Hi Community! This is my very first post here in the Atlassian Community. I’ve been a long-time lurker, but I finally decided to jump in because my team is currently tackling a rather tough nut to crack, and I’d love to get your insights. We’ve been trying to roll out Atlassian Rovo, and honestly, it’s been quite a rollercoaster. I wanted to share my first impressions because I feel there’s a significant gap between the AI hype and the reality of implementing it in day-to-day workflows. Here are the two main friction points we are currently facing: * The feeling of immaturity in the tool While the potential is clearly massive, it often feels like Rovo is still finding its footing:* Inconsistent results: The quality of the answers varies drastically. Sometimes it strikes gold, and other times it completely misses the mark on fairly straightforward queries. * The context dilemma: It requires a nearly flawless Confluence architecture. If there is old documentation or "noise" in the space, Rovo gets confused easily (and let’s be real, who actually has a 100% pristine Confluence?). * The trust barrier (The human factor): our users are naturally skeptical.* The moment Rovo "hallucinates" or fails to find a document that is literally right in front of them, user trust drops to zero. Rebuilding that trust is proving to be an uphill battle. My big question for you: How are you managing the learning curve and team expectations? Have you managed to get your users to see it as a reliable co-pilot, or is it still seen as a shiny new toy that requires constant double-checking? Right now, I’d say our relationship with Rovo is "complicated". It’s like that eager new intern: full of potential and energy, but you still have to explain things three times to make sure they don’t lose the plot. Anyone else out there dealing with this phase of infinite patience? Looking forward to hearing your thoughts and happy to finally be a part of the community!

nav[aria-label="Primary Navigation"] { padding: 0; & ul { list-style: none; width: 100%; display: flex; flex-direction: row; justify-content: start; align-items: start; gap: 30px; padding: 0; & li { margin: 0; } & ul li { list-style: none; } } }

Navigating the Rovo adoption curve

Alvaro Aparicio_9g0djf

Hi Community!

This is my very first post here in the Atlassian Community. I’ve been a long-time lurker, but I finally decided to jump in because my team is currently tackling a rather tough nut to crack, and I’d love to get your insights.

We’ve been trying to roll out Atlassian Rovo, and honestly, it’s been quite a rollercoaster. I wanted to share my first impressions because I feel there’s a significant gap between the AI hype and the reality of implementing it in day-to-day workflows.

Here are the two main friction points we are currently facing:

The feeling of immaturity in the tool
While the potential is clearly massive, it often feels like Rovo is still finding its footing:
- Inconsistent results: The quality of the answers varies drastically. Sometimes it strikes gold, and other times it completely misses the mark on fairly straightforward queries.
- The context dilemma: It requires a nearly flawless Confluence architecture. If there is old documentation or "noise" in the space, Rovo gets confused easily (and let’s be real, who actually has a 100% pristine Confluence?).
The trust barrier (The human factor): our users are naturally skeptical.
- The moment Rovo "hallucinates" or fails to find a document that is literally right in front of them, user trust drops to zero. Rebuilding that trust is proving to be an uphill battle.

My big question for you: How are you managing the learning curve and team expectations? Have you managed to get your users to see it as a reliable co-pilot, or is it still seen as a shiny new toy that requires constant double-checking?

Right now, I’d say our relationship with Rovo is "complicated". It’s like that eager new intern: full of potential and energy, but you still have to explain things three times to make sure they don’t lose the plot. Anyone else out there dealing with this phase of infinite patience?

Looking forward to hearing your thoughts and happy to finally be a part of the community!

Accepted answers

Martin Runge

Hi @Alvaro Aparicio

Welcome to the Atlassian community, and thanks for a very honest first post.

I would say your relationship with Rovo is complicated for the right reason. You are asking it to serve as an oracle for your entire knowledge base, and that is the hardest possible starting point. We hit the same wall.

What changed things for us was spending most of our energy on data, context, and structure rather than on the tool itself, and then narrowing the use case. Today, Rovo is a good assistant for working on Confluence documents, and we use it to produce standardized drafts of statements of work. Narrow scope, clean source material, a human reviewing the output. Trust grows much faster when people see it save real time on one task rather than half-working on twenty.

On the noise problem, I would not try to clean everything. I would pick the spaces that matter, archive what is stale, and accept that everything else stays out of scope for now.

What does your first narrow use case look like? Sometimes picking one and doing it well beats a broad rollout.

Cheers, Martin

All comments

Martin Runge

Hi @Alvaro Aparicio

Welcome to the Atlassian community, and thanks for a very honest first post.

On the noise problem, I would not try to clean everything. I would pick the spaces that matter, archive what is stale, and accept that everything else stays out of scope for now.

What does your first narrow use case look like? Sometimes picking one and doing it well beats a broad rollout.

Cheers, Martin

Alvaro Aparicio_9g0djf

@Martin Runge

Completely agree. There’s a matter of expectations that wasn’t handled properly the first time round.

When it comes to technology, we tend to be overly ambitious and want to implement absolutely every possible approach to a solution.
Perhaps the problem was making Rovo available to any user without first working through use cases with controlled user segments.

To learn, you have to make mistakes

Dr Valeri Colon (Connect Centric)

@Alvaro Aparicio I think you identified the adoption issue in your follow-up: availability and adoption are two different things.

I’ve had better results treating AI adoption as a progression rather than a launch. Start with a controlled group and a specific workflow where success can be measured. Give users guidance on when to use Rovo, what sources it should rely on, and when human verification is expected. Then use feedback and usage data to improve the workflow before expanding.

You also don’t need to clean all of Confluence first. For custom agents, Rovo supports narrowing knowledge to specific sources, which can reduce irrelevant context.

The goal early on isn’t “trust Rovo.” It’s helping users learn where Rovo reliably saves time, where it needs verification, and where it shouldn’t be used yet. Trust tends to follow demonstrated value.

Welcome!

It looks like you're new here. Sign in or register to get started.

Quick Links

Find more posts tagged with

jira

rovo-atlassian-intelligence