Atlassian CEO on Human-AI Agent Collaboration

[MUSIC] Hello and welcome to the Tech Disruptor's podcast. Here we talk with CEOs and management teams about their views on tech disruption and how it is driving their decision making and strategy. I am Sunil Raj Kapooral, Software, Analyst at Bloomberg Intelligence, part of Bloomberg's research department. For those new to our podcast, Bloomberg's research department has 500 analysts and strategists working across all major world markets. Our coverage includes over 2000 equities and credits as well as outlooks on more than 90 industries and 100 market indices, grantees and commodities. Today we are joined by Mike Cannon-Brooks co-founder of Atlassian, the company behind Gira Confluence and Gira Service Management, tools that sit at the center of how software teams plan, build and collaborate. Mike, thanks for coming on to our podcast. Thanks for having me Sunil. Happy to be here. Great. I know Mike Atlassian, I think, over a course of the last one year there has been a lot of changes at the company. Is there anything that you want to kick off with before we go into some finer or deep dive kind of questions? Yeah, look, it's been a big year for the industry, I think, and it's a very exciting time. Look, I would just say we're rolling really well. I think this comes out after Team 26. We have a massive slate of product and customer announcements, partnership announcements. We are deeply integrating AI into our stack and delivering a huge amount of customer value. And on the business financial side, hitting some really great roles, we had a fantastic quarter and just very excited about the customer value we can deliver with all of the amazing AI technologies of the industry is bringing to the front. Great. So one of the hot debates for this year has been agent take workflows. And I just want to get a hang on what is Atlassian doing there? And what are the some of the products that you are thinking of? And how should we be thinking about the product roadmap? Well, a big question. Look, we're doing a ton in the agentic and AI areas. I suppose it depends on which part of the workflows you're meaning. Fundamentally, the Atlassian platform, so a set of apps and collections operate on a singular cloud platform has very deep AI and agentic connections in lots of different ways. So I can list a few of them potentially. We start with our core pillars, I suppose. One of which is the teamwork graph is one of the core pillars of our Rhovo and AI platform. The teamwork graph is important because context is the hardest thing to accelerate at the moment. Models are improving continually as we see almost every day. And that's sort of think of that as the war intelligence. But the ability to apply that intelligence to your business processes, your workflows, your code, your people is something that's right at the core of what we do and what we provide. We have a lot of updates on the teamwork graph in team 26 and how we're expanding that outside the Allassian platform for the first time in many, many ways and providing that to all sorts of agent platforms, a lot of great customer examples, fundamentally allowing your agents to get better answers, to operate significantly cheaper and to also operate faster. So that's the goal the team work graph is to have that organizational knowledge for you. That's something that's going really, really well. It's north of 150 billion objects and connections and growing fast. We integrate with all the major agent platforms and coding agents through to business agents and we have partnerships with all of the major model providers. Why is that important? I think we try to be as integrated and connected with the customers' business processes and workflows as we can. We have always tried to do that. What that means is that different customers will pick different agent platforms for agent work. I suspect they pick three to five for most major enterprises. I don't think it'll come down to a single one. They'll have different strengths and weaknesses, different access to different things. What's important is that our business processes be they structured like in Giro or the service collection or whether they're unstructured like in Confluence or all loom. Those business processes have access to the agents and the platform. At the same time, the agents and platforms have access to the business processes because the future is about human and agent collaboration. We're really deeply involved in that. Let me see huge customer advantage. We've given a lot of statistics in our various earnings calls about the fact that anyone who uses the agent platforms tends to expand their seats on a last-in-a-lot faster. Their activity rates are a lot higher. There's a lot of things that we're doing when it comes to agents on and off the last-in-platform. That's before you get to the fact that you can build significant agents on the platform and studio. We've passed five million agents created by our customers. We believe in bringing again the agents and humans together to help accelerate our customers' businesses. Now that you've mentioned about agents, I think since the beginning of this year, maybe Hot Topics has been around with the agent AI systems. Where does the seat-based models evolve? What does it mean for software developer community? More importantly, what