Atlassian Launches Agentic Multiplayer Protocol for Enhanced AI Collaboration
Atlassian Corp. has introduced the Agentic Multiplayer Protocol (AMP), a platform update aimed at transforming the collaboration between humans and artificial intelligence agents. This new initiative focuses on establishing a shared context and jointly overseen tasks within a unified digital workspace, enabling a more effective partnership between users and AI tools.
Announced during the Team ‘26 Europe event, Atlassian’s vision for AI agents is to integrate them as team members, promoting what they term a “multiplayer” approach to workplace interactions. The goal is for human workers and AI agents to coexist and collaborate more seamlessly in an organized and governed environment.
AMP is crafted to equip each AI agent with the contextual data necessary for productivity, as well as stipulations to ensure safety within the platform. Each agent receives an identity assigned by administrators, which defines their authority and scope of operations.
Jamil Valliani, Atlassian’s Head of AI Product, articulated the concept during an interview, describing the environment as a “multiplayer game” where humans and AI agents engage dynamically, with multiple participants contributing. However, he emphasized that agents are not autonomous entities; Atlassian maintains a guiding principle: “Headless software means brainless software.” The company intends for AI to augment teamwork rather than operate independently.
Rovo, Atlassian’s AI assistant, previously noted for its autonomous capabilities, exemplifies this collaborative ethos. Users dictate objectives, oversee the proposed approach, provide necessary adjustments, and ultimately review the outcomes of Rovo’s work. The introduction of Rovo Work, a new mode for Rovo Chat, enhances this framework, particularly for complex, multi-step projects requiring human oversight.
Valliani explained that when users engage Rovo Work, they grant the AI assistant the autonomy to operate at full capacity. In scenarios where Rovo encounters unfamiliar tasks outside its training data, it has the ability to self-educate. For instance, when tasked with creating an Instagram-ready video reel, Rovo didn’t rely solely on pre-existing knowledge; it explored the requisite instructions, adapted its approach to match the desired output format, and sourced the tools for video synthesis.
Innovative Visual Communication with Loom
Atlassian is also set to redefine how everyday users interact with AI agents through its product Loom, which facilitates visual communication. Users can record videos of their actions on screen, providing clear instructions to both colleagues and AI agents. By visually demonstrating tasks—such as highlighting interface elements, clicking buttons, and entering information—users can convey their needs more effectively than relying solely on verbal or written prompts.
This is significant, as human explanations can sometimes fall short in clarity. Valliani stated, “With Loom, it’ll actually capture you saying all that, and the actual references on screen that you’re pointing to when you say it, and then format it into the prompt.” This approach fosters a more intuitive understanding for AI agents, easing the communication process.
Enhanced Contextual Awareness for Developers
For developers, Atlassian’s recent platform enhancements offer a broadened context and operational environment for AI agents. Rovo Code Search, unveiled at the event, integrates source code into the Teamwork Graph—a centralized intelligence engine that maps relationships among people, code, and documentation. This expanded Data Context encompasses structured information from platforms like Databricks, Snowflake, and Google BigQuery, providing agents with a comprehensive understanding of software development aligned with business objectives.
Additionally, Atlassian’s expanded Model Context Protocol (MCP) server offers a unified interface for both internal and external coding and AI agents. Valliani noted that a significant fraction of interactions involve agents contributing information back into the platform, reshaping Atlassian from being merely a data source to an integrated workspace where AI contributions can be harnessed by both human users and other AI entities.
The MCP server currently supports over 200 tools, handling approximately 15 million tool calls daily, marking a major stride in Atlassian’s commitment to fostering collaborative AI environments. This initiative aligns with the company’s vision of integrating AI agents as cooperative team members, allowing them to actively engage in shared tasks and workflows within their collaborative tools.
Envision scenarios where an AI agent collaborates in real-time within Confluence, Atlassian’s document management system. As agents and humans make live edits to documents, the contextual environment enables agents to adapt their approaches based on the actions of their teammates, ultimately enhancing the collaborative experience.