Please welcome Atlassian CEO and co-founder Mike Cannon-Brookes.
Good day, Anaheim, and welcome to Team '26. Thank you. That's a great start. If you look around in this room and online, there are thousands of you, the builders, the operators, and the problem solvers who serve as the heartbeat of the modern organization. Now, we're lucky to be joined today by some of the world's top performing teams, and you'll hear directly from many of them in all the various breakout sessions this week. For 20 years though, our mission has been simple: to unleash the potential of every team. But today, that mission has a new urgency. We're standing at the edge of one of the most significant reimaginings of work in our lifetime. Today isn't just about software. It's about the birth of a new species: the AI native organization. And Atlassian has evolved for just this moment. So you might be asking, what is an AI native organization? Is it just a traditional business with some chatbots that float around? No. It's a fundamental rearchitecture of value where humans move to the critical frontiers while execution is increasingly handled by autonomous agents at scale. At the frontiers, humans do what only humans can. They define intent, they navigate trade-offs, and they resolve ambiguities. Tools and tasks will change, but our purpose remains the same: solving the problems that matter for all your organizations. Now, let's be clear. This transition will have many growing pains. While model performance improves exponentially, many businesses are still stuck in a linear gear. They're piloting or waiting to see. Now, I would argue that's not caution. That is surrender in slow motion. You cannot wait and see your way through an existential shift in technology. The future will belong to the AI native organizations. However, in 2026, that raw intelligence, it's now a commodity. You can literally buy smarts by the token. Therefore, models cannot be your differentiator. The differentiator is your context. That institutional memory of every project, every goal, and every workflow that's finished that only your team understands. Think of this as a simple formula: acceleration for your business is about context multiplied by intelligence. Intelligence is the engine, but context is the fuel. Context is your collective memory. Every decision from every successful and every failed project. If your AI doesn't know what choice you made in 2024, it can't help you win in 2026. We aren't just building tools. Together, we are designing a new anatomy for teamwork. And the best part, if you're an Atlassian customer, you already have this. It's what we call the teamwork graph. It's not a database or a set of files. It is the connective tissue between your work, your people, and your tools. Think of the pulse of your company. You define goals and manage projects in Jira. You design in Figma and you drive growth in HubSpot. You collaborate across dozens of platforms from Zoom to Teams and of course in Confluence pages and whiteboards. Now, every action you take isn't just work. It's the continuous creation of your company's collective context. For years, we've been connecting these dots across your entire ecosystem. What does that formula look like in practice? Well, we operationalize those variables through the Atlassian system of work. It is the connective foundation of how your teams align to goals, plan their work, and unleash their knowledge. It's the how of the modern organization. But a system is only as smart as the data it runs on. That's where the teamwork graph comes in. It is the data fabric for the entire system, the context. And you can multiply that intelligence through the Atlassian AI gateway. The gateway is your secure portal to the world's leading edge AI models. It's the bridge between your private organizational context and that collective intelligence utilizing best of breed models without ever compromising your data security or sovereignty. And then there's Rovo, the unlock, the accelerator for you and your teams. If we double click into Rovo, we see that it's the interface that turns context and intelligence into actual momentum for your business. It brings context AI to every individual in your organization, the builders, the knowledge workers, the leaders. It brings ambient intelligence to every Atlassian app and carries that context far beyond our walls, injecting your organizational memory into all of the apps you use. Context without borders. Today, you aren't just choosing software. You're choosing what kind of company you want to become. Now, we don't promise a world without chaos. Work will always be a little bit messy. That's where the human ingenuity actually lives. And our job is not to sanitize chaos. It is to give your organization a nervous system that can handle it. So your company compounds with every ticket, every line of code and every decision to make the system smarter for the next day. We are building the memory of the company that you are becoming. Now we didn't come to Anaheim for just slides. We came for conversations. Despite all the tech that we're going to talk about, it's still all about the humans. It's about every single one of us in this room. It's about how we collaborate and work with AI to amplify our capabilities and redefine what's possible for our organizations. That's what we mean when we talk about unleashing the potential of every team. Now, this year is going to be pretty different. We're going to have an awful lot of live demos and we're going to break it down through a series of human conversations. First, we're going to talk about how you can grow, build, and deepen your context graph. Then how do you harness and command all of that context across all your Atlassian apps? And finally, how do you surface that context, not just across the Atlassian ecosystem, but across every other app and platform that you use? For the first of those conversations to talk to you about how you grow your context graph, please welcome Chief Product and AI Officer, Tamar.
