AI series ep #2: The Atlassian Approach to AI in Service Management Q&A

[#1]

👋 Hi all! We recently wrapped up the second episode of our AI series, “The Atlassian approach to AI-powered service” now available on-demand!

Ben Costello, Principal Product Manager, AI, and I shared key insights from Atlassian’s 2025 State of AI in Service Management report, shed light on how Atlassian solves and approaches those challenges, and discussed Atlassian’s AI in Service roadmap.

The webinars were offered in three time zones and were very well attended (Thank you for that!). As a result, we weren’t able to get to all the questions, so we’ve gathered them along with the answers below. Sign up for our next episode, a customer case study, and watch previous episodes on-demand here. 

Thanks again to all who attended and asked such great questions!

Question	Answer
Where can I watch previous episodes?	https://www.atlassian.com/webinars/it/ai-series
What’s the timeline for HIPAA compliance for AI?	We don’t have a timeline to share yet, but there is an open feature request.
Of the 500 pros surveyed in the report, were all users of Atlassian products, or did the survey include respondents from organizations using other service management platforms?	For the survey, our goal is to identify overall trends regardless of tooling, so we don't ask respondents which tools they are using or require that they're using Atlassian solutions. Respondents double opt-in to participate in surveys relevant to their work and are sourced by CITE.
Will Rovo be CMDB aware?	Rovo will be aware of assets stored in Atlassian's Asset repository. See here: https://www.atlassian.com/software/rovo/features?tab=assets#tab-83e103f7-5732-483d-ac05-0a97b48858e4 It’s coming soon, and I don't have a release date to share as of yet. Stay tuned :)
With growing reliance on KB articles to support AI, what strategies do you recommend for maintaining/curating documentation? Are there specific challenges you’re seeing in keeping KB content accurate and relevant as AI-powered service management grows?	Thanks so much for this question. Growing reliance on KB articles to support AI, what strategies are recommended: Definitely leverage built-in automations and AI to keep your content fresh. In Confluence, you can remind the author to refresh their article and have Confluence suggest which updates may need to be made. This takes the manual reminders off the author's plate and even helps them get the article updates drafted to work off of. For me, I set a quarterly cadence for all of the articles that I own, since that's typically when our SOPs change. For specific challenges keeping KB content accurate and relevant: Things move very fast now in the world of AI, sometimes for the better, sometimes not so much. This means that KB articles can go out of date quickly. Leveraging built-in technologies in the tools you already have and building time in your schedule for experimentation is key. This is something I personally struggle with—building in experimentation time—but I have to stop putting it off given how quickly AI is moving. The payoff is that in the long run it will make you faster overall. I know we're speaking specifically about knowledge management, but this is true for all practices. No matter your role, building in time to research, learn, and experiment with AI capabilities is key.
Is there somewhere where I can find a breakdown of the AI costs for a company by company size?	Great question, we don't have a cost calculator available at this time. Industry publications do offer some, but from what I've seen, the estimates vary a lot. Atlassian has reports in the admin center to ensure that admins have all the necessary information to stay on top of their costs and credits.
A large concern for me when it comes to AI is the environmental impact and energy use. Is Atlassian making any considerations for that/have any resources where we can learn more on how impactful Rovo is on the environment?	Great question! Atlassian keeps the climate top of mind and ensures that we're building AI responsibly through various commitments to the environment and strategic partnerships. You can read our sustainability report here for more info: https://www.atlassian.com/company/corporate-social-responsibility/report
How would you recommend setting up an agent that can access Assets information? We have JSM and the Assets app, but when I try to ask Rovo a question such as who owns a certain device, it is unable to find it. I have tried to create an agent to search for this info and select "All Organizational Knowledge" but it seems like Assets is not accessible to Rovo.	Great question. Assets visibility in Rovo is coming soon, so you won't have to build an agent to do that. I don't have a date to share yet, but we're testing internally and are close!
