At Seibert, experimentation with Atlassian’s AI-powered Rovo has taken many shapes. We’ve been building multiple helpful agents internally, exploring the full spectrum of what’s possible, from quick no-code helpers to fully customized Forge apps.
In this short article, I’ll go over the Rovo agents Seibert has created, talk about why it’s worth trying them out, and offer ways you can get your hands on these agents too.
What the difference between no-code, custom-action, and forge-built Agents?
Rovo agents basically come in three different forms, and it’s important to know how each one works. First, no-code agents can be built by anyone without programming knowledge, but their functionality is limited to the capabilities Atlassian provides. Then, agents with custom actions allow connections to third-party services and enable more advanced workflows, but they require an app and developer involvement. Lastly, fully coded Forge agents provide complete flexibility and control within Atlassian, though they are more complex to build and cannot use the built-in no-code actions.
Experimenting with different approaches
The journey started with no-code agents, simple setups that anyone on the team could configure without touching code. From Agile retro helpers that pulled in insights from Jira and Confluence, to backlog analyzers for refining work items, these early experiments showed how quickly value could be created with minimal effort.
From there, the experiments grew more ambitious. Using custom actions, agents were connected to third-party tools, extending their usefulness beyond Atlassian products. And finally, Forge-based agents pushed the boundaries even further, offering full control over behavior and integrations, while requiring proper development effort.
A closer look at the Rovo agents built
So far, about 15 Rovo agents have been created internally, spanning a variety of use cases and technical approaches. Here are some of the key examples:
-
Agile Helper Agents (no-code): Designed to surface sprint topics from Jira and Confluence, and combined with Gemini summaries of Google Chat to give teams a complete picture for retrospectives.
-
Backlog Analyzers (no-code): Used for refining and improving backlog items, helping teams better prioritize and manage their work.
-
Jira Service Management Agents (no-code): Virtual service agents that answer customer questions instantly using a connected knowledge base (like Confluence), reducing ticket volume and improving response times.
-
Third-Party Integrations (no-code with Forge actions): Agents configured to connect Atlassian products with tools like Personio and Moco, enabling smoother HR and project management workflows.
-
Feedback Agents (Forge-based): Fully coded Forge apps, including those showcased during Codegeist, providing advanced customization and control over agent behavior.
-
Templating.app Integration: Templating.app allows Jira users to create reusable issue templates, including Epic hierarchies and subtask templates. In large companies, the sheer number of templates can make it difficult to find the right one. The Rovo agent built into Templating.app solves this by helping users quickly identify the template they need.
-
Internal Sales Enablement Agent (no-code): Built to help employees quickly find product information, battlecards, competitor comparisons, and discount policies. By limiting the agent to trusted Confluence spaces, the output quality is high and always cites sources.
-
Playbook Finder (no-code with Jira action): Focused on surfacing the right sales and marketing playbooks stored in Confluence. If no match exists, the agent offers to create a Jira issue for a new playbook request. Still being refined to avoid over-suggesting unrelated playbooks.
This variety of Seibert experiments illustrates how Rovo can scale. From quick, no-code wins to fully customized solutions that require deeper technical investment.
The pros and cons of each type
Through these iterations, some patterns emerged:
No-code agents
Pro: Accessible to everyone, no coding required.
Con: Limited to the built-in capabilities Atlassian provides.
Custom-action agents
Pro: Broader integrations and more flexible actions.
Con: Requires developer support and an app as a foundation.
Forge-based agents
Pro: Full control and customization.
Con: Can’t use built-in actions directly, and development is more complex.
Automating away manual work
The experiments with Rovo agents have already shown to reduce some manual processes—and in certain cases, even improved them. For example, searching and finding relevant content has become much faster, with most of the manual research now automated and delivered in concise summaries.
Instead of spending time gathering information, developers can focus on interpreting results and moving discussions forward. Rovo isn’t just about automation, and it’s also not about fully replacing human effort; rather, the agents act as collaborators that streamline repetitive tasks while leaving space for meaningful decision-making and teamwork.
Looking ahead – How Seibert can get Rovo to your work
With over a dozen agents already built, the experiments at Seibert are far from over. Each iteration adds new insights into how Rovo can support teams across Atlassian and beyond, whether it’s streamlining retrospectives, powering service agents in Jira Service Management, or connecting seamlessly to HR and project tools.
Seibert’s dedication to providing users with solutions that make using Atlassian products even better translates directly to AI experimentation. We not only create products fit for diverse teams, but also offer custom development and personalized Rovo agent creation.
We’ve been internally experimenting and finetuning agent creation for some time now. This dedication shows that we love mastering new developments in the Atlassian ecosystem, and we’re ready to reflect our internal knowledge on your processes. If you’re looking to implement Rovo agents for your work, find out more about Custom Development here.