How to Build an AI Native Team with Mike Cannon-Brookes

What It Covers

Atlassian CEO Mike Cannon-Brookes discusses how enterprises can advance from AI novice to AI-native status, covering the role of organizational context graphs, agentic workflows inside Jira and Confluence, and why 2026 marks the shift from chat-based AI toward embedded product experiences across 300,000+ Atlassian customer organizations.

Key Questions Answered

• •Enterprise AI Adoption Sequence: Before employees can use AI tools, security and compliance infrastructure must be built first. When Atlassian acquired browser company Dia, only 4 of 13,000 employees could initially use it due to security restrictions. Organizations should map their full enterprise controls before broad deployment, not after rollout begins.
• •Context as Multiplier: Atlassian frames enterprise AI value as intelligence multiplied by context. Their Teamwork Graph indexes over 150 billion objects and connections, including org charts, skills, code repositories, and physical assets. Organizations should prioritize building a unified context layer rather than deploying isolated AI tools that lack access to organizational knowledge.
• •Headless Agent Access via CLI and MCP: The Teamwork Graph CLI ships with 60–70 command sets built specifically for agents, enabling coding tools like Cursor or Claude Code to query Atlassian's full semantic code index and org data. Teams should expose their knowledge graphs through MCP servers so agents retrieve pre-processed context rather than burning tokens on repeated reasoning hops.
• •Measuring Output Quality Over Token Consumption: Leading enterprises track engineering throughput, flow, and output quality rather than token usage volume. Atlassian's DX acquisition helps organizations with 5,000–10,000 engineers measure whether AI coding tools actually improve productivity. Teams should define output quality benchmarks before evaluating AI tool ROI to avoid being misled by usage metrics.
• •Skill-Sharing Loops Accelerate Adoption: Because no organization has employees with more than a few years of AI deployment experience, Atlassian runs internal sharing programs where staff post Loom videos documenting both successes and failures. Teams should institutionalize structured failure-sharing channels alongside wins to compress the collective learning curve across the entire organization simultaneously.