TLDR
Atlassian CEO Mike Cannon-Brooks argues AI will transform rather than eliminate jobs, comparing the shift to bank branches that evolved rather than disappeared. He emphasizes that design and experience — not model selection — are the key to unlocking AI productivity, and describes Atlassian's agentic multiplayer protocol (AMP) and teamwork graph as infrastructure for human-agent collaboration.
For you
PRO members can sign in and request more information — or their own version of this digest, tuned to what they care about.
Sign in to unlockSurprising Insights
Bank branches have grown in the last decade despite online banking predictions, because they fundamentally changed their function — a parallel for how AI will transform rather than eliminate jobs.
Atlassian's teamwork graph now has roughly 1,000 people working on it, up from five eight years ago, making it a massive internal investment invisible to most customers.
Cannon-Brooks explicitly avoids calling Atlassian an 'AI company' or saying they 'sell AI products,' framing AI as just another ingredient like mobile or internet was previously.
A recent user survey showed people 'hated' working purely through command-line/terminal interfaces for enterprise software, finding it unproductive compared to visual interfaces.
Key takeaways
Jobs are being restructured rather than eliminated; Cannon-Brooks believes AI will create more jobs than it changes, with the nature of existing roles evolving progressively — software developers will still exist but work completely differently within five years.
The bottleneck for AI productivity is design and experience, not model capability; companies need new interfaces that support multiple collaboration modalities (chat, delegation, pairing, multiplayer) rather than simple chat interactions.
Atlassian's teamwork graph, built over eight years, connects knowledge, work, communications, code, and data into a single enterprise context graph that agents query via MCP, reducing token usage and improving answer quality through compounding learning from each agent session.
Model blending and harnesses matter more than chasing the latest model; Atlassian routes tasks to different models based on strengths and cost profiles, treating model progress as a commodity layer available to all vendors equally.
Headless interfaces are a 'path to death' for software companies; Cannon-Brooks argues visual interfaces remain essential for productivity and that MCP/CLI access should complement, not replace, continued investment in adaptive, human-centric design.
The competitive threat from model labs building applications is overstated; delivering customer outcomes at scale with compliance and design expertise is a non-trivial 'art and creativity problem' that pure technology companies struggle to replicate.
Notable quotes
“I think the restructuring answer is probably far more accurate. I think, look, my view is net-net, it'll create far more jobs than it changes.”
▸ Notable quotes · 3:10“I'm not afraid of losing my job to AI. I'm afraid of losing my job to someone who's really good at using AI to do my job.”
▸ Notable quotes · 5:28“I don't like to say we sell AI products. I know that sounds weird. But I think we sell technology products. I think we sell products that help companies run workflows. AI is a new ingredient in the technology mix.”
▸ Notable quotes · 6:17“If you believe the majority of your usage will be through a headless interface, and therefore you don't need to continue to work on your design and experience of your software, I think that is a path to death. I think that is a very brainless move.”
▸ Notable quotes · 20:10“Our cost of writing code has come down as well. So it's our speed of running as a business, our speed of delivering for our customers. All of that is broadly available.”
▸ Notable quotes · 19:17“Model progress is amazing. We are seeing continued progress. Our job as a vendor is to harness that progress and deliver it in better outcomes. That doesn't mean just jumping to the latest model.”
▸ Notable quotes · 15:45Chapters
0:35 — Introduction and AI's impact on work: Alex Kantrowitz introduces Mike Cannon-Brooks at Web Summit and asks how AI is changing the nature of work based on Atlassian's data across Jira, Trello, and other products.
2:37 — Jobs transformation vs elimination: Cannon-Brooks argues AI will restructure jobs rather than eliminate them, using bank branches as an analogy — they grew but fundamentally changed function. He believes net job creation will be positive.
5:06 — How jobs will change: outcomes over tasks: Work will shift from task execution to outcome definition and verification. People will need to think at higher levels of abstraction, focusing on goal-seeking and continual improvement loops rather than mechanics.
7:22 — Design and experience as the bottleneck: Despite model progress, efficiency gains aren't flowing through because interfaces haven't adapted. Atlassian shipped AMP (agentic multiplayer protocol) to support multiple human-agent collaboration modalities in familiar interfaces.
10:06 — Collaboration modalities: chat, delegation, pairing, multiplayer: Cannon-Brooks outlines four interaction patterns: chat (Q&A), delegation (agent acts on behalf), pairing (side-by-side work), and multiplayer (multiple humans and agents collaborating). Each requires different interface designs.
12:42 — Teamwork graph as enterprise context layer: The teamwork graph connects knowledge, work, communications, code, and data into a single graph with inference layers. Agents query it via MCP for better answers, lower token costs, and compounding learning from each session.
14:54 — Model commoditization and blending strategy: Models have different strengths and cost profiles but aren't fully commoditized. Atlassian uses model blending and harnesses to route tasks, treating model progress as an even playing field available to all vendors.
19:58 — Headless UI debate and competitive threats: Cannon-Brooks calls headless interfaces 'brainless' and a path to death, citing user preference for visual interfaces. He downplays threat from model labs building applications, emphasizing design and compliance as hard-to-replicate differentiators.
Speakers
Alex Kantrowitz — Host
Mike Cannon-Brooks — Guest