What Is the Atlassian Teamwork Graph? Features & How Rovo Uses It

Enterprise knowledge has never been more fragmented. Teams work across Jira, Confluence, Slack, Microsoft Teams, GitHub, Salesforce, Google Workspace, and dozens of other applications, creating information at a pace that far exceeds their ability to organize it. The result is a familiar challenge: critical context exists somewhere, but finding it—and understanding how it connects to other work—remains difficult.

To address this problem, Atlassian introduced the Teamwork Graph, a foundational intelligence layer that connects people, knowledge, goals, projects, and work across the Atlassian platform and integrated third-party tools. Rather than storing information in isolated systems, the Teamwork Graph creates a unified network of relationships that helps teams discover context, surface relevant knowledge, and improve collaboration at scale.

The importance of the Teamwork Graph has grown significantly with the rise of Atlassian Rovo and AI-powered experiences. Rovo relies on the Teamwork Graph to understand organizational context, connect information across applications, and deliver more relevant search results, recommendations, and AI assistance. In many ways, the Teamwork Graph has become the backbone of Atlassian’s vision for AI-enabled teamwork.

Let’s take a closer look at how the Teamwork Graph works, what makes it valuable, and why it has become such an important part of Atlassian’s AI strategy.

What Is the Atlassian Teamwork Graph?

The Atlassian Teamwork Graph is a unified data layer that connects your work, people, goals, code, and knowledge across Atlassian and 100+ connected apps. It is not a dashboard you look at. It is not a report or a visualization tool. It works quietly underneath the surface, powering the search results, AI recommendations, agent decisions, and workflow automations that users experience every day.

Think of it as a living map of your enterprise. It understands how Jira issues connect to epics, how those epics support business objectives, who owns those objectives, and where the supporting documentation lives. By continuously mapping these relationships, the Teamwork Graph provides the context needed for search, automation, analytics, and AI-powered experiences.

This connected understanding is what makes Atlassian’s AI capabilities possible.

The scale of the Teamwork Graph reflects the complexity of modern organizations. Today, it connects more than 150 billion objects and relationships across the Atlassian ecosystem, with billions of new data points added every week. Changes made in connected systems are continuously reflected in the graph, ensuring that the context powering search, recommendations, and AI experiences remains current.

It connects across:

• Atlassian products: Jira, Confluence, Jira Service Management, Loom, and Bitbucket
• 75+ third-party tools including GitHub, Google Drive, Slack, Figma, Salesforce, Notion, Microsoft Teams, and SharePoint
• Custom sources via Forge connectors, more on that in the Team ’26 section

Everything in Rovo, from search to chat to agents, runs on top of this graph. This is what makes it the most important piece of infrastructure in the entire Atlassian platform.

Atlassian Teamwork Graph Features: What It Can Actually Do

The core Atlassian Teamwork Graph features are not separate products you turn on. They are capabilities baked into the graph itself that power every AI experience across Atlassian. Here is what each one does.

Unified Context Across All Your Tools

The graph collects work items, pages, ideas, service requests, projects, goals, code, and messages from every connected app and brings them into one shared data model.

• Every update in one tool reflects across the graph automatically.
• A Jira ticket change, a Confluence page edit, a new Loom recording- all of it compounds the graph’s knowledge in real time.
• Teams get the full picture of any project without switching tabs or asking someone to forward a link.

This is the foundation of Teamwork Graph. Everything else the graph does depends on this connected layer being accurate and up to date.

Semantic Search Across Every Connected Tool

The graph powers Rovo Search with meaning-based queries. It understands intent, not just the exact words typed into a search bar.

• Ask “what blocked the Q3 payments release?” and get results pulled from Jira issues, Confluence post-incident reviews, GitHub commits, and Slack threads all at once.
• Results are ranked by relevance to your actual work, not by recency or keyword frequency alone.
• Permissions are fully respected throughout. Everyone only sees content they are already authorized to access in the original source tools.

This is fundamentally different from typing the same query into five different tools and manually comparing the results.

Intelligent Cross-App Recommendations

The graph does not just retrieve information when you ask for it. It surfaces relevant context before you ask.

• While working on a Jira epic, it can recommend the Confluence doc most connected to that work.
• While reviewing an incident in JSM, it surfaces past similar incidents and the decisions made at the time.
• It suggests relevant teammates based on who has contributed to similar work before.

These recommendations get sharper over time as the graph accumulates more context about how your organization works.

Real-Time Context That Compounds Over Time

The graph is not a static snapshot. It is continuously updated and self-improving.

• Changes propagate within 10 minutes of being made in any connected tool.
• Multiple billions of new objects are ingested every single week.
• Every Jira update, pasted link, and connected integration adds more signal to the map.
• The longer it runs, the richer and more accurate the context becomes.

This compounding effect is what separates the Teamwork Graph from a simple data warehouse or search index.

AI Agent Context and Awareness

This is where the graph’s value becomes most visible to end users.

• Every Rovo agent runs on the Teamwork Graph. Without it, agents would have no idea what your organization is working on, who owns what, what decisions have been made, or how different projects relate to each other.
• With the graph, agents can act like informed teammates. They know which team owns a service, what Jira tickets are open against it, what the on-call runbook says, and who made the last architectural decision about it.
• This is what makes Agents in Jira work as genuine teammates rather than generic bots responding to keywords.

