Atlassian CEO on AI Agents, Layoffs, and the SaaSpocalypse Debate

Atlassian co-founder and CEO Mike Cannon-Brookes spoke with The Verge's Decoder podcast in an interview published Sept. 28, 2026, about AI, enterprise software and the debate sometimes called the SaaSpocalypse. The Verge
He spoke from Australia, where Atlassian is based. He started the company with Scott Farquhar and has been on its board since October 2002. Atlassian
Atlassian makes Jira and Trello. These are widely used tools for planning work, tracking tasks and coordinating teams. They live inside daily workflows.
The broader context here is why that placement matters. Because Jira and Trello sit where work is planned and tracked, Cannon-Brookes sees directly how AI changes team structure and how companies buy and run software.
The interview also covers layoffs Atlassian made earlier in 2026. Cannon-Brookes said the company needed a different mix of skills. He gave no roles or numbers. He described it as a question of composition, not simply getting smaller.
There is earlier background to this debate. On March 6, 2026, Andreessen Horowitz published a podcast called 'Atlassian CEO on the SaaS Apocalypse, AI Agents & What Comes Next' with Cannon-Brookes, Alex Rampell and Erik Torenberg. It focused on a selloff in SaaS stocks. Andreessen Horowitz
The Decoder interview returns to that topic six months later. The name has shifted slightly, from SaaS apocalypse to SaaSpocalypse. The question is the same. Will AI agents, software that can act on its own to complete tasks, replace apps paid for per user, or will existing systems of record, the official places where work history is kept, absorb agents and continue?
The broader context for architects is access and accountability. The issue is less whether AI can write code or draft a ticket. It is about state, meaning stored data, plus permissions, audit trails and ownership. Collaboration tools hold that information. Any agent working for a team still must read and write in the same place, under the same rules.
Looking at what this means for technology leaders, the layoff point matters more than the SaaSpocalypse label. Saying the company needs a different mix of skills points to a change in how teams operate. Teams built around installing SaaS, connecting tools and sorting tickets by hand work differently from teams built around supervising agents, checking their work and handling exceptions. Job titles may stay the same while the daily work changes.
In my view, that change explains why the SaaS debate often talks past itself. One side talks about pricing and seats. The other talks about workflows and results. Both can be right. Established platforms can lose growth in user counts yet handle more transactions if they become the control plane, the place where agent actions are logged and approved. The risk is not sudden disappearance. It is slow movement of value from the user interface to the infrastructure underneath.
Worth flagging for practitioners is who defines done. Jira and Trello have long set that definition for engineering and operations teams. If agents start closing tasks on their own, clear rules for tasks, approvals and responsibility matter more, not less. That gives an advantage to vendors that hold the record, if they offer clean APIs, which are standard ways for software to connect, plus policy controls and visibility into what non-human users do.
The optimistic read is straightforward. Better coordination tools have helped small teams ship more in the past. If agents cut the cost of routine coordination, people can spend more time on judgment, design and customer problems. That shift disrupts staffing. It helps output, if governance keeps pace.


