Atlassian Corp. Plc. today launched Rovo, a new generative artificial intelligence knowledge discovery product that extracts data from a company’s internal tools and can help find information stored in them and then act on it.

Powered by Atlassian Intelligence, a generative AI offering the company launched a year ago, that brought AI services and models into its products, Rovo helps companies find information lost amid multiple products, learn quickly and act on behalf of employees like a “virtual teammate.”

“The first thing Rovo does is it brings together all those third-party SAS tools, as well as your custom-built in-house tools, and lets you put it all into one search index for all of your employees,” Sherif Mansour, head of artificial intelligence at Atlassian, told SiliconANGLE in an interview.

It can do this because it has a semantic understanding of all the tools based on the company’s information based on what Atlassian calls its proprietary “teamwork graph.” Mansour said this is the company’s “secret sauce” that makes Rovo, and all the other AI capabilities, work so well, by providing personal context to everything that it does. 

Rovo will surface different search information for an employee in marketing than it will for someone who is working with an engineering team. It will do this partially because of the information the employee has access to – as Rovo respects security “allow and deny” lists set by information technology teams and administrators – and also because Rovo is aware of whom the employee interacts with and what they search for. This also means that if someone has private documents, Rovo will not use the data in them if another user searches.

“And we’ll use signals such as the topics you’ve worked on, what you’ve recently edited or changed across Atlassian and non-Atlassian apps,” said Mansour. 

Getting up to speed quickly with Rovo

Because Rovo is aware of the entire context of a company’s organization, projects, goals, new teammates and more, it can also insert itself where employees work – which means that it can also help with a deeper understanding of anything that an employee comes across.

Within search, it can surface information about the organization with information cards that reveal extra information about projects using contextual information by providing additional citations that might be missed in traditional search. That means information gets updated dynamically as new documents are added throughout projects and the organization.

It also exists where people work, such as in their documents. For example, Rovo can act as a glossary for company jargon and terms of people, projects and things that might be confusing.

Say a company has a project named “Blueberry,” but an acronym PPG. A new employee might have absolutely no idea that PPG stands for “Portal Payments Gateway,” since it’s an acronym, but Rovo can highlight this as a glossary term. If that comes up in a document, Rovo can underline it so that I can be clicked on and it will then reveal relevant information about the term.

Since this is generative AI, it also comes with a conversational capability. If a user wants to follow up with a question, Rovo can answer within the context of the application and it will stick to what it knows. It will also cite sources within company data to support its claims, Mansour said, making it an ideal companion to find internal company sources.

Onboarding virtual teammates with Rovo

Allowing generative AI chatbots to do work for you is becoming a hot topic lately and now that Rovo can help find data and help learn from it, it can also act on it. This capability is known as Rovo Agents, and teams can build these agents themselves or call them up out-of-the-box from the Atlassian Marketplace.

One example of an agent is Jira Backlog Buddy. This Rovo agent can look at a Google doc or in Confluence and translate that into a bunch of tasks in Jira, which is Atlassian’s project and issue tracking system. At the start, Backlog Buddy will suggest groupings for work within the project of how the tasks might be laid out and how they could be placed within the Jira project, which takes away a lot of the tedium that might have taken an entire day for the team to hash out.

“The user is always in control here,” said Mansour. “So, it’s not going to go ahead and create these things. It said, here’s some suggestions. Do you want to tweak them? Or do you want me to go ahead and create that for you? So, in this example, we’ll say, OK, cool. Let’s just create this suggested grouping of work inside my project. Then it goes ahead and does that.”

Developers can create Rovo agents with code or without code by using prompts and attach them to Atlassian Marketplace apps, which provide agents with the skills that they need to interact with data to take action. Mansour said Atlassian is aiming to release about 20 Rovo agents that will be available for users to start using immediately.

Image: Atlassian