All right, welcome. Are you ready to have some fun with a few thousand of my closest friends here and do some live demos where absolutely nothing can go wrong?
I am game. In the age of AI, things are moving so fast that I'm very pleased we're doing live demos so that we can show off all the amazing progress that we've made. So Mike talked a lot about context and I think it's one of the most substantial things that we're working on. If you don't know, he is a super fan of the teamwork graph, constantly sending DMs and Loom videos to me and the team like, 'Did you see this? This is super cool.' So why don't we start by you showing one of your favorite examples.
All right. So we thought we'd start with a little comprehensive demo, end-to-end demo of an actual example that I sent to Tamar a little while ago, maybe two, three weeks. So, as I mentioned, we're doing all these demos live. So, a couple ground rules so I don't get any trouble with my legal team later. Firstly, we're going to use all real data. Secondly, we're going to show off our products in some really interesting and new ways. Here we have the new Rovo Chat, continually improving, lots of great stuff. And the first thing I'm going to do is I'm literally going to tell it that I am live on stage at Team '26 and demoing Rovo Chat. And we're going to ask it to remember to take out anything that would be classified as sensitive about a customer. We're going to swap real names with cartoon characters. So that's for PII. And if money comes up for any investors in the audience, we're going to replace all the numbers with like ridiculously hilariously large cartoon numbers. And most importantly, like make us look good. Like try to just give us a little little bit of love. And we can see that that memory has gone in there. So then it quickly... so here is the prompt that I'm going to put in. So I have a customer meeting tomorrow with ServiceRocket. ServiceRocket has been a customer of Atlassian for more than 20 years, a partner of ours. Thank you. And a member of the community for more than two decades.
Shout out to all the community members in the audience. Woo.
A great example of looking at more than a 20-year relationship plus recent interactions. And we can see immediately... we can see that the teamwork graph is showing up. You can see the icon so you know exactly which apps the data is being pulled from.
That's right. And you'll see a lot about the new Rovo Chat's planning, reasoning, skills, tool use. You'll see a lot of examples. You can read some of the background. But first let me go back to those memories while that's working away. We've improved Rovo memory massively in the last six months. Indeed, it now has two different types of memory which I can manage. First, my implicit memory. So it's using the teamwork graph continually every single day, every single thing I've done to learn about me and my job and use that to give me answers that I am looking for. And secondly, it's got explicit memory. So there you can see at the top the memory that I've just added that it's going to save me from any embarrassment on my legal team right here on stage with actual live company data that during any Rovo Chat demos it's going to put in some cartoon characters and replace those numbers.
And with the explicit memories you can delete it. So after he's done you can delete that memory. And Mike, did you know that in my Rovo memory I have an explicit memory for you?
Okay. I told Rovo that if I'm preparing something for Mike, there should always be a TL;DR followed by tons and tons of details.
That sounds like what I would be into. We can see a lot across this trace. We can see lots of the planning, reasoning. You'll see skill uses going past a huge number of new Rovo features which you can see at all different sessions. And as Tamar mentioned, you'll see a lot of data from the new connectors, in this case Salesforce, but also Jira, Confluence, Bitbucket, Teams, etc.
And we've made a ton of enhancements to our connectors. We've had last couple months over 50 enhancements. So for example, Google Drive, we now can index images and the text in images. So it's now searchable along with comments and tables and slides. For Microsoft Teams, we not only index the messages but also the transcripts of your meetings. You mentioned Salesforce. We now index the campaigns, the cases, the context, all with field level controls. GitHub, PRs, deployments, index issues. I could go on and all of this is much faster. So all of the data is now synced real time, near real time in minutes. It will be updated and a full sync is way, way faster. Our goal is to make sure any change is inside the teamwork graph in under 10 minutes. Massive amount of speed improvements. Kudos to all of Tamar and the engineering team across the place. We're now ingesting more than multiple billions of objects every single week into the teamwork graph across all of your organizations to give you the context that you need and lots more admin controls for all the admins out there of all of the different connectors.
