The Atlassian Community Forums are currently in read-only mode. We will be relaunching on a new platform on September 22 (read more here). We apologize for the extended downtime. For concerns or questions, please email communitymanagers@atlassian.com. See you on the other side, on the new Atlassian Community Forums! :)

×

Forums

Articles
Create
cancel
Showing results for 
Search instead for 
Did you mean: 

🚀 Now rolling out: Agent actions in Automation & Agent Accounts

As AI agents take on more responsibilities across our work, they’re quickly becoming active digital teammates. But to truly rely on them, admins need complete control over what agents can access, and builders need agents that can actually get work done end-to-end.

Today, we’re excited to announce the General Availability of two major capabilities that will roll out over the next couple of weeks: Agent Accounts and Agents acting in Automation.

What’s new

Agent Accounts

With agents taking action, permissions matter. That's why we're introducing Agent Accounts. You can now choose whether agents use their own unique identity and scoped access, or continue running under the permissions of the invoking user.

agent-accounts1.jpeg

Just like users, you can list all your agents in Atlassian Administration and grant them product access by adding them to groups.

team-triage.png

Agents acting in Automation

Until now, adding a Rovo agent to an automation meant the agent could think (analyze text, summarize, or classify data), but it couldn't act. This meant that automating agents required complex, multi-step automations that were hard to build and even harder to manage.

Now, agents can take actions directly inside your automations – set a trigger, choose your agent, give it a prompt, and let it handle the work end-to-end.

agents-in-action.png

See exactly which account the agent runs as, what tools it has access to, and choose whether it's allowed to take action.

What this means for your teams

  • Build more powerful workflows with less effort: Collapse complex multi-step automations into a single natural-language prompt. For example, replace 10+ branch decision trees with a single agent that reasons and acts, delivering the same outcome in significantly fewer steps.

  • Unlock use cases that weren't possible before: Agents can now create issues, update fields, write pages, and more, because they can reason and act.

  • Grant access on your terms: Give or revoke an agent's access to Atlassian apps and third-party tools through Atlassian Administration, the same way you manage users today.

  • Track every action with full audit clarity. When an agent acts, its name appears in audit logs, so you always know who did what.

How to get started

Getting started is easy, whether you’re building something new or upgrading an existing flow.

  • To set up agent accounts: Admins can set up a default access policy to ensure all newly created agents are granted access to the right apps. You can manage agent accounts in Atlassian Administration under the Rovo tab.

  • For new automations: Builders can add the "Use Agent" action, choose their agent, craft their prompt, and flip the "Let this agent take actions" toggle on, all directly in Studio.

  • For existing automations: Your current automations won't change automatically. To enable actions, builders can edit their automation, select the "Use Agent" step, and switch the toggle on. Make sure to review automations to remove any redundant downstream steps that the agent can now handle directly.

We understand that control and compliance are essential. In addition to automation-level toggles, admins can disable agent actions across all automations via Studio admin settings. When disabled, this setting acts as a global override across the entire instance.

Both new features are rolling out gradually over the next couple of weeks – keep an eye out for them! To learn more and get started, check out these support pages:

TAGS
AUG Leaders

Atlassian Community Events