People talk a lot about Rovo and Atlassian's AI, but rarely about what makes them possible. Beneath the surface lies a foundational, quiet and decisive layer: the Teamwork Graph. Understanding this tool means understanding why Atlassian's AI can answer questions that your tools, taken in isolation, could never handle.
In a typical organization, work and knowledge are split into silos: tickets in Jira, documentation in Confluence, code in Bitbucket or GitHub, conversations in Slack or Teams, files in Google Drive or SharePoint, designs in Figma. Each of these tools knows its own content, but none knows the links between them.
The result: information exists, but context is lost. A Jira ticket doesn't "know" it's linked to a decision documented in Confluence, that it produced a pull request in the code repository, or that it advances a company goal. That context lives in people's heads — and vanishes with them. The Teamwork Graph is Atlassian's answer to this fragmentation.
The Teamwork Graph is a graph-shaped representation of all of an organization's work: work items, knowledge, people, teams, goals — and above all the relationships that connect them. It isn't a classic database of rows and columns, but a network of nodes and links, where value lies as much in the connections as in the data itself.
This graph structure isn't a technical detail: it's precisely what lets a machine understand context. Knowing that item A is linked to B, which depends on C, which belongs to team D and contributes to goal E — that web of relationships is what turns isolated data into usable knowledge.
The graph rests on two pillars: entities (the nodes) and relationships (the links, or edges). The Teamwork Graph makes these a living map of your organization.
An entity is an identifiable object from the world of work. The Teamwork Graph distinguishes many types:
This is where the real value lies. A relationship describes how two entities are connected: a ticket "implements" a goal, a page "documents" a project, a pull request "resolves" a ticket, a person "owns" a team, a ticket "blocks" another. These links, captured automatically or declared explicitly, allow you to evaluate a global context rather than mere keywords.
A database provides reference values. The Teamwork Graph shows you how these values are related within your context. This is the missing context—and what AI leverages.
A graph limited to Atlassian products alone would have partial value, because real work always spills beyond a single vendor. The strength of the Teamwork Graph is its ability to integrate external sources through connectors.
In practice, the Teamwork Graph can index and relate data from third-party tools — external code repositories, office suites, team messaging, design tools, and many others. The goal is to unify work context wherever it lives, rather than forcing the whole organization into a single tool.
This openness also has a major practical consequence: the quality of the graph — and therefore, of everything built on it — depends on the sources connected to it and their condition. Connecting a messy source inevitably injects that mess (confusion) into the Teamwork Graph.
Indexing entities and links isn't enough; you also need to understand their meaning. The Teamwork Graph adds a semantic layer that interprets content and establishes matches even when the words differ.
This is what lets a search understand that a "payment outage" and a "billing incident" might be the same thing, or relate a goal framed by leadership to the concrete work of the teams contributing to it. This semantic interpretation, combined with the graph structure, is the foundation on which AI can reason — not merely search for words.
Here's the most important aspect from a security standpoint, and the most often misunderstood. The Teamwork Graph is designed to respect each entity's original permissions. Information is never supposed to surface to someone who didn't already have access to it in the source tool.
In other words, the graph doesn't create new access rights: it inherits the existing ones. That's an essential guarantee — but it has a fearsome corollary. If your permissions are misconfigured at the source (an overly open Confluence space, a Jira project visible to everyone), the graph will faithfully propagate that mistake, and AI may expose, with full "technical legitimacy", information that should never have been broadly accessible.
The Teamwork Graph is not leaking sensitive information: it simply reveals the shortcomings of your governance with regard to the management of your permissions, which no one was aware of.
The Teamwork Graph isn't an end in itself: it's the foundation on which Atlassian's AI capabilities — grouped under Rovo — rest.
None of these capabilities would be reliable without the Teamwork Graph. An agent unaware of the links between a ticket, its documented decision and its delivered code can only offer fragmented answers; grounded in the Teamwork Graph, it offers reasoning grounded in real-world context.
Atlassian frames the Teamwork Graph within a broader vision it calls the "System of Work": the idea that work, knowledge, people and goals form a single, connected system, rather than a collection of juxtaposed tools. The graph is its infrastructure — the invisible plumbing that holds the whole together.
For an organization, the stakes are strategic: it's the difference between a suite of apps used one at a time, and a platform that understands how work actually fits together. The first answers questions about a tool; the second answers questions about the organization.
Understanding the graph also means understanding what it demands. Three implications deserve immediate attention.
The good news is that preparing for the Teamwork Graph is the same as good Atlassian hygiene — work that pays off regardless of AI.
The Teamwork Graph is one of the deepest evolutions of the Atlassian platform, because it shifts value from individual tools to the relationships between them. But that value isn't automatically won: it's earned through data quality, permission rigor and structural clarity.
At Nimbax we prepare Quebec organizations to leverage this layer by working the foundations first: governance, content hygiene, permission audits and compliance. The Teamwork Graph is powerful and our role is to ensure that you unlock its full potential.
