, right? With getting all that data in, those all just save so many just tedious, repetitive steps. And I think that will get better and better and better over time. Yeah, I think with, when it comes to, you know, chat and when it comes to things like agents and the studio work,

[00:12:31] what we're seeing is that folks often when they kind of take that information and they have to go in and start acting on it and they have to synthesize it and understand it, that can be really challenging. Sometimes you're just, you know, this like daunting book of knowledge, right? And, you know, it can just feel a bit intimidating to go and say, hey, like, how do I just make sense of all this? And that's where I'm seeing a lot of folks using chat is just to sort of get those first couple of steps, like, you know, take this big, massive knowledge and just start putting your arms around it. And I think that's also something that when you talk about ROI, you know, you kind of talk about

[00:13:01] saying, okay, like that's, you know, that there's a lot of human creativity required there, but just trying to strip away the noise and really focus on what matters, that's where the agents can help a lot. You know, one of the demos I've been showing has saying, hey, look like, you know, take all this data about performance and help me just quickly understand where I should zoom in on, like which region is underperforming. And normally you'd go put all that into a spreadsheet tool and go and make your charts and it might take you a few minutes or an hour. Here, it's another, yeah, we'll do that for you so you know where to zoom in, right? Because the act is focused

[00:13:31] on this region. And I think that that's the type of benefit that we're seeing now as folks are able to go and remove, you know, save so much time and get to the real, like, analysis part. The types of stats we see there are just truly mind-boggling. I think now by our internal measurements, we test very aggressively internally. You know, we believe that at least 90% of queries and asks of chat are successful, right? Which is pretty stunning, right? And that's all grown in the last just few months alone.

[00:14:00] So imagine a few more months of us tweaking and improving, right? You'll see even better stuff. And now people are building agents on top of that capability, which will help them solve their own problems, I think quite well. I think the latest stat I ended with is we believe that by 2026, the efficiency gain for a typical human will be 4X, right? You know, it was like the latest study that we found. And we can go find the data for you there, but that's, I think, what we were quoting yesterday from our research. And I would imagine businesses that don't take this seriously could quickly fall behind

[00:14:29] at equally a rapid rate. But something else that I read that you said there was that your customers are not just saving time and money, they're also raising the quality of their work and opening the door to new ideas. What are you seeing here? Is it also making companies more innovative at the same time? Yeah, I think that when you start saving a lot of this time and giving people their energy back, right? It's not just your time, it's your energy, right? You drain all those best hours of your day, just trying to chase the process. Then they actually have the room to go and be creative and try new things.

[00:14:59] And it also gives people a voice. There are folks who maybe are not great at expressing themselves in writing or they need that extra bit of help visualizing something. And I think what the agents we're seeing out there now, they help with that, right? So folks who maybe didn't know before how to go and take this thought and translate it into something workable or create that prototype. I think that's the kind of stuff we're really excited to see is folks, you hear more energy and you have more voices able to speak up and share in something that's tangible and effective their ideas, which I think will be inspiring.

[00:15:29] And from everyone I've spoken to here, another thing that stands out is everyone seems really passionate about everything because I think it's because you're not just providing solutions or services. You use these products yourselves on a daily basis and that really comes out there. And there'll be a lot of people listening that are much, the very beginning of their particular journey and you've seen all the mistakes and seen the challenges and come through the other side. And I would imagine that many people listening, having the time to reinvest, to upskill staff and ensure that nobody gets left behind.

[00:15:59] How did you overcome that at Lassian? You're absolutely right. We use it aggressively internally. We have at least 3,000 agents in Zed Atlassian, right? And that can't possibly be all engineers creating them, right? It was the HR people, the finance people, the legal team, all of them creating these agents and experimenting. I think what we found a lot is that they're often beachhead teams, teams that just have a culture around trying things early, maybe because of pain. They have aggressive goals and they're trying to find ways to break through

[00:16:29] or because that's just how they're inclined. They just have a more innovative spirit or a bit of risk-taking appetite. And we found that identifying those teams and empowering them quickly and putting a bit of a spotlight on them really helps. You know, that's what I think because I think folks always get scared or sometimes they're scared of failure or whatnot. But when you go and say, hey, like, let's have these teams go and maybe show us a few different agents they built or a game they had, even if it's small, you know, celebrating them and sort of saying, okay, now what do we do next? How do we go and take it further?

