During Team ‘26, Atlassian made one thing very clear: The next stage of enterprise AI will not be won by the company with the most chatbots or agents. It will be won by the teams that can efficiently connect AI to the way work actually happens, and that’s precisely the principle of the Teamwork Graph, with all the AI surrounding them to help it grow. Get to know all the announcements during one of the most important annual events of Atlassian: Team '26.
The founder's keynote is usually the main stage where most announcements are made. Mike Cannon-Brooke, Atlassian co-founder and CEO, presented in a very unusual way: a live demo. He showed how the change will look for enterprise work around the idea of an AI native organization. This is not a traditional business with artificial intelligence added on top. Instead, it’s an organization where humans define intent, make judgment calls, and resolve details, ambiguity, hallucinations, and other defects. Meanwhile, agents help execute the work across tools, teams, and workflows.
It might seem to be a very ambitious vision, and that’s not new to Atlassian, given the announcements shared at Team '26 show how Atlassian is planning to translate this vision into something practical:
Mike Cannon-Brooke announcing the launch of the Teamwork Graph
The Teamwork Graph is Atlassian’s context layer that connects people, goals, projects, documents, code, decisions, and tools. As it's from now officially launched; following its announcement a few months ago, it includes more than 150 billion connections, giving humans and AI Agents a shared understanding of how work moves across the organization.
For those teams using Jira, Confluence, Loom, Rovo, and other Atlassian apps, this is more than a technical update. This is signaling what’s the path Atlassian System of Work is heading: Toward an AI system that doesn’t just answer questions but understand context, takes action, and helps teams to move forward from planning to delivery with less friction.
The central message of the speech across the main keynote and all the other keynotes, sessions, and workshops is that AI models alone are no longer enough to create a competitive edge, given that all of them are built the same, with the same capabilities for everyone. As Atlassian’s CEO explained , intelligence can now be accessed on demand. What matters more is the institutional memory and knowledge behind the work: the decisions made, the goals defined, the failed attempts, the project's history, and then the relationships between all of them. And here’s where precisely the Teamwork Graph stands out.
It’s not just another database. Instead, it’s a map of work across Atlassian apps and all other connected SaaS apps, including the obvious, such as Jira, Confluence, Loom, Figma, GitHub, Slack, Salesforce, Workday, and others. Atlassian describes it as the connective tissue between people, work, goals, tools, and decisions.
Why these announcements matter for project managers, PMOs, product teams, and business leaders?
These announcements matter for the overall PMO because AI without context creates more noise. As simple as that. It can summarize the wrong document, miss a dependency, ignore ownership, or generate work that doesn’t match the team’s priorities. On the other side, AI with context has a better chance of understanding why a decision was made, who owns the next step, what has already been tried, and where the work fits into broader goals. And that’s what makes the Teamwork Graph different. Atlassian is opening beyond just Atlassian apps.
It was announced that the Teamwork Graph CLI, a BETA, gives developers, admins, and coding agents a command-line way to access context from the graph. The CLI is designed for technical users and agent workflows to help AI tools query work items, document ownership, relationships, and extra content without relying on disconnected product APIs.
It was also introduced the Teamwork Graph tools in Rovo MCP Server, also a BETA. This matters because the Model Context Protocol (MCP) gives compatible AI assistants and agents a standard to connect with external tools and data. This is important because Atlassian is opening this context layer so agents across the stack can act with a better understanding, ownership, history, dependencies, and permissions.
For the PMO and project leaders this could mean fewer isolated AI experiments and more governed and connected workflows. For technical team, agents can work closer to the actual software delivery lifecycle. For admins, it also raises the importance of permissions, governance, and data hygiene. And here’s where’s precisely interesting starting to implementing a solution built in Jira for your projects, and the PMO Collection for Jira, alongside the PMO Solution, presents an easy path to adopt and adapt to this new connected paradigma.
The PMO Collection it's built to keep your project management connected in Jira
As another central piece of the whole event, Rovo was reintroduced as the interface that turns context and intelligence into action. Officially, Rovo is already used by more than 90% of Atlassian enterprise cloud customers and by 75% of the Fortune 500, making this announcement stand out, especially because of the new Rovo Studio experience, which is now generally available.
