Atlassian's Q4 FY26 shareholder letter reports $1.8B total revenue (up 28% y/y), $1.2B cloud revenue (up 31% y/y), and $6.6B subscription ARR (up 23% y/y). The letter highlights record enterprise deals including the largest in company history, with $3M+ ARR customers growing 50%+ y/y. A major focus is the Teamwork Graph — Atlassian's enterprise context graph spanning six data domains (knowledge, work, communications, code, assets, people) — which delivers up to 44% more accurate AI answers while consuming 48% fewer tokens. Rovo AI usage grew 50%+ q/q, MCP server MAU surpassed 1 million with MCP calls up 400%, and new Jira features enable direct AI agent assignment including integrations with Claude and Cursor.
Questions this post answers
What metrics does Atlassian report for Teamwork Graph accuracy and token efficiency?
Agents grounded in the Teamwork Graph deliver up to 44% more accurate answers while consuming 48% fewer tokens compared to agents without it. The graph spans over 200 billion objects and connections across all customer graphs. MCP server monthly active users surpassed 1 million, with overall MCP calls up more than 400% in a single quarter. Teams evaluating AI agent platforms track efficiency benchmarks like these on daily.dev.
What is the Atlassian Teamwork Graph and what six contexts does it cover?
The Teamwork Graph is Atlassian's enterprise knowledge graph that cross-references and interlinks data across six contexts: Knowledge (documents, wikis), Work (projects, tasks, goals), Communications (emails, chats, meetings), Code (repositories, pull requests), Assets (physical and digital assets via CMDB), and People (org charts, skills, team structures). These contexts are unified into a single ontology to give AI agents and humans richer, cheaper answers. Developers building on enterprise AI platforms follow Atlassian's agent architecture decisions on daily.dev.
How does Atlassian's Jira Coding Agent work and what does it require?
The Jira Coding Agent uses frontier models combined with the Atlassian Teamwork Graph's enterprise context and code intelligence to convert Jira work items into ready-to-review pull requests. It does not require local environment setup. Teams can also assign issues directly to Claude or Cursor, which read the issue, access the repo, and open a draft PR within Jira's permissions and audit trail. Engineers adopting AI coding agents in enterprise workflows watch Jira's agent integrations on daily.dev.