So, it looks like we've got a result. Let's go down and see if I zoom in up here. We will see if our first live demo is going to work. So firstly, it says it's prepared by your brilliant CEO. So that's Rovo making me look good. You want it to talk to you in really nice ways. It can do that. Clearly runs the best organized jury boards on the planet. We get a different result every time. So this is fun for us as well. I can see here that... Oh, it looks like it's worked. Okay, so first part of the stress, the memory has worked well. I don't think the annual revenue is probably $347 billion. And I'm pretty sure the account owner is not Foghorn Leghorn. So, memory has worked. Again, kind of a trivial example of Rovo memory to help it sync into your memories. Many, many things that you can do in a real world business case. We can see instantly links like here, I can go to the account, which is the most important thing. And as I mentioned, lots of new rendering options. So, in this case, we've generated a chart or Rovo has for us. Think of it as a real-time BI dashboard of a 20-plus year customer relationship that took data from Salesforce and from Confluence databases and built me a chart in real time of showing me the things that I wanted to see. Multiple charts in fact, lots of other opportunities we have open. Again, memories coming in to help us there. Data you'll see from across Teams, lots of other places, live intelligence all the way down to details of my meetings tomorrow with Scooby-Doo, which is going to be great. It's going to be really exciting. Full stakeholder map of the company and a series of use cases. And then we can see that has come from 61 different sources across the graph in about 3 minutes there, giving me a 20-year customer relationship in one single query.
An amazing example of the breadth of the context of the customer's data that you can access and this saved Mike so much time preparing and there obviously an infinite number of examples you could use with the teamwork graph and this is all available now for all of you. But what I'm really excited to show is how do we build the teamwork graph? Whether you're migrating from data center or you're already in cloud, all of your data is already in the teamwork graph. So let's see how this happens behind the scenes. So let's say you create a Jira work item. It goes into the graph and then you assign it to Bell and now it's linked to Bell in the graph. And then you link the work item into a Confluence page and then all of the people who read, edit, comment on the page are also now linked in the graph. You can add a Loom recording and the transcript of that Loom recording is in the graph and searchable. But work doesn't just happen in Atlassian apps. It happens everywhere. You might add a Figma link to the Confluence page or a Google doc to the Jira work item or you bring in your HR data from our partner Workday or any other system and who you talk to and who you meet with comes through the Microsoft and Google connectors. We've added assets, one of Mike's favorites, to the graph. And something I'm really excited about is code is being added to the graph. So we can index your code and you can ask questions about it. You'll be hearing more about that later. And all of this is helping you grow your graph. And importantly, agents can access everything in your graph. Customers are already running 5 million agent invocations every month. And this is growing very rapidly.
Indeed, something we're super fired up at. Thank you for all of you. 5 million agent accesses each month is an incredible number already. And that's what an AI native organization is going to look like. Agents are showing up everywhere. Indeed, they're across all the surfaces now of the Atlassian platform. They're embedded in your Jira workflows picking up tasks. You can write alongside them. They can write alongside you in Confluence pages and whiteboards. They can get assigned work just like any other teammate. You've put them into many, many millions of your automation rules to handle those repetitive tasks and speed up your organization. You can chat to them in Rovo, but also in Slack, Microsoft Teams, and Google Gemini, wherever it is you hang out. And most importantly, we've made them intuitive to access just like you are familiar with. For example, jumping into a conversation, just @ mention an agent anywhere in any platform surface and it will appear. That is what we mean when we say you're choosing the type of company that you want to become in the AI era. With the Atlassian platform, agents and teammates are already truly integrated across all the surfaces you work on. It's up to you how you use them.
And many of you are building agents in Rovo Studio and you're connecting those agents from your favorite apps to Atlassian via MCP. As you develop more and more agents, it's a living map of how your organization thinks, works, and grows. And that is your enterprise's world model. So Mike, I know assets is one of your favorites. It's a rich new source of data in the graph. Assets for all of you can range from the physical, a spare part or a machine, or to components or services. So you told me recently a great story about the Atlassian Williams F1 team and how they use assets. So why don't you share that with us?
