Atlassian and OpenAI expand partnership to turn enterprise knowledge into action

Atlassian and OpenAI announced an expanded deal under which OpenAI's frontier models, including GPT-6 Astra and the GPT-5.6 series, will power agents across Atlassian's platform and its Rovo assistant. Rovo pairs those models with Atlassian's Teamwork Graph, which links people, projects, documents and decisions.

The relationship runs both ways. Atlassian says more than 3,000 of its developers use Codex, and OpenAI says it will keep relying on Jira to manage critical workflows. The companies are also exploring Jira integrations that would let teams assign work to AI agents, track their progress and review results.

We think this is one of the more sensible places to put an agent. Most of the drudgery in knowledge work is not writing; it is assembling. OpenAI's example is a product manager asking whether a launch is on track and getting an answer drawn from Jira tickets, Confluence pages and discussions, with blockers and missed milestones flagged. If that works reliably, it saves real hours, and it grounds the model in a company's own records instead of the open web.

The concern is the same feature seen from the other side. A context layer that connects everything is also a context layer that exposes everything. The announcement says ChatGPT and Codex connect to project data subject to appropriate permissions, and that matters enormously. Permission systems in large companies are often messy, and an agent that can summarize across them will surface whatever those permissions get wrong, faster and more confidently than any human browsing would.

The longer arc is the part we're watching most. Assigning Jira tickets to agents, then measuring their impact on cycle time through Atlassian's DX productivity platform, is a recipe for managing AI like a workforce. The companies say humans stay in control. In practice, once agent throughput becomes a dashboard metric, pressure builds to hand over more work and review less of it. That's where quality slips and accountability blurs.

Our read: useful, grounded, and likely to spread quickly through teams already living in Jira. The guardrails that will matter are boring ones: clean permissions, clear audit trails of what an agent did, and review steps that don't get skipped when the metrics look good.

The collaboration began in 2023. Atlassian has also launched plugins that bring Jira work items, Confluence content and people into ChatGPT and Codex prompts, along with CLI plugins for the Teamwork Graph.