Atlassian brings AI agents into Jira with open beta launch

Atlassian has launched an open beta of agents in Jira, allowing teams to assign work to AI agents, collaborate through comments, and embed them into workflows. The company is also expanding its support for Model Context Protocol to strengthen its open AI ecosystem for enterprise customers.

The new capability allows teams to assign tasks to Atlassian Rovo agents and third-party agents directly within Jira. Instead of running AI experiments in isolated tools, organisations can now manage agent-driven work in the same place they track projects, bugs, and releases.

Tamar Yehoshua, Chief Product and AI Officer at Atlassian, said the nature of work is shifting.

“Work is changing fast: people are now orchestrating across agents, tools, and cross-functional teams. Without clear coordination that can easily turn into chaos,” Yehoshua said. “We’re focused on helping teams turn that complexity into real productivity.”

Agents become part of the Jira workflow

With agents in Jira, teams can:

• Assign work to Atlassian Rovo agents and Model Context Protocol (MCP)-enabled third-party agents.

• @mention agents in comments for in-context collaboration.

• Add AI agents directly into workflows so they can design, execute, and update tasks while humans remain in control.

The agents operate within Jira’s existing project settings. That means they follow current configurations, permissions, audit logs, and approval processes. For enterprises concerned about governance and oversight, this structure keeps AI activity visible and traceable.

In simple terms, AI is no longer working in the background. It shows up like a teammate with a task owner, comment history, and workflow status.

Investment in an open AI ecosystem

Alongside the Jira beta, Atlassian announced new investments in Model Context Protocol, or MCP. MCP is a standard that allows AI agents to access tools, data, and workflows in a consistent way.

According to Atlassian, enterprise customers account for nearly 50% of all Rovo MCP Server usage. Customers on paid Atlassian editions represent 93% of usage, indicating that large organisations are leading adoption.

Building on this trend, the company introduced two updates:

• MCP skills in Rovo: Rovo agents can now connect with MCP-enabled third-party applications such as Amplitude, Box, Canva, Figma, and Intercom. This allows agents to pull live data and take actions across tools.

• Rovo MCP Server general availability: Atlassian now offers a hosted MCP server that provides a single connection point to Jira and Confluence for compatible AI clients. These include Claude by Anthropic, Cursor, Google’s Gemini CLI, Lovable, and WRITER.

This approach positions Atlassian as an open platform rather than a closed AI stack. Enterprises can choose the agents and clients that fit their needs while keeping Jira and Confluence as the control centre.

From experiments to accountable teammates

Atlassian is framing this shift as a move from AI pilots to operational use. By embedding agents directly into workflows, companies can track what agents do, who assigned the task, and how outcomes move through approvals.

For young professionals and developers already using Jira daily, this could change how tickets are handled. Instead of just assigning work to a colleague, teams can loop in an AI agent, comment on its output, and refine results in real time.

The open beta of agents in Jira is now available, as Atlassian continues to expand its human-AI collaboration strategy at enterprise scale.