Atlassian today announced Rovo, its new AI assistant, during the Team '24 conference in Las Vegas. Rovo takes data from first-party and third-party tools and makes it easily accessible through new AI-powered search tools and other integrations into Atlassian products. Most interesting, however, may be the new Rovo Agent, which can be used to automate workflows in tools like Jira and Confluence. The great thing about these agents is that anyone can build them using a natural language interface. No programming required.
“We like to think of Rovo as a large-scale knowledge model for organizations. It's a knowledge discovery product for all knowledge workers,” said Sherif Mansour, Head of Atlassian Intelligence Products at Atlassian. told TechCrunch. “If you look at what knowledge workers have to do, they go through a process like this: They have to find a job. They have to learn and understand it. Most people who have some sort of desk job go through that loop. What's exciting about Rovo is that we're finally getting to the point where generative AI is going to be there for teams. I think it will help accelerate what we can do.”
Atlassian Team `24 Las Vegas
The foundation of Rovo is Atlassian's Cloud Teamwork Graph. This is the same graph that is also the basis of his Atlassian Intelligence, the company's long-standing effort to bring his AI teammates into its products. This chart aggregates data from Atlassian's own products and numerous third-party SaaS tools. And in some ways, the proliferation of SaaS tools is making him need applications like Rovo. Every tool tends to have its own data silos, making it difficult for employees to find the information they need.
Image credit: Atlassian
Rovo said Rovo revolves around three pillars of teamwork. It's about helping teams find and collaborate on their work, helping teams learn, and helping teams take action.
Atlassian already aggregates all this data, so in some ways enterprise search is a low-hanging fruit. But it's also a tool that can quickly prove useful to users, eliminating the need to constantly switch contexts to find information. Third-party tools supported out-of-the-box include Google Drive, Microsoft SharePoint, Microsoft Teams, GitHub, Slack, and Figma.
Companies that use many custom tools often have the option of building their own connectors as well. For example, Atlassian itself has built a connector that brings in internal developer documentation. According to Mansour, just making that documentation available to him on Rovo saved developers an hour or two each week. This was a significant time savings over what the same developer reported using his AI code generation tool.
As Mansour emphasized, aside from building the AI infrastructure that powers Rovo, the biggest technical challenge is building all these connectors and managing the permissions set by the company's IT and security teams. Make sure to comply. “When I search, I get different results than my search. We make sure it's tailored to you and respects your permissions. [shows] What you have access to. ”
Image credit: Atlassian
We wouldn't be in 2024 if Rovo hadn't also emerged as a chat service. Because you also have access to all this data, it's a relatively simple task to use search augmentation generation (RAG) to feed data into large language models and have them provide customized answers.
Even when using RAGs, large language models are still susceptible to hallucinations (although using RAGs greatly reduces the chance of the model going off-script). To help users trust the results, Rovo always cites his sources, and in most cases (like slideshows and his Figma designs) he even has interactive previews.
One of the interesting features Atlassian has built into Rovo is the ability to detect and explain company terminology. For example, there are Chrome extensions that automatically underline and explain specific company-specific terms when you're reading Google Docs. This feature is powered by Rovo's semantic search engine.
virtual teammates
Finding information is one thing. Taking action on it is another thing. This is where Rovo Agents come into play. In a way, this is an extension of what the company did with his Atlassian Intelligence. In fact, the company also describes his Rovo agents as “virtual teammates.”
“Rovo Agents transforms teamwork with its ability to integrate large amounts of enterprise data, decompose complex tasks, learn on the fly, and collaborate with human teammates to make critical and complex decisions. ,” Mansour said in today's announcement. “Agents are more than just an improved version of a chatbot. They bring specialized knowledge and skills to a variety of workflows and processes.”
Image credit: Atlassian
This means you can generate, review, and edit content for marketing uses, product specs, Jira issues, and more. Users can also build agents that answer specific questions and recommend best practices. But more importantly, it allows users to automate tasks based on when Jira issues progress, for example, or to help users clean up their Jira backlog or organize their Confluence pages. All humans participate.
“We strongly believe that the future of teamwork is for teammates to work together with virtual teammates, or agents,” Mansour said. “There are a lot of them, and you'll be interacting with them in your daily workflow.”