To get the most out of Jira and agents together, Atlassian is announcing several updates to this part of its System of Work offering today. Agents will take on a more active role in Jira, and it will be possible to use Rovo Agents’ capabilities throughout the rest of the stack.
AI Agents can be extremely useful. However, it is important not to just let them loose everywhere, but to make them part of an actual team. At least, that is what Atlassian has in mind. Shortly before today’s announcements, we spoke with Sanchan Saxena, SVP and GM Teamwork Collection at Atlassian. “The difficult part is not finding use cases [for agents, ed.],” he says. According to him, the problem lies mainly in the amount of chaos agents create. “What we don’t want is agents that do 10 times the work but also create 10 times the chaos,” he says.
Agents become an integral part of Jira
To allow agents to do their job without creating extra chaos, Atlassian has made them part of its own platform. This means that people in Jira can assign work to agents, but also use @mentions in Jira comments to put them to work. If you want to do some extra research on something you’re working on, you can easily have AI agents do it for you.
Of course, the intention is that people use the right agent for the right task. That’s something Atlassian takes care of behind the scenes. Jira acts as a kind of orchestration layer between people and agents. Agents also have access to the Teamwork Graph, on which virtually everything Atlassian is built. This means that agents can also incorporate all the context and interrelationships from the Teamwork Graph into their work. The fact that Atlassian integrates agents in this way also means that it is clear how they work. Everything is verifiable, even if an audit is carried out.
When we ask Saxena about responsibility, he is clear: “People are always ultimately responsible.” That also means that everything must be well organized. It must be clear who is responsible for what. He readily admits that this can certainly be a challenge when scaling up. That is why Atlassian assigns agents to specific projects, among other things. This provides a better overview and also enables people to take responsibility for them.
Rovo MCP Gallery
The second announcement today related to Jira is the Rovo MCP Gallery. This is what you would expect it to be based on the name. A range of MCP servers from other tools that Rovo Agents can connect to. Quite a few are already available at launch. Amplitude, Box, Canva, Figma, GitHub, and New Relic are just a few names. It is also possible to offer your own internal systems to agents via MCP in this way, or other Atlassian products.
Agents can retrieve live data via the MCP Gallery, but also take action on behalf of a user, for example to create tasks or update data. Ultimately, the MCP Gallery should provide more context from outside. There is already a lot of context available within the Atlassian ecosystem, but external context can provide even more insights. It also ensures a more coherent way of working. After all, a single task should now only produce a single answer. If you have to extract it from all kinds of different tools, you end up with a lot more. That’s not always useful.
The third-party tools offered in the MCP Gallery are also built by those parties. Of course, Atlassian cooperates where possible and where necessary, Saxena indicates. However, the ultimate responsibility for building them lies with the third party. He is not concerned that other suppliers will not build enough MCP connectors. “Jira is one of the most requested MCP connectors,” he says. For many organizations, it is one of the most important tools when it comes to workflows. In addition, there is of course also an advantage for third parties. This way, their customers can also gain more insights from the data that is imported as context.
Read also: Atlassian’s System of Work vision takes shape with Teamwork Collection and Rovo AI Agents