Sure. Look, like 50% of our customers, you out there in the audience, have real world physical objects as a core part of your business. Logistics, manufacturing, building things, cars, trucks, satellites. And 100% of our customers have some form of physical assets. In our case, that's things like meeting rooms and projectors and laptops. Assets allows you to bring those objects, physical real-world objects, into the graph. In Formula 1, every single component matters. If you miss one, you're not going to win a race. Can't do that if you've only got three wheels on the car. So that's what Atlassian Williams have done. They have modeled all of the components of a Formula 1 car as first-class objects in Assets. They're no longer in spreadsheets or in some sort of disconnected system. They now show up in the teamwork graph. But that means they are connected to their work, their people, their code, service requests, projects, and all the knowledge that Atlassian Williams has from their 48-year history. Now, why is that so powerful? Well, between each race, if you don't know about Formula 1, the team's constantly making upgrades. They're constantly changing the car, like upgrading a suspension bracket, connecting one of those wheels. But when they're upgrading that, they want to know, is it worth it to do the upgrade? And can it get done in time for the next race? Now, that's a hardware question, but it's also a knowledge question based on their history. So, one of their race engineers can now ask Rovo exactly that. Rovo will pull in the information from lots of different sources related to that specific part that's stored in Assets across service requests and projects and knowledge bases. Then it makes actual recommendations about prioritizing the upgrade while making certain trade-offs. Now the important part is that answer crossed system and team boundaries. No single person at Atlassian Williams actually had that context, but the teamwork graph did and they accessed it via Rovo.
I bet when you were building the teamwork graph, you never thought about a use case like this.
We certainly did not think about that as a use case. But it's the amazing thing about our job and building platform technologies for all of these amazing customers to build things that we never expected. It's one of the best parts about the job.
I agree. I love it. So the teamwork graph can help your organization keep track of assets and work but also people. This is especially useful for me. I don't spend my days building cars. I spend my days building teams. So many of you may identify with this challenge that I often face. If you're working on a project which is blocked and you need to find the right people with the right skills to put the project back on track. So, a combination of the teamwork graph, the Talent app, and Rovo can help you do this. In our Talent app today, you can connect your HR system to get your org data, but we've taken it further. We've added talent profiles. We now infer skills based on what people actually do. Code commits, Confluence contributions, Jira assignments. You get a full 360 view of what somebody actually works on. So if I see an at-risk project, I can ask Rovo to explore available talent. This invokes a Rovo query which uses the teamwork graph to determine the kind of expertise required to unblock the data. After some reasoning, it gives a list of potential engineers who can help. From there, I can deep dive into their profiles in the Talent app. I can see their focus areas, levels, and skills. Again, this is inferred from actual work. I want to call out that this is both insanely useful for someone like me in a leadership position and something incredibly unique to Atlassian. It's something you can only do on the Atlassian platform because of the connectivity between the work, the people, and the code. So, we've shown you how you can grow the teamwork graph. You can take those connectors. You grow it with every single workflow and process and action that you take across our and other systems. We've added new whole categories of content from people to assets to code. And we're now inferring data on top of the teamwork graph to continue to make you smarter, compounding that value across every tool and every workflow. Thank you to Tamar. That was awesome. Great start for us today. Thank you. All went well.
So now we've shown you how you can continue to grow and expand your context graph. Let's show you how we can harness all of that context that you've built and put it to work in our apps. To do that, let me welcome Sharief, our head of AI and product craft.
Doing good, good. I was listening to you and Tamar backstage.
Always. We're small group. Small group of friends here, so it's all intimate and nice. Yeah.
You know, Mike, I was thinking backstage when you were talking about the connectors with Tamar. When I think about the context that comes from those connectors, we often think about things like the business context, spreadsheets, strategy documents, presentations, and so forth. But if you're an organization that builds and maintains any sort of technology as part of your core business, then actually code is a huge part of your context. And Mike, you always remind me that Atlassian is all about connecting technology teams and business teams together to achieve amazing things. So this provides semantic code intelligence to solve some of the toughest challenges that your teams will face. It means you can now ask questions across thousands of repositories and get answers in an instant. And it's not just simple pattern matching. It actually understands your code, every line of your source code and every file's intent. So Mike, we've been using code search internally at Atlassian for the last few months and I thought this would be a good candidate for our first demo. Should we do it?
Absolutely. It's one of the most amazing and hard technical things that our engineering team's ever built and it's pretty amazing. Let's show it off.
