An employee can finish a task faster while everyone around them still waits for context, approval or a handover. Atlassian’s October founder update takes aim at that gap between personal productivity and the performance of a team.
On 6 October, at Team ’26 Europe, Atlassian introduced its Agentic Multiplayer Protocol, or AMP. The company described three areas of focus: collaboration inside existing workflows, visible agent identities and activity, and context with governance. It said the rollout was beginning that day. Its proposal included scoped permissions, activity records and human review points.
GFT interpretation: this is an attempt to make AI work legible to the people who did not write the prompt. A useful result is harder to trust if colleagues cannot identify its owner, understand the information behind it, or see what changed. Shared visibility can therefore be part of the product’s value, rather than an administrative extra.
Founders building agents should consider a simple scenario: a colleague returns after a day away and needs to understand what the agent has done. Can they inspect the work, locate the responsible person, correct an error and resume the process? A conversational interface alone does not necessarily answer those questions.
There is also an operating choice for buyers. They must decide which actions require approval, which can happen automatically and how exceptions reach a person. Those choices should reflect the consequence of a mistake. Drafting an internal summary and changing access to a sensitive system do not deserve identical controls.
An announcement cannot establish that coordinated agents will improve every team’s output. The meaningful test is whether work becomes easier to review and complete across handovers. Atlassian’s update is interesting because it places that organisational problem near the centre of the AI conversation, where founders and customers can examine it.