When you engage with a Rovo agent does the data move to an LLM?	The detailed answer about all privacy, etc., can be found here: https://www.atlassian.com/trust/atlassian-intelligence But essentially, we use a zero data retention model with third-party providers to ensure no customer data is retained. We use a mix of internal and external models for Rovo; OpenAI is one of them. Thanks for the query!
How does data residency work with Rovo? If I ask it a question, does it go offshore (out of country)? And is there documentation to show what does and doesn't go offshore?	Here are the details of how we support data residency: https://www.atlassian.com/software/rovo/guides/admin-guide/rovo-data-usage-privacy#access-audit-log
Currently, Rovo Agents cannot analyze attachments in a Jira Work Item by themselves. I need to download the attachment and then "feed" it to the agent for analysis. Are there any plans to automate this?	This is on our roadmap and should be released soon. Stay tuned.
I saw that it is possible to connect SharePoint to Rovo: https://support.atlassian.com/organization-administration/docs/connect-sharepoint-to-rovo/, but the documentation says that we have to give permissions to read all sites. Is there the possibility to limit that to only some sites?	Because Rovo respects the permissions of the user, it will only surface sites that the user themselves has access to. But I can share your feedback with the team working on this :)
Can we expose Confluence and Bitbucket to AI code editors like Cursor AI or ChatGPT business plan via MCP? Are there official MCPs available for Confluence and Bitbucket?	You can already expose Confluence via MCP; Bitbucket is also planned. For more details, see:https://www.atlassian.com/platform/remote-mcp-server and https://support.atlassian.com/atlassian-rovo-mcp-server/docs/getting-started-with-the-atlassian-remote-mcp-server/
Hello! Will Rovo be able to scan Confluence databases? And conversely, will the Confluence database be accessible via API for consultation?	Rovo will already scan your Confluence sites, and conversely, you can expose Confluence to other AI tools via MCP. See our docs for more details:https://www.atlassian.com/platform/remote-mcp-server
Hi Kate, is there any plan for a virtual service agent to create complex tickets based on request types with forms? Ask customers questions and based on the answers fill out the form and create a ticket.	There are a couple of answers. There is a new "Raise a request" skill which allows you to create tickets and fill in different fields, including custom fields. We have a team working on forms integration in Rovo. I think they are still designing the solution, so keep an eye out early next year for releases.
We have Rovo and a number of the default agents in use, more so for creating weekly status reports for Atlas Projects. However, on viewing the instructions provided above: as the Org Admin, in Atlassian Administrator settings I cannot find the following menu items: To add Rovo to your portal, go to AI Agents > Manage in Agent Studio > Channels. From there you should be able to select the portal you'd like Rovo to show up.	Those settings are found inside the project itself; look for space settings. Each agent is attached to a specific project, so you will have to do this inside each project.
Are questions to Rovo "private and limited" to our account or are they used for training the LLM in general?	We never retain any user data or do anything like train general models with it. See our trust center docs for more details:https://www.atlassian.com/trust/atlassian-intelligence
Is Rovo an independent AI engine, or do you use external engines?	We use a mix of external and third-party models.

[Zohaib]

Thank you Kate for the detailed snapshot. It's really helpful for me who couldn't able to attend the seminar to catch up. 

I hope I can get clarify more about this question: 

Hello! Will Rovo be able to scan Confluence databases? And conversely, will the Confluence database be accessible via API for consultation?

I understand that Rovo already has capabilities to read through the Confluence pages, however, from my experiences, it is still not able to understand the Database document in Confluence, which is this feature: (https://support.atlassian.com/confluence-cloud/docs/create-and-edit-databases/) 

We are using the Database extensively in our site as an alternative to table. However, there is no other thing more we can do with the Database, except viewing from UI (neither Rovo agent can analyse the content, or automation can integrate with it). 

I look forward to details from the team how to use Rovo with Confluence Database if it is supported