Atlassian lays groundwork for humans and AI agents to work side by side
SiliconANGLE Kyt Dotson ● Covered by 2 sources
Atlassian is making AI agents work inside the same tools as people. It’s trying to give them roles, limits, and shared context instead of letting them roam free.
Based on reporting by SiliconANGLE, Kyt Dotson — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Atlassian is pushing its AI strategy past chatbots and into something closer to a staffed workflow. At Team ’26 Europe, the company announced Agentic Multiplayer Protocol, a platform update meant to let humans and AI agents work in the same digital space with shared context and shared tasks. The pitch is simple: treat agents less like detached tools and more like teammates, but keep them inside a governed system.
That governance is the point. Head of AI Product Jamil Valliani told SiliconANGLE the company wants agents to have enough context to be useful, but also an identity, authority and scope assigned by administrators. Atlassian’s internal line is blunt: “Headless software means brainless software.” In other words, autonomy only matters if it still fits into human teamwork.
Rovo, Atlassian’s AI assistant, already has some of that flavor. Users set the objective, review the plan, steer it when needed and check the result at the end. Now Atlassian is adding Rovo Work, a new mode for Rovo Chat that is meant for complex, multi-step jobs that humans approve along the way. Valliani said sending a query to Rovo Work gives it permission to “unleash itself fully,” which is a bold way of saying the system is built for longer tasks, not quick answers.
The more interesting part is what happens when the model runs into something it doesn’t know. In one example, a product manager asked for an Instagram-ready reel, and Rovo went looking for the right instructions, figured out the needed output format and picked up tooling for video synthesis. Valliani also said power users can get Rovo to generate custom skills, though not every learned behavior becomes something exportable.
Atlassian is also trying to replace a lot of awkward prompt-writing with something more natural. Loom, its screen-recording product, can capture a user pointing at interface elements while saying things like “Make this larger, move this, turn it green,” then format that into a prompt. For developers, the company is widening the context around code with Rovo Code Search, Data Context for systems like Databricks, Snowflake and BigQuery, and a bigger Model Context Protocol server. That server now exposes 200 tools and handles roughly 15 million tool calls a day, turning Atlassian from a place agents fetch data from into part of the workspace itself.
My take — AI-written commentary, not fact-checked reporting
This is the sane version of the agent hype cycle: give the bots a desk, a badge and a manager, not a blank check. Atlassian seems to understand what a lot of vendors still miss: useful AI is usually collaborative, supervised and a bit boring. The open invitation is fine; the open aisle to wander off into chaos is not.
Read more about this at: SiliconANGLE