All right, let's jump here. So, I'm going to ask it a question here about our codebase and this is going to use our new code intelligence skill that we have. Now, for a bit of context folks, many years ago, I used to work in Confluence and what often happens every few years still today is that we'll do a design system upgrade and we'll need to revisit all the places in our code that use a particular style and upgrade it. A fairly significant change. So over here I've asked it, hey, tell me all the buttons in our Confluence codebase that don't use the latest style. Who are the people working on it and what are the style guides? So you'll notice here I haven't told it where the Confluence codebase is and I haven't told it what the latest style is either, right? So and just connect me with people so I can understand where work is at today. Now Mike, this is running in our internal codebase which is pretty large, right?
It is indeed. In real time, we're looking across Bitbucket Data Center, Bitbucket Cloud, and GitHub, three places that we store our code. We have more than 11 million code files being searched and 1.5 billion lines of code in the time it takes Sharief to come back with this answer made over 25 years.
Yeah. And what's incredible here, one of the things I want to call out is this is a really good example of me with a business ask or requirement. Hey, we're about to do a design system change. So that tells me, hey, where do we want to go? And I'm asking the code, what do we have today? So, where do we want to go? What do we have today? And this is helping me bridge those two worlds together right here within moments. All right, Mike, we got some results. Let's check it out. I already like it's used an emoji button. Get it button. AI's got some humor sometimes. It's identified the legacy library and the new library. So, I haven't told it where the new library is. It knows where to make the changes. It's got some internal posts on where the new updates are. Oh, and this is quite nice, Mike. It's given me an audit of all the button styles across the codebase and the percentage breakdown of how much they're adopted today. So, a huge piece of information that can help me make a decision on how hard this change is and how broad this change is. It's also got specific areas that I might need to focus on on migration. Oh, and check this out. People, teams that I need to work with and what their roles are on each of that and even the Slack channel so I can contact that team for some more detail. Pretty incredible power there.
It's absolutely unbelievable. Like the amount of work that it has just done to look across one and a half billion lines of code, find the Confluence codebase, find the context, find the design system and write that full document all with clickable links to code if you want to deep dive into them in that 2 minutes is insane.
Yeah, it's pretty cool. Can you give us one more example of what else we can do? So one more. You got one more example. This folks I have to publicly announce might be a career limiting move on my side. Been a good run, man. So, Mike, why don't you tell our small group of close friends here when was the last time you committed code in Atlassian?
I'm guessing that you know that it has been a while. I'm going to guess late 2000s, like 20 odd years.
Okay, close to 20 years. To date us in front of a lot of people. So I thought it'd be fun earlier backstage to ask what to-dos from Mike Cannon-Brookes are still in our codebase and has anyone done anything about them recently and just to make Mike feel young again. I've also asked it to include which year this stuff happened for. And so here's some recent results. Now I actually quite like what it does here because Mike is also sometimes known as MCB, Mike Cannon-Brookes. Sometimes people like to call him Uncle Mike. It's found all the different aliases for Mike. Actually, a really good example of it using the context that is across our knowledge base and our conversations in Slack and all that to pull that in.
People planning and reasoning. Smart, smart way of doing it.
Let's see what it's come up with here. I found two fossil-tier to-dos. Jeez, it's gone a bit emoji wild here today. Just to... we needed the dinosaur there. Thank you. That's when my hair was a different color walking around. All right. Icon on Java to-do make this private. All right, Mike, you got a bit of work to do. I like just to make it easier for us to scan the AI here. It's bolded. This is over 21 years old now. And it's no discussion found about implementing it. Well, let's see how we go with the next one. Public variable survive 21 years. I think it's all right. The next one here is the new code plug macro plugin. So, for those of you who use the Atlassian editor and Confluence and Jira, it has a way to format code in there. This is actually one of our first ever macros in there. And so, Mike was one of the original authors of there. It looks like he had to support Confluence 2.8 as a to-do and still hasn't done it. Which is pretty funny. Now the good news is it's picked up here that that's an old version of Confluence. So, it's actually worked that out. Worked out that that's deprecated. So, for those who don't know, Confluence 2.8, we're talking pre-cloud. In fact, we're talking pre-data center. So, there's before data center even existed as an example of how quickly we can search a 25-plus year codebase.
