Transforming IT: Embracing AI and Efficiency at Atlassian – Atlassian Team ’25 NA

This is Textron tv, And we're back at Atlassian, team 25 in Anaheim. I'm with, uh, Tal Serif, who's the senior Vice president and head of corporate engineering. Welcome to the show.

And what is corporate engineering? Sure. So, you know, if you think about traditional it, many organizations have what you could think of traditional it, where they would buy a lot of software.

Well, we're taking an approach of thinking beyond that and, uh, specifically looking for not just off the shelf solutions that anybody can buy, but what are the specific problems that we can solve for ourselves that we can then turn around and, and also solve for other customers as well. And so I'm happy to talk to you about one of those things, uh, that we, That kind of plays within this, this ongoing it discussion of buy versus build. Right.

Exactly. Okay. Um, yeah, maybe you could just for just a couple of minutes maybe discuss that and how it applies to Atlassian.

Yeah, Absolutely. So my predecessor really had built a culture of a hundred percent buy, and that worked for a certain size company, but then you get to a place where you really want to understand that the full connection. And if you think about how we've framed it on our show in terms of the teamwork graph and understanding how it, how everyone uses end to end, whether it's our applications or just the knowledge that spreads across the company, we wanted to be able to solve our own problems and challenges around workforce planning to understand who's working on what, what skillset do you have?

Mm-hmm. The relative allocation of people working on those projects. And to do that in a way that was useful for ourselves.

And then we were able to show it to some of our, uh, customers and found that they too were facing some of those same challenges and they asked us to, uh, bring it to market. So we're, we're excited to make an available in our early experience program, um, here at the show today. Well, so how does your team tackle challenges with a product-based mindset?

I think that's kind of where talent comes in. Absolutely. Can you maybe talk a little Yeah.

Explain that. You, You can think of any business challenge that anyone might have, and, you know, historically many companies would look at, um, would use something like a spreadsheet or something to pull together who's working on what. And that would be a, a herculean effort, right?

You'd spread it across everyone. They'd all enter in this information into a spreadsheet, and it would be up to date for that moment in time. But as soon as that was done, it was immediately out of date.

And so what we ended up finding is like we are able to build talent to not only answer those immediate questions, but then be able to see them over time. And so immediately answer like, if a, a challenge or an opportunity like AI comes in and you wanna, um, move around resources, the best possible resources in the company to build a robo agent or to build search for robo, well, who do we have in the company who has worked on search before? Who do we have in the company who could build the robo connector?

And so how do you find the people with that talent and, and have them work on the most important highest priority work for the company, uh, and, and solve that as quickly as possible. So that's, that's why we were able to, to both build that and why we needed to build our own tooling, because it just doesn't exist. Everyone has that problem.

It's, it's interesting because there are so many AI related announcements. I think this week there was, there's it from Google Next. I mean, you can, you can name almost any company.

Salesforce just made another announcement. Microsoft is making one, uh, Dell. So we've heard a lot about agents at the conference this week.

How has, uh, your corporate engineering team deployed agents? So we have a, we have a very much a, a build culture, a developer culture here, Atlassian. And so, uh, Atlassians have built over 2000 of our own robo agents.

And that can be anyone from building an agent to say help with onboarding. So obviously when, when someone joins a company, they may have questions that they need answered that are, are common questions. And you know, what those answers to those common questions might be in something like a Confluence page.

And so why not take all the knowledge that sits in a Confluence page or in Jira tickets, answer people's questions in real time, whether that's onboarding, travel and expense, figuring out what the right talent is to work on a particular piece of, uh, uh, of solution. Answer customer questions. So, so people have built low code, no code VO agents across the company to answer any number of questions.

And they can do it in a low-code, no-code way, where literally, I don't have to be a developer. I don't have to talk to a developer, but I can see, you know, I spend 40% of my time answering the same travel and expense questions. Why not let a robo agent answer those common questions so that you get your answer quickly?

Yeah. And I can spend my time answering the harder questions or, or spending my time more efficiently on, on the, the more important elements of my role. It's like The way that I, there's a series of enhancements that you all made and the robo search fund was interesting, or even Trello AI to-do list this idea concept that we spend so much time, we waste so much time trying to find something that we know we stored somewhere, whether it be Slack, email, where, where, where, wherever it might be.

Or maybe it's in a Word document, I don't know. And, and in a sense, I think you mentioned earlier that like a quarter of our time is basically spent Looking for Information, looking for things that we know exist, but we don't know where they're where they are. Absolutely.