do you think about the collapse in sentiments, especially for the SaaS space? Look, I feel like we've repeated this a lot. I'll repeat it again. Firstly, as a business, on the top line, we're executing extremely well. Last quarter, revenue in the cloud was up 26%. We increased our NRA for the third quarter running to well north of 120. Our RPL or our long-term commitments from customers grew over 40% year-on-year at a 6 billion plus run rate. Customers are buying and adopting the Atlassian platform at larger and larger levels. They are buying more seats and more applications from Atlassian on a continual basis, despite news to the contrary. The reason they're doing that, and as we've shown, those using heavily using agentic software creations or a cursor or a cloud code, or those using various of the more business-centric agent platforms from Google or Salesforce, they actually expand their seat count faster. They're able to accelerate their business. They're able to get a lot more done. That's the data we're seeing on the ground. So we've given a lot of stats about that. I believe that is what the case will be. Now, there will be a whole mix of results here. There's no doubt about that. There will be some customers who really profit from the AI transition. They invest in it. They lean into it. They have their data organized. They can operate a lot more efficiently and quickly. That is what we're seeing in that end of the market. There will be some who don't manage to make their transition. And that's, I think, the way of most technological transitions. We can also see that in job stats. If you look at product management or development jobs in the US, they're growing very fast. The demand for developers is not disappearing. In fact, it's increasing because you've made them far more productive. There are many, many more ideas for software we want to build. Then there are people who can do it currently. And so you're seeing an explosion of software and applications and processes. And that requires a lot more people coordination, which is what we're very good at. In terms of the market itself, look, obviously, I think the market has got this wrong. It's our job as a company to continue to prove that. We are putting up, we've had three great quarters in a row. We continue to, I think, put up very good top line numbers for investors. The rest is a little bit up to them. We'll control what we can control. We will help our customers get better, bring AI to them every day and show that in our results. Great. So one of the things that Atlasin has been working over the last year or at least talking a lot about as being about a row. I think which is an AI agent that is specifically designed to work across different knowledge bases for customers and help them with their workflows. workflows. So. What should we be taking out of that rollout and where is it today in terms of the demand cycle and what has been the customer reception? Look, the customer reception has been fantastic. So Rovo is our AI platform. So it enables you to, it includes chat and search. Well class chat and search tools dedicated to the enterprise, enterprise data. That includes everything from tool use to memory to long-running chats, multi-model. It allows you to do a whole host of enterprise controls on your AI work. At the same time Rovo allows you to build agents, it allows you to consume agents from other agent platforms, access the LICN platform with MCP and CLI and other things to take those agentic directions off. We've brought agents into all our applications. So Rovo is a platform, not an app in and of itself. So you can bring agents into your JIRA workflows as native collaborators. You can bring them into your conference whiteboards to enable them to dynamically whiteboard with you. Those agents can come from a lot of different platforms. So think of Rovo as the AI connecting glue. That's all built on top of the teamwork graph. Again, the teamwork graph is an organizational context graph that has been built for many years now. It tracks north of 150 billion objects and connections with 10 billion plus changes happening every single day as customers run those business processes. So we understand how your meetings and your work items and your projects and your messages and your calendar items and your customers and your code and pull requests are changing and how they are connected. We bring people to the entire equation and teams with a weighted graph that understands your organization. That is the most valuable thing that can be provided to any model. The deployment of it is going very, very well. So we've passed 5 million Rovo monthly active users. That's been growing 50% quarter on quarter. Agenteic automation is up five to seven times in the last six months alone. In service collection, it's a massive thing to automate service. We have tens of customers examples that we provide from a salesman's using Rovo across tens of thousands of employees to massively improve the productivity of their product and engineering organizations alongside other tools, which is always important for us. Cisco took 75 different tools and combined them into a single Lassian platform on consolidated workflows in Rovo leading to massive spend reduction and also huge efficiency increases. So 75% of the Fortune 500 are now running Rovo as a part of