Atlassian: Humans and AI Agents Collaborate 10 Million Times a Month

People and AI agents now collaborate 10 million times a month on Atlassian's platform — a figure the company says has doubled in six months. Atlassian disclosed the number in a Thursday shareholder letter posted alongside its first-quarter fiscal 2027 results.

The milestone landed beside a familiar SaaS beat. Total revenue hit $1.772 billion, up 24% year over year. Cloud revenue reached $1.302 billion, up 31%. Subscription annual recurring revenue climbed 23% to $6.812 billion, while remaining performance obligations rose 44% to $4.746 billion. GAAP operating margin came in at 11%, and free cash flow reached $463 million.

"Enterprise AI has hit an inflection point," Mike Cannon-Brookes, Atlassian's chief executive and co-founder, said in the release. "Model power is table stakes. The advantage goes to platforms that bring people and AI agents together with shared context, deep collaboration, and confidence in the outcome."

The quote is the clearest signal yet of where Atlassian sees the enterprise AI market heading — away from raw model capability and toward platforms that can orchestrate humans and agents on shared ground.

How fast is Rovo actually growing?

Fast. Rovo agentic automations are up 9x year over year, according to the letter, as customers move past one-off AI queries into automated workflows. Users of Rovo Agents have doubled in the past six months. Rovo-assisted actions per user are up more than 3x since the start of the calendar year.

Adoption reaches into the riskiest customer segment, too. More than 80% of highly regulated cloud customers have Rovo enabled, Atlassian said.

What is AMP, and why do agents need identities?

Atlassian is pitching the Agentic Multiplayer Protocol, or AMP, as a layer where agents show up as visible teammates inside Jira, Confluence, and Loom. Each agent gets a distinct identity, governed authority, and an audit trail. The design pushes agents out of private chat windows and into the systems of record where work actually happens.

Coding agents are already living there. Codex, Claude Code, and Cursor create and update Jira work through Atlassian's Model Context Protocol server. That MCP server now serves more than 2 million monthly users and handles more than 15 million tool calls a day, the release said.

The MCP traffic is not a side channel — it's a growth engine. After adopting the server, users create 50% more work in Jira and 75% more in Confluence on average, Atlassian reported. Some 98% of MCP users are also active on the broader platform in the same month. Customers using the MCP server are growing ARR at roughly twice the rate of non-adopters, per the letter.

What is the Teamwork Graph?

The context layer underneath all of this is the Teamwork Graph. It now maps more than 250 billion objects and relationships, up from 150 billion six months earlier, and connects to more than 80 third-party tools. Atlassian also said it expanded its OpenAI partnership so ChatGPT and Codex can tap that graph.

The graph matters because Cannon-Brookes's thesis depends on it. Agents can only collaborate with shared context if a platform holds that context — and Atlassian is arguing its graph, built from years of Jira tickets and Confluence pages, is the moat.

How does packaging carry the AI story?

Product packaging is doing some of the work. More than 3,000 customers have upgraded to Teamwork Collection, a 7x increase from a year ago. Those accounts run twice as many agents as Jira-only customers and complete 35% more work items per Jira user. Service Collection passed 70,000 customers, with ARR growing north of 30% year over year.

The pattern is straightforward: customers who buy the broader collections use agents more and get more work done, which gives Atlassian a measurable story to tell on upsell.

What does Atlassian expect next year?

For the full fiscal year, Atlassian targets subscription ARR growth of about 18.5%, total revenue growth of about 13.75%, and cloud revenue growth of about 26%. Second-quarter revenue is guided to $1.825 billion to $1.835 billion, with cloud growth of about 27%.

The AI-agent push continues product work WebProNews tracked earlier, including always-on agent loops in Jira that move backlog items toward pull requests. Thursday's numbers put a scale claim behind that roadmap: agents are already in the monthly collaboration count, not only the keynote slide.