[00:16:58] I think helps a lot. The leadership buy-in also helps a lot. Mike, our founder, he built a lot of agents and he is always using search and he's always using chat. You know, literally even today, like, well, it was probably on a lunch break or something, right? Just sending us feedback saying, hey, I tried these searches and, you know, what do you think about adjusting the query language this way, that way? So I think that level of leadership buy-in and that leadership involvement and inspiration also plays a really key difference in inspiring, you know, that employee adoption, that risk-taking mentality quite a bit.

[00:17:28] So there's other factors, but I think those are the two that I see a lot come up as like positive ingredients that really accelerate this transformation. So for any business leader that's listening to our conversation today, they've officially got access to Rovo for completely free. Where should they be starting on their journey? You know, they can't just go head first and try to solve all the world's problems at once, but where should they start to ensure they don't get overwhelmed and they do get the most out of everything? Yeah, there are a couple areas I'd suggest. The first is, if you don't have a search tool that you like,

[00:17:57] if you're having a hard time finding information, just plug in the Rovo connectors, start using search. It should make your life way easier, way faster. And if you're paying a lot of money for a search product today, you know, you could probably save some money there, right? So consider that. That's pretty straightforward. Everyone benefits from search. They use it multiple times per day. It's a very easy place to start and start seeing benefit. The second is, when it comes to using, you know, agents and thinking about this whole sort of agentic transformation and the buzzwords around it, the best thing I find is, you know, of course, like, you know, like I said, you know,

[00:18:27] you inspire teams to find the beachhead teams that will early adopt. But I think also as a business leader, you can look at the processes that are already in place in your team. And oftentimes, people have written automations, right, to try to make them smoother already. And automations are almost like really miniature agents in some way, right? They're like the earliest versions of agents. They're like simple, simple decision trees. But there are often good places to start, right? Because you might say, okay, this automation might succeed, you know, some percentage of the time, but and it runs maybe multiple times per day. Well, gosh, what if I could just

[00:18:57] make that automation a bit better, right? You know, have an agent go in and help with some of the areas where it often fails by introducing some reasoning, the ability to go make some better decisions versus simple yes or no type decisions. That is something that doesn't require a whole team to go and learn how to build an agent. It requires to have access to that workflow and start working it in. And what you find is that if you even improve it because automations run all the time by definition and they help move processes along, once you're able to

[00:19:27] deliver those improvements, everyone in the team who relies on it realizes the benefit, right? Because they experience it. When you see the agent showing up saying, oh, by the way, I made this decision or by the way, like this agent took care of this for you all behind the scenes. Nobody had to think about it. It just showed up. It makes people curious and excited because they see the benefit. They didn't have to think about it twice, but it definitely prompts conversation. It prompts more thinking about other things that can improve. So we think that's also a really cool place for any leaders if they're trying to kind of get that immediate ROI, that first sort of dopamine hit of success.

[00:19:57] And obviously, we're in a week where there's been so many big announcements, but I've always got one eye on what's going on around the corner. What's next? Where do you go from here? Is there anything else you can share around Lassian's overall AI strategy and what we can expect? For sure. So there are a few things. One is that we continue to make all of these core tools that we've now infused in our platform, the search, chat, and studio tools way, way better. So for example, we announced Deep Research, which is coming out in a few months in chat. So the ability

[00:20:26] to really go way back in your company's history and understand this project might have had different incarnations, different people working on it, gathering all that data together, making sure you can make better decisions, also grounding that on public knowledge, right? So I could be able to look at web knowledge. That's all stuff that is cooking right now that will come up in the next few months. We're also working to make chat and agents more multimodal. So one of the demos on the floor, I believe it's out there, is you can take, if you want to create a workflow, for example, in Jira,