Yeah, actually Mike, that's a really good point because for those of you out there who are running data center and doing your cloud migrations right now, this is an excellent example of all that knowledge you have in the data center and instances you have that goes to your teamwork graph when you migrate to the cloud. So all that knowledge, the learnings, all of it comes with you when you migrate to the cloud. And just to finish us off here again, it really loves these emojis. Do you want us to raise a ticket, Mike? So you want to finish that off after this? Maybe vibe code it or something.
Let's do that. We'll vibe the answer to that now. All right. Hopefully that's an entertaining example, but you can see the power there.
That's right. I hope you can see in these slightly humorous examples, right? Firstly, I'm off the hook. I think that code is going to be okay. But on a serious business note, searching one and a half billion lines of code with all of your logic over the last 25 years and combining it with people, with context, the insanity of writing those reports in real time. That would have taken some of our senior engineers a few days to go and pull all the information and write it with that amount of clarity. Just amazing what we can do nowadays. And had the business context of knowing what we're trying to do.
So, we've now got... Thank you. So we had your pull requests, we've got your repositories in the teamwork graph. Now we have all the source code that can be semantically indexed and understood in the teamwork graph, broken down, cross-referenced with a world-leading code search engine and understanding the intent of all of those files. What does that mean? It means any AI coding agent that you run will get better quality results. It will get faster results and your CFO will like this, far cheaper results because it will use far less tokens to get the same answers. Any coding agent will benefit from connecting to the teamwork graph. Now coding is only a small part of the software development life cycle and we've been building for software teams for over 20 years. As you can see from some of those examples, AI is rewriting that playbook pretty fast as I'm sure we're all aware. Sharief, you talk to customers every single day. Tell us about how those workflows are shifting, what we're doing for them.
Yeah, Mike, everyone is feeling the change. We've all designed our business processes around one fundamental assumption, and that is code takes time to produce. But if reality is if code is no longer the blocker, then planning and judgment and making sure we are all building the right thing the right way is key. And so at Atlassian, we've been building some experiences to help your teams use the context that you already have across your organization to help your software teams make better plans with AI. Want to jump into an example for this one?
Absolutely. All right, let's do it. So over here I've got my Jira and my team's working and the demo I'm about to give you folks is some of the stuff's already shipped. Some of it we're working in progress. So I'll call that out as we go. And I'm sure my boss here, Uncle Mike, will want to geek out as we go along. All right. So, imagine we work for a financial services company and we're adding a new personal investment dashboard. You know, the kind of dashboard that shows you how my finances are going and all that kind of stuff. We want to add that to our app. Now, unless you're working at a brand new startup, all of us are dealing with code complexity. Tech debt, legacy code, deprecated libraries, maybe some founder code in there as well. And so you don't want to vibe code your change into production here or one-shot it. This is a fairly big change with a bit of risk. So this is a really good example of putting humans in the loop and why that's so critical and this is how the teamwork graph is going to help them. So on my board here I'm going to use the capability we're working on to be able to create an AI plan right from my project in Jira. And I'm going to give it a high level overview of what we're trying to do here. Working on the financial overview feature here in our app to show on the dashboard how my personal finances are going.
Straight to the teamwork graph.
Correct. Yeah. Pulling in all those extra sources of information. Now I haven't given it much business context here. I've said we're just working on this high-level dashboard and it's pulling in the additional context from the conversations and sources here, Confluence pages, Google Drive documents, GitHub pull requests and so forth. And as it keeps going here, it is going to invoke the code intelligence skill that we just saw earlier.
Again, all those skills can be invoked either with the slash command if you want to call them out explicitly or you can include them in agents you build in Studio, all reusable automation rules or here it's inferred automatically that it needed to use the code intelligence skill.