So you're exactly right. And so whether that's a robo agent answering a question to figure out what's that, what is the right Jira ticket? What is that right confluence page on a particular project or doing some deeper research, we also announced the ability to do deeper research and actually being able to answer a harder question form the foundation to the answers of that question, put that into a Confluence page so that we could work on it together.

So it, it could be anything from a chat based, uh, you know, I have a question I'm gonna ask in a chat. I have a search, I wanna search for the right confluence file the right Jira ticket, the right slack in state ticket, or hey, I wanna, I wanna do that deep research and put together, this is a definitive answer to this policy or this process, or how I might accomplish this particular task. How's the, uh, reaction been or the assessment of your employees when they've been using these, these diff various agents?

Um, were, did they find it particularly useful or easier that, was it easier for them to use than they expected? And were they more open to using ai? Because that's the recurring theme is that people feel threatened by it until they actually experience it.

And I think maybe, I don't know, maybe I'm getting ahead of myself, but if 2025 is being designated as a year of AI agents, maybe as more people use them within their environments, they become more comfortable in that accelerates the use of, of these agents? Well, I think there are a couple things. One, letting non-technical people, non-developers building their own agents, that's an incredibly valuable capability that we're adding.

And I may know my business problem better than I'm able to describe it. And if I can build a prompt to build my own robo agent to answer that problem, or if I see a, a, you know, like I was describing to you travel and expense, if I see the same questions, what can I expense in terms of move going to this team event? Answer that question once, put it on a page, build an agent that then answers it for thousands of people who might ask the question, right?

I can do that as a non-technical user. I could, I can, I can build that in a low code no code way and help everyone in my company be more efficient with their time and get their questions answered more quickly. Um, so Nvidia, CEO Jensen Wang recently said it has become the HR for ai.

Does, does that resonate? Absolutely. When I think about, um, how many robo agents get built, like I, I think I mentioned over 2000 Rob, two agents have been built inside of Atlassian.

Think of those as like 2000 more employees that I can put on a particular problem, right? Particularly if I build an agent that solves a specific problem, whether that's answering a help desk ticket, answering an HR ticket, and answering a travel and expense question. Anything.

So it's like a VO agent might just be, uh, very, very adept at answering just one, uh, repeated question that is, that is repeated over and over again and people forget, you know, we have 20 different people asking the same question that that agent basically resolves their their issue. Maybe that's their one task. The the agents.

It could be. I mean, it could be, or it could be they could do other things, but Yeah, I, I typically think of it as something an agent is gonna answer more than one question, and it might answer a large set of questions, but you're absolutely right. I myself may only have one question, but think about hr.

Yeah. And, and how all the tickets around onboarding an employee may show up in HR or password reset. Um, right.

Like think of all the IT kind of service related questions you might ask. There's a huge history of all the questions that people ask. Company Passwords are a major password, a major, you have to go to the IT person.

I've been asked to, to restate my password for security reasons I don't remember. I have it saved somewhere, or I changed it multiple times where I don't know the latest version. So that's, you know, and it's, and it, I've wasted our IT person's time a few times that Well, I was gonna ask you, what, what's your best piece of advice for IT teams to succeed in ai?

I would say lean into ai, right? I think all of us are feeling we're at this amazing inflection point Yes. Where we know AI has so many capabilities and can answer our common questions.

But I think there's, I think you even said it, sometimes there can be a little bit of trepidation Yeah. About leaning into it and taking advantage of things like robo, where you can build a low-code, no-code agent to answer questions to solve your day-to-day business problems. Instead of sitting back and waiting for someone to build you the perfect solution, you can actually learn, iterate, solve the problems you have day to day, and help us evolve, help help your, um, own employees and customers evolve in the use of that technology to solve their own questions and problems.

Well, you're gonna, I think based on what you announced today, Atlassian has announced and what they're, what you plan to do, I think it's gonna make work sim more simplified, more organized, more efficient. And I think a lot of companies are gonna strive to go in that direction and just kind of reduce the clutter or the unnecessary time we spend chasing things that are just, you know, in a sense, just meddlesome. So, thanks, uh, very much for being with us, Tal.

Thank you so much. Thanks for being, I really appreciate It. Thanks for being, it's good seeing you.

And, uh, we'll be back with more from Atlassian, team 25 in Anaheim.