[00:20:55] you can maybe draw it out as if you were on a whiteboard, right? You know, drawing out your workflow, take a picture of it, and an agent will actually go and create the workflow for you, base of the picture, right? So that's like maybe a taster of like, you know, where we're trying to, the types of things we're thinking about on like how far agents can go. We're certainly also working on making sure that as many people as possible are enabled on building, you know, agents and other AI accelerated workflows. So that's where you'll see Studio really just getting better and better and better over the coming months. Like I think we, we had our first version out now, but I think you can expect

[00:21:25] like every month you'll see like, you know, better and better enhancements. And we're always working on improving search check. We're going to make those connectors better. There'll be, you know, even more fine-grained options on how to configure them. The hints we give you in search, the things we do to go and make, make sure you have experience that you find what you're looking for. We are shipping improvements every week there. So I think you'll see a lot of exciting stuff. We've also, for folks who are, you know, more have greater security or regulatory, you know, needs, we've been working on self-hosted, or sorry, not self-hosted, Elaston hosted LLMs

[00:21:55] so that they can have an additional layer of security guarantee if that's a concern for them. Again, that's all in the next few months. Wow. And I think one of the big standout things here is you no longer buy a service or a product and then that is it for two, three years. It continuously evolves every week or every month. Anywhere in particular, anybody listening should be checking out to stay up to speed with those hints, those tips, and those constant evolution. Yeah, for sure. I mean, we're doing as much as we can in the product itself, right,

[00:22:24] to help people learn and discover new things. So if they're an Elaston customer right now, they can, when we turn on Rowa for them, they'll start seeing new buttons, suggestions. We try to be really clever about in product suggestions and making sure things show up at the right time to help people out without being annoying. We're also working on a proactive AI experience so that instead of a, you know, having to know, okay, I issue, I click here or I try this agent out, there's always a robo-suggestion available for you and you can go and quickly look at it,

[00:22:54] peek at it, say, hey, like, you're the agents I recommend, that could be helpful in this context. Here's a couple of chat suggestions if you want to go start a chat and maybe ask, you know, understand this content better. We're going to make that sort of suggestions corner or that suggestions button be really powerful for folks so they can always know that they can take a peek there and see what the latest and greatest is. Because I don't want folks to have to go and, you know, well, yes, we'll do blogs, we'll do all these sorts of things, but we know that most customers, they use the product every day, that's the best way to go and, you know, tell them about something new and we just have to pick the right moment in time where it can be most helpful to tell them that. Awesome.

[00:23:24] Well, I'll also add links to other areas so people can stay up to speed with everything, but I appreciate how busy you are during the event. You've been on stage so just a big thank you for stopping by. No, thank you for coming here. It's been a blast and I'm looking forward to hearing feedback from your listeners. I think as we heard from my guest today, Roro isn't just about chasing hype. It's far from that actually. It's more about simplifying work, amplifying human potential and giving time back to people who need it the most.

[00:23:53] And with agents that triage bugs, streamline documentation and even interpret team rituals, what we're seeing here is AI as a collaborative partner, not a replacement. And I think this is something that we really need to hammer home. And for any leaders wondering where they can begin, I think his advice is clear. Start with search, experiment with agents and let adoption build naturally. And the future of work won't be built

[00:24:23] by AI alone. Let's dispel that, miss. It'll be built by teams, teams that know how to use it. But what do you think? Are we ready to trust AI as a teammate? Are you ready? And how are you thinking about adoption in your organisation? As always, email me techblogwriteratoutlook.com tweet me at neilchews and you'll also find me on Instagram and LinkedIn at neilchews too. Please, share your thoughts.

[00:24:53] Let's keep this conversation going in the Tech Talks daily community. and I'll be back again tomorrow with another topic that we can dissect together. And I'll be beamed back into your ear balls tomorrow morning. Bye for now. علinton fi Thank you.