Yeah. And it's found this is a fairly significant change. It looks like six repositories it'll go through. Oh, and it's asked me for a question I need to answer here. Demoing another new feature of Rovo. So, your AI is pretty dang smart, but sometimes it reaches a point where it thinks, am I going to go down this path or that path? And if that's close to like a 50/50 call, for example, now Rovo will ask you rather than going a long way down the wrong path. So, you can keep guiding it and steering it and helping it go in the right direction. In this case, it's asked a really important question when it's understood the business context and the code. Yeah, it looks like there's two ways to store data in this app and it's referred to the code to understand that, but it's also referred to the business context on what the preferred way is. And when you're doing a demo in front of thousands of people, always go with AI suggestion. So, we'll just go ahead and do that. So, it's given me a plan here. So, it's created a proposed technical architecture plan of how to implement this change. I can see the files, the repositories it's proposing to change and it's giving me sources and references so I can make sure it's doing the right thing. I'll go in and modify them if I need to and it's got all the requirements for the change and this part here Mike, my favorite part. So get this, it found a Confluence whiteboard that someone in my team has previously created for the architecture of this app. It then copied it, used the context of the change in the teamwork graph to propose the new architecture. That is insane. It is very, very cool. And of course it's about human and AI collaboration. It's given you a start. Sometimes it'll get it totally right. Other times it'll give you the start. Maybe you've just a hint that you need to update the architecture diagram that otherwise would have been a legacy diagram and been old. In this case, it's actually added a sticky note to tell you what it has actually done. Yeah, nice little change log feature there. Now, this is a good example, folks, of a feature that you actually all have today. You can go to your existing Confluence whiteboards and give it some additional context from anywhere in your graph. Ask it to update it based on that or copy someone else's whiteboard with your new project proposal and see what it comes up with there. So, that's a good example there of stuff you already have today. Now, it keeps going here and recommending all the files that it needs to change and additional sources for that.
And gives you a little hint if you look at the top of how much it estimates that this will cost in tokens.
Yeah. And it's using that based on the code intelligence capability that it has and it knows what models we use. So it's giving me an estimate of how big this change is. All right, let's accept this plan. And now it's asking do we want to break down the work into smaller chunks for our team. So we'll go ahead and do that. Now this particular part of the demo uses the intelligent work breakdown skill that you all have today except it's using that with a new code intelligence skill that we demoed earlier. So what it's doing here, it's recommended a bunch of tasks for me to do. It probably knows that I often do the easier ones. Some tasks for Claude Code and a task for a custom agent. So it's using all the heuristics that it has to recommend work items for specific people, teams, and agents to pick up next. So we'll go ahead and create them on my backlog. And now we're off to the races. So we've broken down a pretty high level chunk of work into smaller, more deliverable tasks. And because our team has already got agents orchestrated in Jira with automation, Claude Code automatically picks up the tasks here and gets to work on them straight away and I'll pass it back to my team members in review when it's done. It's a great example of using agents, automations, workflow steps all together in a way that works for your teams to get you moving as quickly as possible. And let you orchestrate your agent workflows right there from Jira.
So to give you a quick recap of what we've just seen because that was an awful lot in about four or five minutes. We showed an example of how Rovo used the knowledge context, the documents, the architecture diagrams, the text that you have written alongside the people context and the code context and used that understanding to build a plan. We can help and edit it, steer the plan, make sure it's correct. Turn that plan...
Everywhere into a set of work items to update our architecture diagrams all the way through starting the execution with automations and agents and workflows in a couple minutes. And that is what makes Jira amazing in the agentic era. It is your AI control plane across both agents and human workflows. It puts them together in ways that make sense for your specific team. It understands what's happening and helps orchestrate those complex multiplayer multi-agent environments that we're all heading into. And Mike, we're really pleased to announce that agents in Jira is also now generally available today.
So today you can connect any agent to Jira using MCP. We also have the GitHub Copilot agent available for you out of the box. And in the coming weeks we're working with our partners Claude Code, Cursor, and OpenAI's Codex to get them implemented as well. So, let's just watch a recap of everything we've just seen so far.
Thank you. And like everything we're showing you today, the teamwork graph has been working behind the scenes on your context. Every workflow, every decision compounds the value of your Atlassian platform. The plan, the code, the specs, the work items, the architecture diagrams, all got written back into the teamwork graph. What that means is the teamwork graph is now smarter than it was at the start of that demo and will continue to compound over time. So, we talked a lot about how it's powering software teams. We're seeing more and more knowledge workers and business teams accelerating on the Atlassian platform with that context. Marketing teams, HR teams, finance teams. So, probably time we hop across to those.
Yeah, Mike. And it's true for Atlassian. We've seen all sorts of teams run many use cases with Rovo that are quite inspiring. So, today I want to walk you through an example of a sales enablement team that do some of these workflows today. Let's do it.