their Agenteic and AI fabric, I would say. Rovo is a fantastic set of technologies. We're deeply invested. We partner with all of the foundation model companies and other big enterprise companies. Our job is to help surface your content, your workflows, your processes in the right places and make them faster, XLR, your business, which is what Rovo is helping to do. Right. So, Agenteic has a lot of breadth and terms of its offerings. I mean, you cater to the software developer community. Then you also are catering for knowledge workers. And then you also have this IDSM and then there are a lot of collaboration tools. So is your product product breadth really helping you to pick up more customers or maybe add more value to the customers as well as get more out of your customers in terms of the wallet share. And then the second thing I want to understand is given your depth of products and the breadth of the products. Are we at a point where it is becoming a bit of a technical liability or do you think it is a moat? Yes, is the answer the first question there? Every customer I speak to wants to be a deeper strategic partner of Elacene. They want to expand their use of the Elacene platform. They see us as a strategic vendor. So the platform that underlies all our apps and collections is what they are buying into. They're buying into the Elacene platform. They're buying into its connectivity with their other apps. We've always had a deep philosophy that we are not the control center for your organization. We are a very important part of the fabric of software that your organization runs on. We integrate, we partner, we collaborate, we connect to many, many of your other applications, whether that's Microsoft Teams, whether that's Google Docs, whether that's Figma or GitHub or Salesforce. Our job in unleashing the potential of every team, which is our mission, is that team productivity, that team collectiveness that requires your business processes and your workflows to run faster and better. That has always required data exchange with those other products in the AI world that only gets accelerated. The reason people opt into the Elacene platform is the speed you can get it up and running, the updates, the efficiency, the user experience quality, the breadth and depth, and the connectivity of the way that we think about teams. So you listed a couple of great examples there about how knowledge workers, business teams, service teams, and technology or software teams and leadership teams all have to work together. The reason we have organized our apps into collections around those different types of teams is that they actually all have to work with each other. So the team work graph, two of the major announcements we have at team 26, we're adding code to the team work graph and we're adding assets to the team work graph and we're adding a lot of things about people to the team work graph, or have done over the last six months. Why does that matter? Well, the team work graph has long had, for example, in code. It has long had pull requests and repositories in the team work graph. We have now semantically indexed your entire code base. Now this obviously helps a coding agent because coding agents don't deal well with hundreds and hundreds of code repositories or many, many millions of coding files. We can use the team work graph to index that, to make your coding agents more efficient, faster, cheaper and better. And that's what we're seeing in all of the statistics. So bringing the actual code fully semantically indexed at massive scale and connecting it to your people and other objects. But more importantly, we connect it to your business data because what your coding agents need is to be able to understand what's in your Google documents and your customer database as much as they have to understand what's in your code base. Similarly, your business questions asked your business agents increasingly did to understand not just what's in your serum system and what's happening with a given project in terms of your project management tools. They need to understand what's happening in a code base to answer business questions. And so our ability to span across those two areas is incredibly important. Our system of work is a philosophy of how business teams and technology teams should get together to unleash technology driven technology, chord organizations in every industry. We've also added people, the skills, their understanding their teams, the org chart, because that also is incredibly important to be able to answer questions. And lastly, service, whether that's IT service management, whether that's employee service management, whether that's customer service management, your agents in service across any of those fears need to understand your code base. They need to understand your people. They need to understand your org chart. And they need you to understand all of your business data to be able to give the best possible answers. That's what we have in the team work graph as a context and why customers are adopting for the Alassian platform in larger, larger numbers. So I believe Atlassian reaches 80% plus of the Fortune 500 companies out there. And I asked this question, given your depth in terms of the customer reach, and potentially you having the best view of what is happening in the tech sector, especially in