Atlassian's AI agent Rovo generated a lot of buzz at this week's Team '25 conference. Rolled out six months ago, it has had an enthusiastic welcome from customers, some of whom were already happy to talk about their experiences and offer advice and encouragement to others.
In a shared session, David Todd, Director of IT & DEX at 1E, recently acquired by TeamViewer, talked about how agents have reduced ticket handling times by as much as 80%, and automated some big, tedious jobs. Meanwhile Frederick Frenzel, Director of PMO at HarperCollins Publishers, discussed how the global publishing house was using agents to manage project requests, streamline processes and reduce backlogs.
1E provides system management tools to support what it calls the Digital Employee Experience (DEX). It was acquired by remote connectivity vendor TeamViewer at the end of last year. David Todd says he first saw Rovo at the Team '24 Europe event last October, and immediately saw how it could benefit his team. He goes on:
We use Atlassian across our business, and fully embrace the System of Work approach. I think that's really helped us as we get the most benefit out of our Rovo agents when we have so much data in that system of work available to us.
He saw the use case for it, and after a quick demo at the show he could fully understand how easy it would be to adopt and really get the value out of it. He comments:
I didn't want this to be a top down strategy, because that doesn't work. I don't understand everyone's business processes. So one of the first things we did was [set up] some Centers of Excellence for Atlassian, bringing people from other departments like HR, marketing, and finance to help us out.
Todd held sessions for the entire company to join and ask questions. One key request was about getting Rovo agents to write code. The company's Tachyon product has its own client endpoint management language called Scale. Todd says that although this isn't difficult to learn, you still have to know how to write it -- so could Rovo learn it? He goes on:
Whilst the meeting carried on, we did it, and we connected [Rovo] to guidance for where we showed customers how to write Scale. I said, create me an agent that is an expert in writing scale. I want you to explain how it's written, and put comments in the documentation you created, and that was pretty much it. And we did that during the call and demonstrated straight away.
He said that people were coming up with ideas on how to start to use it, and then they took that away to their teams, and started to discuss their processes and how they can improve, championing it from within.
1E now uses Rovo agents, not just to resolve support tickets, but also to prevent support tickets even coming in by automating knowledge base article creation, and then providing solutions to the customer. Todd explains:
AI is brilliant, and it's not a replacement for people. It enhances what we can do and also takes away the things that we don't want to do. Writing knowledge based articles is boring. It's not helping the customers directly. I want our service desk agents to be helping customers directly. The huge change is freeing these guys up to do what they can do.
Frederick Frenzel says the AI agent has already had a similar effect at HarperCollins, bringing significant productivity increases to the Project Management Office. But it was important to set up the right guardrails. He comments:
At the end of it we're a regular media organization. So when we look at a tool like Rovo it's a really big concern of ours to make sure that we're looking out for our content providers and making sure everything's secure, and the right access is in the right spot.
In early access to Rovo, the company went into test mode, building up use cases, starting from the project management team. He explains:
That's everything from taking our handwritten notes and starting small, and trying to take all those kinds of mundane, tedious tasks out the way.
Before the Rovo agents were adopted it took weeks of planning, and getting a lot of people to come together, to lead and move things forward. Frenzel continues:
Our Day-to-day Demand to Deliver (DDD) is everything from the business intake, which in our business is our customer, just like our content providers have the customers that buy the books. When we have that initial entry point, we're using JIRA Service Management to bring in our intake of requests. So it's an internal customer bringing that in, and it went through a series of classic project management steps. Leadership is approving those requests at the secondary step to make sure they can fit within the goals of the organization.
The volume of requests was an issue, as he explains:
So we worked with Atlassian to focus. They worked with us to ensure that when we looked at this design solution, we really focused on the key aspects, and found what the biggest blocks were. Our scale was there, but it was just that. There were too many requests at the end of the day.
Frenzel has already deployed a few custom agents to help out various processes, which he says are making a big difference. He gives the example of the first ticket submission recap agent. When the business submits a request, the agent evaluates it and pulls together what that business person is trying to achieve, thus streamlining the process.
Using agents for prioritizing backlogs has been one of the biggest wins. He explains:
We have to make sure our goals are aligned. The weekly recap ticket agent was able to look at all those requests and prioritize them in a particular order. We were able to choose. edit and 'thumbs up, thumbs down' where needed, and it tied to the goals that we set for the organization...
"I remember an entire backlog when I was doing the cleanup. It was 50 requests that were under evaluation that we had to do and get to discovery. I ran the agent, and it would have taken hours for someone to do single artifacts for each one. It brought in all 50 of them in 15 minutes.
Both Frenzel and Todd are enthusiastic about the impact Rovo has had both on their processes and on collaboration. However they emphasize that it is critical to involve multiple teams and business leaders when beginning to implement AI agents. Todd concludes:
