3 lessons from rebuilding Atlassian’s product development lifecycle for people and agents
Avani Prabhakar · 10 November 2026
Read the originalAtlassian is redesigning how its nearly 10,000 research and development staff build products with AI agents, aiming people at decisions and keeping shared specifications the agents work from.
Avani Prabhakar, Atlassian's Chief People and AI Enablement Officer, wrote that the company's 2026 developer survey found AI saving less time than a year earlier. In 2025, 68% of developers said AI saved them more than ten hours a week. This year 26% do, and 49% spend four or more hours a week managing AI.
- An automated pipeline finds security vulnerabilities, proposes fixes and runs tests before an engineer decides whether to ship. Agents prepare fixes for 52% of vulnerabilities, saving about 50 engineering days a week.
- Teams record what to build and how in shared specifications that become Jira tasks agents carry out and update. In one Confluence pilot, two features shipped in 2 and 4 developer weeks against estimates of 30 and 40.
- More than 50 product managers and designers who had never written production code shipped over 100 engineering-approved changes in three months, with engineers scoping and reviewing the work.
- Atlassian is changing roles in a set order: first how AI changes tasks, then which skills people need, and only then the roles themselves.
A product organisation could look first at the handoffs between product, design and engineering, rather than adding AI to each job, as Prabhakar advises. The piece is the company's own account and promotes its playbook.
Have an automated pipeline propose and test a fix for each flagged security vulnerability, so an engineer starts from the fix and decides whether it is safe to ship.
Record decisions about what to build and how in a shared specification that becomes the tasks AI agents carry out and update.
Derived by Working Surface from the article; more in the Playbook. Source line: When agents can generate code, designs, and prototypes in minutes, the bottleneck moves to deciding what’s worth building and whether it’s good enough.
13 November 2026: AI agents earn more freedom when a team limits what they can reach, from data to design components, and adds a check that does not depend on a glance.