terms of the large enterprise spending, I'm curious to know what are you seeing there and what are the things that stand out to you and which might be telling you a story that could be different from the market sentiment. We see a huge amount of customer experience, data and anecdotes across not just Fortune 500 in the US, although we do have I think north of 85% publicly listed as customers. We see it across, you know, a large scale business in Europe and Asia. in Japan, in South America, increasingly. So we do have a massive unique viewpoint into how hundreds of thousands of companies and millions of teams are operating. I would say the headlines in terms of AI adoption and excitement are certainly true. We see that in every customer I go to, they're running a lot of pilots, they're understanding AI, there's a flood of tools and software coming out, and it's very exciting, it could be a bit daunting. They are all accelerating and running extremely fast, all those parts of the truth. What they are struggling with is the, I would say, the talent changes that are required in their business, the way that their business operates, their processes, their workflows. These are harder things to just change quickly. And they are all trying to write, how do you make your organization go from AI novice to AI native, as we sometimes say, that requires more than just buying an agent framework. There's a lot of business re-engineering to do. There's a lot of talent changes. So if I look internally, we have a huge number of skills programs and training programs and sharing programs and brown bag lunches to have staff sharing how they are using AI, our AI through the ROVO platform, but also other, we buy plenty of other AI tools, how they are deploying them, where they are finding advantage, where they are finding challenge. It's a very exciting time for customers in terms of the abilities that they are gaining. It also requires a lot of people changes and fundamentally, like a lot of these changes some time. But there's no doubt that the acceleration of businesses result of the agentic world is happening, but I would say we're still very early in the impact of that. The effect on Atlassian's business is very good, as I said, our AI is one of the best things that's ever happened to our business in terms of being able to solve customer problems in the way that we've done for two decades now. So we're all in and doing very well there. You talked about skill development, but on the other hand, one of the things that we are seeing increasingly, this year has been rebalancing of workforce and I know Atlassian has announced earlier this year, about 10% or so workforce reduction. And then we have had similar signals from larger tech companies. What do you think of this rebalancing is needed, given that you are seeing a growing demand for AI? And what is the key takeaway here? Look, I think it's obviously those are the toughest questions you have to do with the toughest decisions to make, right? We try to be very clear in our announcements, right? We have a constrained environment for investing, and we need to invest far more in certain parts of our organisation, the areas that we are winning. So we're doing really well in enterprise sales. You can see that in our IPO numbers, doing really well in the AI parts of our platform. You can see that in the rovo adoption. And we need to continue to invest in the areas where we are winning, and that requires a change in the team construct that we have as a whole. And the skills mix we need continues to change, right? So we are trying to adapt rapidly to the changing environment, right? And I think the best companies are doing that. I think the bar for what great looks like as a software company, as you see in the public markets, as you commented on, has gone up a lot, right? The bar for what growth rates have to be, what profitability has to look like, and the speed of value creation has continued to raise. So it's up to us to say, which we do say, we're going to meet that bar. And that's how we're trying to continue to build the company and set up for this, you know, decade of AI-driven change. It's very tough that it impacts people right, but it's trying to make the best decision for lasting future and for our customers future ultimately. We have five more minutes left, but then I have two big picture questions before we move on to our closing section. One big picture question is, how are you thinking about the long-term bets, especially, I mean, Atlassian had been bootstrap for many years before going public, and now it's a public company. How does that change your thought process in terms of long-term bets? That's my first question. Okay, long-term bets are very clear actually, right? We continue to grow in the enterprise. That's not just betting on enterprise sales and compliance and scale R&D. It's betting on how we partner and serve the largest companies in the world with their most difficult problems. That part of the business continues to grow very well. I think that's a long-term bet, right? We're investing in those relationships with customers every quarter. That is a long-term bet that makes a lot of sense. Secondly, we continue to invest in a rovo platform and in the teamwork graph and bringing all of the AI partnerships that we have to business workflows, which have a type of team that is. I think that intelligence becoming more of a commodity that context becomes incredibly important, which is what we have in rovo and that connectivity to all of our partners is a really important part. We continue to invest long-term in some more ambitious things. Where browser is doing it incredibly well, dear reports are launching. It's revolutionary way to look at your SaaS applications, for example. I think we have a lot of interesting small bets. You would say that have huge long-term potential, but fundamentally it's about investing in our customers in AI and in a prize and in a broad system of work and bringing those teams together and helping with those business workflows. Those have been our long-term bets for many years and continue to be the same way. We've new technologies to be able to deliver those bets better than ever before. Great. Now that you mentioned about browser, one thing that I need to check with you is that I think increasingly the word is moving towards headless infrastructure. So how are you thinking? I know this week or maybe last week we saw more announcement from some of the major tech companies on the headless infrastructure going into the headless infrastructure wave. So how is that? Lassian thinking about that. Yes, this is a new term. Welcome to the speed in our industry changes. The old school part of me is like APIs. I'm currently important and always have been. Continue to be so we can call them headless if you want. Look, they're incredibly important. We think about as integration between tools. Our platform, the Timorgraph, we've shipped a few things to Tim26. The Timorgraph CLI or the Command Line interface is amazing. It's a completely new way to look at your Timorgraph across meetings and documents and your knowledge across your people, across your physical assets, your services, your code. So there's no doubt that the APIs provided broadly. We have the CLI, we have our MCP server, which is doing really well, allows agents, agentex software, AI models to access our platform and to contribute to our platform, reading and writing both. That is very exciting. We've been there since the start and we'll continue to do that. We've always had strong APIs and strong extensibility. I think that is really, really important. I think the challenge is when people start to say, "Oh, okay, I was going to replace the browser." Dea uses MCP servers to do some of its most amazing work. It is a client talking to backend APIs and the experience, the design layer, the human connectivity, how humans experience that technology, whether that's in Jira, whether that's in a video in Lume, whether that's in the Dea browser, that design and experience is still incredibly important. So the headless era, if you want to call it that, does not mean that the design and experience and the interaction with the software is not important. It just means you can do a lot more work with agents on the API side, which then enables those user experiences to change and be that much more efficient, right? So Dea is perhaps an example of where headless is both true and to say that it's going to be the only way software is consumed is I think very false. Mike, this brings us to our closing section that I call as three straight arrows. Your answers could be one lineers. Are you ready? Sure. All right. So my first question is, what is the one metric that you care about every day when it comes to a classian? Monthly active users. Software needs to be consumed and used to provide value. value I believe very anti-shelfware. If our monthly active users, we can break it down many, many ways. So the Therogo, Mao, Giro Mao, MCP Mao. The active use of our software is a sign of our customers' belief in us, our use of our platform and our strength. Right. So second is, what is the one single most direction where you think you would be able to increase your mode? Is it on the CRM side? Is it on the ITSM side? It is on the context side. We have a massive investment in the Timograph in providing the best context to every team that we surface and providing a context to other agents as well as we are expanding that Timograph to be agent accessible and human accessible. That is fundamentally your knowledge, your people all connected in very unique ways to give you better, faster and cheaper answers for any agentic work. That is the biggest investment that we have in that area. Great. Third and final question here is, what is the one top of your mind book recommendation? One top of my book recommendation. Look, the book I read that's touched me most, which is not a book called Orbital. It's a very short book by, I think it's Samantha Ha, the apologies if I got her name wrong, called Orbital. It's a sort of a meditation on reflections on life as seen from a space station. That sounds ridiculous. It's just a question of apologies, it's just a description. But it truly is one of the most moving and interesting books I've read recently. Sort of a little in the interstellar kind of a vein, but just beautifully written. It's almost like a 150 page poem, I would say, more than a book, actually, but it is actually a book. It's just written in such a beautiful way. Great. I think I'll add that to my reading library. Thanks again for coming on to our podcast. This has been a great discussion and thanks very much.