AI Governance for the Atlassian Enterprise: An Operating Model for Rovo Agents

Wednesday, September 30th | 2:00 PM ET

Rovo has agent governance controls now. Inventory, split permissions for using AI versus building with it, audit logs, credit visibility, approvals on agent changes. Most orgs haven't touched them, so the defaults are what's running. This session walks through them in a live instance and says which ones matter.

The admin console doesn't decide which agents should exist in the first place. It doesn't separate the ones that can run unattended from the ones that need a person checking the output. It won't flag an agent that has started being confidently wrong, and when the person who built it moves teams, nothing picks up the accountability. None of this hurts at six agents. All of it hurts at sixty.

There's also the population that never appears in the inventory. Atlassian is opening the work graph to tools outside the platform, which is the right call, but it means what's reading Jira and Confluence is no longer the same list as what an admin can see. Most governance plans haven't caught up to that.

The session ends with a control checklist, a classification that applies to agents already in the instance, and a first move that takes days.

Get the operating model before your agent count outgrows your governance. Secure your spot now.