All right, let's do it. Now, this one, folks, takes some time to run. So, we pre-recorded it yesterday for you. So, let me run through it. So, picture this. I run a sales enablement team and we just had our offsite and we went through a whole lot of documents during the offsite. Lots of decisions, spreadsheets. Here are some of them. And we need to collect all this information, a week's worth of work into something that's digestible that the rest of our teams can understand. So, I'm going to ask Rovo to do this for me. I'm going to ask it to pull together a Confluence whiteboard. And in this prompt, I'm going to ask it to summarize the Q3 planning offsite we just had with my team and I will give it some of my preferences. And pro tip, by the way, folks, if you always want to look good in front of any manager, just say you also want it in a 2x2, they love that stuff. So, notice I haven't told it where to look for the Q3 offsite, it'll just work that out. So, it's getting to work here and you can see Rovo in action processing and searching the team to understand the prompt I gave it and work out the context. And it starts streaming live on the dashboard. It's grouping our strategic priorities into that awesome 2x2 grid there. And then next up, it'll look at all the decisions we've made, give me a summary of them, and one of my favorite things, if there was a meeting recording during that time, it'll embed it right in line in addition with the Google documents that we referenced at that time. So, it's incredible amount of power here, summarizing all that work for us. And now it's actually streaming some of the charts we made at the offsite with the new Confluence remix feature. So it's got all like quarterly look back with the numbers there. So all of this happened within seconds and the artifact is immediately useful to the whole team.
Now Sharief asked for a Confluence whiteboard there but you could have asked this for a Confluence page. If you wanted a more textual format it would have done the same thing with all the remixes, all the links etc. Rovo will tap into the teamwork graph wherever it is that your teams like to work and whatever type of content they want to consume. Yeah, you can even ask it to create a Google document with this or my recent favorite, ask it to populate a Confluence database for you. It does an incredible job. All right, let's switch gears here a bit. This sales enablement team tracks all their work in Jira and they do their work with their human teammates as well as multiple agents. So, some of these agents they work with are the Canva agent for managing creative assets, the Gamma agent for creating beautiful decks and presentations, as well as a custom brand guidelines agent that they've created themselves. Now, they've designed their workflow to enable them to orchestrate their work with agents. So, to do this, they simply drag and drop the cards into the steps of the workflow and the agents automatically get to work. So Mike, you know all those demos that a lot of people are seeing of engineers with like lots of consoles open. I'm going to fire off an agent here, fire off an agent here. Has that one finished and come back to them?
Yeah. All of you have this today in your Jira boards. It's an incredible power that available to all of your teams. So let's dive into a deeper example of how this works. So this team is getting ready for their revenue kickoff. They call it RKO. So RKO26 is coming up and they need to draft a deck that has all the opening remarks. So they simply drag the work item into the status for the Gamma agent and the agent gets going. Now on this work item, we've added a bit of context here, the description of what it needs to use as its context, the design guidelines we'd like to use for the presentation. We've even attached an image we wanted to use so it's consistent with the visual look and feel. And the Gamma agent has started to work. Just the agent takes all that context and then within moments it'll come back with a plan and then I can accept it. I can modify it. I can go back and forth with it if you like and then you can confirm it when you're ready. Now at any time I can get out of this work item and go check on the progress of other agents. Now if I ever want to come back to it it's as simple as tapping on the agent like I would just chat to another teammate. You see, agent assignment in Jira creates another Rovo chat session which you can resume anytime. So if I'm using Rovo mobile on my phone, I can continue that agent session from wherever I am. Could be on the bus, could be in a taxi, could be on a plane, I can keep it going and I can keep it going. Same with Rovo on the desktop or the Jira Confluence mobile apps that also have Rovo embedded in them.
Yep. Switch apps, go to Confluence, keep chatting to another agent, do something else, come back. The sessions are always there with you.
So it looks like Gamma has now finished with this new deck. Let's take a look at it and we'll open it up and it's got that image that we asked it to use from the Jira item. It's followed the design guidelines and used the information we've attached on that Jira item as context there. So as you can see agents in Jira are an incredibly powerful tool for supercharging all types of teamwork, not just technical teams. Again, amazing examples of agents in Jira live for all of you today. And that was recorded last night on actual live production data.
