Throughout its Group ’24 convention in Las Vegas, Atlassian at present launched Rovo, its new AI assistant. Rovo can take knowledge from first- and third-party instruments and make it simply accessible by a brand new AI-powered search software and different integrations into Atlassian’s merchandise. Essentially the most attention-grabbing half, although, will be the new Rovo Brokers, which can be utilized to automate workflows in instruments like Jira and Confluence. One nifty side of those brokers: Anybody can construct them utilizing a pure language interface. No programming required.
“We like to consider Rovo as a big information mannequin for organizations. It’s a information discovery product for each information employee,” Sherif Mansour, Atlassian’s head of product for Atlassian Intelligence, informed TechCrunch. “Once you have a look at what a information employee has to do, they kind of undergo this means of: I must discover a piece of labor. I must be taught and perceive it. After which I take an motion. Most individuals which have some kind of desk job undergo that loop. I believe what’s thrilling about Rovo is that we’re lastly on the genesis of generative AI touchdown that that helps speed up what we will do in that space for groups.”
Atlassian Group `24 Las Vegas. Picture Credit: Atlassian
The premise for Rovo is Atlassian’s “cloud teamwork graph,” the identical graph that types the inspiration of Atlassian Intelligence, the corporate’s year-old effort of bringing an AI teammate to its merchandise. That graph brings collectively knowledge from Atlassian’s personal merchandise and quite a few third-party SaaS instruments. And in a means, it’s the proliferation of SaaS instruments that necessitates purposes like Rovo, as a result of each software tends to have its personal knowledge silo, making it tougher for workers to seek out the data they want.
Picture Credit: Atlassian
Rovo, Mansour stated, revolves round three pillars of teamwork: serving to groups discover and join with their work, serving to these groups be taught after which serving to them take motion.
In a means, enterprise search is the low-hanging fruit right here, since Atlassian is already aggregating all of this knowledge. Nevertheless it’s additionally a software that ought to show instantly helpful for its customers and hold them from having to always change contexts to seek out info. A few of the third-party instruments which can be supported out of the field embrace Google Drive, Microsoft SharePoint, Microsoft Groups, GitHub, Slack and Figma.
Enterprises, which frequently have numerous customized instruments, may construct their very own connectors. Atlassian itself, for instance, constructed a connector that brings in its inside developer documentation. Merely making that documentation out there in Rovo, Mansour stated, saved builders an hour or two each week — a better time financial savings than what those self same builders report from utilizing an AI code-generation software.
As Mansour careworn, the most important technical problem — except for constructing the AI infrastructure to energy Rovo — is constructing all of those connectors and making certain that they respect the entry permission set by an organization’s IT and safety groups. “Once you search, you get a unique set of outcomes to my search. We make it possible for it’s tailor-made to you and respects your permissions — and solely [shows] what you have got entry to.”
Picture Credit: Atlassian
It wouldn’t be 2024 if Rovo didn’t additionally come as a chat service. Because it additionally has entry to all of this knowledge, it’s a comparatively simple process to make use of retrieval-augmented era (RAG) to feed a big language mannequin with it and have the mannequin present personalized solutions.
Even when utilizing RAG, massive language fashions are nonetheless prone to hallucinations (although RAG significantly reduces the probabilities of the mannequin going off script). To make sure that customers can belief the outcomes, Rovo at all times cites its sources, and more often than not (with slideshows and Figma designs, for instance), there’s even an interactive preview.
One attention-grabbing characteristic Atlassian additionally constructed into Rovo is its skill to detect and clarify firm jargon. There’s even a Chrome extension for this that can mechanically underline and clarify a sure company-specific time period as you learn a Google Doc, for instance. This characteristic is powered by Rovo’s semantic search engine.
Digital teammates
It’s one factor to seek out info. It’s one other to take motion on it. That’s the place Rovo Brokers are available. In a means, that is an extension of what the corporate did with Atlassian Intelligence. Certainly, the corporate describes Rovo Brokers as “digital teammates,” too.
“Rovo Brokers will rework teamwork with their skill to synthesize massive volumes of enterprise knowledge, break down advanced duties, be taught as they take motion, and accomplice with their human teammates to make important and complicated choices,” Mansour writes in at present’s announcement. “Brokers aren’t just a few souped-up model of chatbots. They carry specialised information and expertise to all kinds of workflows and processes.”
Picture Credit: Atlassian
Which means they’ll generate, assessment and edit content material for advertising use, product specs or Jira points. Customers may construct brokers that reply particular questions or advocate greatest practices. However extra importantly, they’ll automate duties based mostly on when a Jira difficulty progresses, for instance, or assist customers clear up their Jira backlogs or manage Confluence pages — all with people within the loop.
“We’ve a powerful perception that the way forward for teamwork is teammates working alongside digital teammates — brokers,” Mansour stated. “There’ll be lots of them and also you’ll be interacting with them in your day-to-day workflows.”