There was also introduced Max, a new reasoning mode in Rovo Chat. Max functions by breaking down complex requests into multi-step plans, then executing tasks across various tools and returning the results for team review and further development. It's advisable to watch the live demo during the founder's keynote, as the presentation was engaging. The value of Max will depend on how reliably it can connect those pieces without creating more manual review for the team. Watch the whole founder's keynote:
For Jira users, one of the clearest announcements was the general availability of Agents in Jira. As we mentioned in the title, it’s useful for both human and agent workflows. The idea is simple but relevant: agents should not operate in a black box. They should be visible in the same system where teams already plan, track, assign, and govern work.
This new experience has been designed as a unified workspace where teams can build agents, automations, and apps grounded in the Teamwork Graph. The important part is that Atlassian is not limiting this to just developers.
These Agents in Jira allows teams to assign work to Rovo and third-party agents, to interact with Agents in comments, and bring them into workflows, as well as the auditability and admin control part of this model, matters for enterprise adoption. As part of a broader shift from individual prompting to collaborative and traceable agent work inside Jira, GitHub Copilot, Claude Code, Cursor, and OpenAI Codex are underway.
The collaboration across these two apps, named Confluence Remix with Rovo, is another BETA that allows teams to transform written content into visual formats such as charts, timelines, maps charts, quadrants, and more. Besides the slides, it will allow for generating presentations from Confluence content while using the Teamwork Graph.
The Agent briefings in Loom will allow teams to explain requirements, budgets, design, or feedback to the app, and it will translate that multimodal context into a structured prompt or action plan that can be converted into Jira work items. Here you have a sneak peek:
Remember at the end of last year, during our webinar introducing the PMO Collection for Jira?
Yes, in that same webinar, Atlassian was giving hints about the now officially new Product Collection that's built around Jira Product Discovery, Feedback (a new app), Rovo, and product analytic integrations. It’s now confirmed to be in early access with Pendo as the first product analytic integration.
Why does it make sense to make this collection public now? It’s a matter of timing, as AI accelates engineering output, bottleneck shift, and what to build becomes harder, and this collection is the answer to that need. 
The new Feedback app captures customer input from sources such as support tickets, sales calls, CRM records, Slack, etc, and uses AI to organize them into actionable insights which can flow into Jira Product Discovery, helping teams to connect customer signals into prioritization and delivery, faster.
The Enterprise edition of Jira Product Discovery is expected to be available in the coming months.
Atlassian announced several capabilities for engineering organizations with DX, navigating AI-native software delivery, such as AI Code Insights, Agent Experience, and Pulse. All of these together are meant to help to understand where AI-generated code is being used, how agents are performing, and to check if investments in AI are bringing real outcomes.
This new browser, from Atlassian’s The Browser Company, shows how Atlassian is heavily focusing on AI (as if everything we have mentioned so far hasn’t, right?). It’s thought for those people who use many tabs and tools. And as you can see in the live demo, during the keynote, it’s possible to proactively get daily summary briefs pulled from the Calendar, Slack, Team, open tabs, and of course, with the help of the Teamwork Graph. Its enterprise version might include Guard.
Live announcement of the Dia browser during Atlassian Team '26
If Atlassian gets to pull this one out, they would be creating an extra layer to be included in the System of Work, connected with the Teamwork Graph, in a secure, controlled, and governed way.
Overall, this was a Team very focused on AI announcements and showing different ways to connect this ecosystem of tools. For Jira users, this means Agents will be the new paradigm. For Confluence, knowledge will be easier to work with. For product teams, the feedback and decisions are directly connected to prioritization and delivery. For engineers, the oversight of AI and its outcome will bring better conditions. And overall, it aligns with Atlassian’s vision, creating an operating layer for human-AI connected work, and the success of that vision will depend on execution, adoption, governance, and, of course, the quality of work, alongside this “new” teammate named artificial intelligence.
The architecture structuring this whole connection of tools was the biggest announcement during this team: The Teamwork Graph, which he has heard about for a few months now, is now a reality. This is a context layer with AI as an interface, and the future is not very far from the present. The outcomes depend on the context you’re handling today, making teams work with the best processes, context, and structure to move faster, smarter, and overall, work better together.
No Comments Yet
Let us know what you think