‘The accountability is with Mike’: Atlassian CEO’s AI agent rule | Fortune

Atlassian cofounder and CEO Mike Cannon-Brookes argues that AI agents in the workplace must be identifiable and governed, with ultimate accountability resting on the human end user rather than the AI itself. Speaking to Fortune at the Singapore Grand Prix, he stressed that "the accountability is with Mike" and that an agent's actions should be traceable in document histories just like a colleague's edits, so people can see who did what when. On Oct. 6, Atlassian rolled out its "Agentic Multiplayer Protocol," a control plane that turns AI agents into visible, governed team members with distinct profiles and owner attribution.

Cannon-Brookes is critical of new personal AI agents such as Meta's Muse and OpenAI's Dots, which hide what the agent is doing behind an invisible interface. He says that may be acceptable for consumers but not for workplaces, where permissions, security, and compliance matter, citing risks such as employees using privileged internal documents to generate content that could leak information. He also argues most AI interfaces are too hard to use, comparing them to Microsoft Word losing its toolbar, and points to Atlassian's 2024 Rovo assistant and new design patterns that let users link tools like Figma, Codex, and Claude.

On Australia's tech scene, Cannon-Brookes says the country has an amazing ecosystem that is globally oriented from the start, noting Atlassian's first customers were in England and Sweden and that Australian firms benefit from alignment with the U.S. Atlassian, known for Jira and Confluence, reported $1.8 billion in quarterly revenue, beating estimates, after cutting about 10% of its workforce in March as it pivoted to AI and enterprise sales.

(Source：Fortune)