Atlassian: AI agents in Jira now as full team members

Contents

Atlassian has launched an open beta for "Agents in Jira," integrating AI agents as full team members in projects. Similar to human team members, users can now assign work items to agents based on Atlassian's AI assistant Rovo, as well as third-party agents via the Model Context Protocol (MCP). The tasks then appear in boards, sprints, and releases with status tracking.

As Atlassian reports in its blog, the new feature is designed for three use cases: In addition to direct assignment of work items, users can involve agents via @mention in comments for iterative collaboration, for example, for summaries or research. Agents can also be embedded in workflows, allowing them to automatically perform actions when status changes.

"Work is changing fast: people are now orchestrating across agents, tools, and cross-functional teams. Without clear coordination that can easily turn into chaos," explains Tamar Yehoshua, Chief Product and AI Officer at Atlassian. Existing features like Atlassian Intelligence are primarily chat-based or generative, while the new agents are designed as traceable entities.

Teamwork Graph as Data Basis

The technical foundation for agent integration is Atlassian's Teamwork Graph, a unified data layer that connects work items, people, and knowledge from Jira and Confluence. This makes Rovo agents context-aware, enabling them to recognize task relationships or make recommendations, for example. Administrators define available agents, permissions, and completion criteria, and Jira handles progress tracking.

The agents respect project configurations, permissions, audit trails, and approval flows. Atlassian provides central governance through administrators and adherence to cloud security standards, including OAuth 2.1 and usage logs. The enterprise infrastructure guarantees 99.9 percent availability.

MCP Integration for Open Ecosystem

Furthermore, Atlassian is investing in the Model Context Protocol, an open standard for AI agents to access tools, data, and workflows. Via MCP Skills, Rovo already connects third-party apps such as Amplitude, Box, Canva, Figma, GitHub Copilot, HubSpot, and Intercom. The Rovo MCP Server, which has also reached general availability (GA), allows AI clients like Anthropic's Claude, Cursor, Google Gemini CLI, Lovable, and WRITER to access Jira and Confluence data.

According to Atlassian, enterprise customers account for about 50 percent of total Rovo MCP Server usage, with paying customers making up 93 percent. Approximately 40 percent of monthly active users come from the enterprise segment. The company points to over 350,000 customers worldwide, including over 80 percent of Fortune 500 companies such as NASA, Rivian, Deutsche Bank, United Airlines, and Bosch.

Availability and Limitations

The open beta of Agents in Jira starts immediately. An end date for the beta phase or a date for GA has not been announced. The feature is exclusively available to Jira Cloud users. Customers with on-premises installations (Data Center) will not receive the new features, partly because Atlassian is gradually phasing out support for the server variants.

The cost of using Agents in Jira is included in Premium and Enterprise licenses. Atlassian does not charge separate fees for agent assignments. Rovo agents are included in the licenses; 1000 assisted conversations per month are free, with additional conversations costing $0.30 each.

Rovo MCP Server already works for Confluence. Atlassian indicates plans to expand agent integration to other products in the stack in the future, but there are no official concrete plans for Confluence agents yet.

(fo)