Security and Permission Enforcement

The graph handles security at a level most people do not realize is running in the background.

• It runs last-mile permission checks before any data reaches a user or an agent.
• Data from connected tools enters the graph with its original access controls fully intact.
• Users only see what they are entitled to see, regardless of which Atlassian product or AI feature they are using.
• Built for enterprise-grade compliance from the start, not added as an afterthought.

How Rovo Uses the Teamwork Graph

The simplest way to put it: Rovo is the AI layer that sits on top of the Teamwork Graph. The graph is the memory. Rovo is how you put that memory to work.

Every single Rovo product draws directly from the graph:

• Rovo Search queries the full graph to return semantic results across all connected tools. When you search, you are not searching individual apps. You are searching the graph, which has already connected all of them.
• Rovo Chat answers questions using the graph as its knowledge base. It does not pull from the public internet. It pulls from your organization’s actual work data, your Jira tickets, your Confluence docs, your decisions, your goals. That is why it can answer “what is blocking the mobile redesign launch?” with a specific, grounded answer rather than a generic one.
• Rovo Agents execute real tasks using full graph awareness. They know who owns a project, what the related Confluence docs say, which Jira tickets are in progress, and what decisions have already been made. That full context is what allows an agent to act rather than just respond.
• Rovo Dev understands codebases at an architecture level by combining source code analysis with Jira and Confluence context pulled from the graph. When a developer asks, “Where is the authentication logic defined?”, Rovo Dev does not just search filenames. It understands the system and can answer precisely because it has the full organizational context behind the code.

Without the Teamwork Graph, Rovo would be a capable but generic AI tool, similar to any other chatbot or search product. With the graph, it becomes something specific to your organization. It knows your work, your people, your history, and your context.

That is the difference.

What Team ’26 Changed for the Teamwork Graph

Before Team ’26, the Teamwork Graph powered Rovo and Atlassian products entirely behind the scenes. Developers and external AI tools had no direct access to it.

At Team ’26, held from May 5–7, 2026, at the Anaheim Convention Center, Atlassian opened the Teamwork Graph as a programmable platform for the first time. Four major capabilities were introduced:

Teamwork Graph CLI (Open Beta)

A command-line interface built for AI agents and developers.

• 300+ commands spanning 380+ tools
• Works with Claude Code, Cursor, and OpenAI Codex out of the box
• Read and write support with single authentication across all connected tools

Teamwork Graph Tools in Rovo MCP Server (Open Beta)

Two new tools – getTeamworkGraphContext and getTeamworkGraphObject – let any MCP-compatible AI client access your live Atlassian context.

• Works with Figma, Replit, ChatGPT, and Claude
• External agents can now reason and act using your actual Jira, Confluence, and connected app data

Teamwork Graph Connectors via Forge (General Availability)

Any developer or partner can now build custom connectors for proprietary systems, legacy platforms, or internal tools.

• Permissions from the source system are preserved
• Connected data automatically lights up in Rovo and Atlassian Analytics

TeamworkGraph.com launched a new destination for admins and developers to visualize and manage their organization’s graph data.

Pricing: The CLI and MCP Server tools are free today. Future pricing will be tied to Rovo credits, with 90 days’ advance notice before billing starts.

Implementing the Teamwork Graph Successfully

The Teamwork Graph is designed to connect information across your organization, but its effectiveness depends on the quality of the data and systems connected to it. Simply enabling connectors is not enough. Organizations need a clear strategy for data access, permissions, governance, and ongoing maintenance.

Before deploying the Teamwork Graph at scale, consider the following:

• Connector coverage: Ensure critical systems such as Google Drive, Slack, Microsoft Teams, GitHub, Salesforce, SharePoint, and internal knowledge repositories are connected.
• Permission management: Verify that access controls are configured correctly so that users and AI tools see only the information they are authorized to access.
• Data quality: Remove outdated, duplicate, or poorly organized content that could reduce the accuracy of search results and AI responses.
• Cloud readiness: Many Teamwork Graph capabilities are optimized for Atlassian Cloud, making cloud migration an important step for organizations still running Server or Data Center deployments.
• AI governance: Establish policies for AI usage, data access, compliance requirements, and auditing before rolling out AI-powered experiences across the organization.

Organizations that invest in these foundations are more likely to see accurate search results, reliable AI recommendations, and greater user trust in tools such as Atlassian Rovo.

Final Thoughts

The Teamwork Graph is not a feature of Atlassian. It is the foundation that makes every AI feature in Atlassian actually work. Without it, Rovo Search returns keyword matches. Rovo Chat gives generic answers. Rovo Agents act without context. With it, every AI experience in the platform becomes specific, grounded, and genuinely useful to the people using it.

Team ’26 took that foundation and opened it up. Any developer, any AI agent, and any organization can now build on top of 150 billion connections’ worth of organizational context.

That’s a significant shift in what is possible.

The teams that set this up properly and connect the right data sources from the start will get the most out of every AI investment they make on top of it.

Frequently Asked Questions