If 2024 was a year of mass adoption of general-purpose chats powered by large language models (LLMs), then 2025 is predicted to be a year of highly specialized AI Agents. These Agents offer more focused capabilities and deeper integration within specific applications, promising to revolutionize our work.
An Agent can be defined as an AI assistant instructed to perform a specific task within your app. For instance, imagine an Agent that automatically generates a Confluence page with client details from Salesforce whenever a new opportunity is created in Jira. From a technical perspective, an Agent can be an autonomous entity within a software system that interacts with its environment to achieve specific goals.
Within the Atlassian suite, Rovo Agents will drive a new wave of automation for companies of all sizes, enabling AI to handle complex tasks that current rule-based automations struggle to complete and delivering advanced efficiencies to a wide range of users.
This article aims to explain the current AI capabilities of Atlassian, including Atlassian Intelligence, Rovo, and Rovo Agents, and how Agents can unlock hidden value by executing business tasks and workflows.
Atlassian Intelligence for smarter automation
Atlassian Intelligence is a set of AI features experienced by users across a range of Atlassian products that is included in Premium+ editions of plans. Once enabled on your organizational account, it provides features like content creation, summarization, insights, and automation, helping to enhance workflows and improve decision-making.
For instance, in Jira, Atlassian Intelligence can suggest relevant issues or provide insights into project progress, while in Confluence, it can help summarize lengthy
Core features and benefits of Rovo
While powered by Atlassian Intelligence, Rovo is a standalone product available as an add-on for all Atlassian plans. Rovo is billed per user and does not require Atlassian Intelligence to be enabled. Once you get Rovo, you can access Rovo Search, Rovo Chat, and Rovo Agents.
With Rovo Search, project managers can do a comprehensive search across Atlassian and third-party applications, such as Google Drive, Sharepoint, etc., helping them retrieve information quickly.
Rovo Chat, on the other hand, offers a convenient UI for interacting with data from different sources, asking questions, and getting insights and personalized answers based on the data it has access to. Support teams can use Rovo Chat to interact with data from various platforms and get personalized responses, enabling quicker and more accurate support.
Rovo Agents automate repetitive tasks, improving productivity for both individuals and teams. For developers, this means Rovo Agents can handle repetitive code-related tasks and track issue status across projects.
The difference between Rovo Chat and Rovo Agents
Agents differ from Chat by having greater functionality; Agents have access to more actions that can be performed. When you write a prompt in chat, a limited number of actions can be requested, such as “Add comment to page,” “Assign issue,” “Add comment to issue,” “Create issue,” “Transition issue”, “Create page,” and “Edit issue.“
Configuring Agents, however, allows you to select from a wider range of actions, including “Delete issues,” “Add to sprint,” “Create epic,” “Find similar incidents,” and more.
This wider range of actions allows Agents to automate more complex workflows. For instance, an Agent could be configured to automatically triage incoming support tickets, assign them to the appropriate team member, and suggest potential solutions based on similar issues in the past.
Who benefits from Rovo Agents and how they work
By design, Rovo Agents can be created by all users in the organization. This democratized approach increases Agent adoption across the organization, empowering teams to build custom AI solutions without requiring specialized development expertise.
Once created, Agents can be either public or private, meaning they can be discovered and used by other users in your organization or accessible only to you.
Different Agents can be used for different organizational roles. Project Managers can automate task assignment and project tracking, developers can automate issue creation and code reviews, saving time for actual coding, and support teams can improve response times by automating ticket triaging and assigning tasks to the appropriate Agents.
For Rovo Agents, this flow diagram above shows how Agents work to automate and improve interactions within apps like Jira. The Agent acts as a go-between, handling the complex finding and processing of data. Hence, users get helpful, data-driven answers without gathering information from multiple places. Using the LLM, Agents can understand and helpfully respond to user questions, even with unstructured data or complex requests.
No Code Agents
There are two types of Agents: No Code Agents and Custom Agents. No Code Agents imply you only use a prompt to instruct an Agent. As a user, you can create your Agent by giving it a name and writing a prompt. Then, specify a data source, such as Atlassian products and any other available out-of-the-box connectors for 3rd-party tools (e.g., Google Drive, SharePoint, etc.). You can also select actions that an Agent will take, such as “Create an issue,” “Create a page,” “Assign issue,” and more.
No Code, prompt-only Agents, are great for things like prototyping or doing simpler actions. When you build a No Code Agent, it can be used by all users in that particular cloud instance.
Custom Agents
On the other hand, Custom Agents are powered by prompts and your own Forge App. Forge is Atlassian's development platform for building Jira and Confluence Cloud apps. This means you can write your app with a more complex Agent logic that will leverage not only existing Atlassian products and connectors but also external data sources not yet available in Atlassian.
To achieve that, an organization can either develop an app using its own internal capabilities, use Atlassian partners' help, or have third-party Marketplace developers develop it.
With Custom Agents, you have more freedom than with No Code Agents since you have access to REST APIs for all Atlassian tools and third-party data sources. Once you build a Custom Agent, you can publish it in the marketplace or make it private to be shared by private links for one or more organizations.
What is not considered a part of Rovo
Virtual Service Agent is not considered part of Rovo since it has more deterministic workflows and supports defined use cases.
The Rovo GitHub Copilot extension enables interaction with Atlassian data within IDEs but is not considered a Rovo Agent. It’s a third-party integration that facilitates communication via APIs with external platforms like GitHub, DataBricks, etc.
Building Custom Agents with Forge for greater flexibility
Let’s say you want to pull data from an external source with a connector not currently listed in the Atlassian Marketplace, and you need an Agent to perform 4-5 steps to complete a goal.
It might not be as reliable if you try to achieve it with a No Code Agent, where you simply write a prompt and select actions for an agent. Every time it runs, an Agent interprets the prompt with specified actions. Still, there would not be a direct programmatic relation between a prompt and actions, which might result in inaccurate and inconsistent outputs, especially for complex queries.
To address this, building a Custom Agent with code could be a better approach. With Forge, you can use in-house connectors or any other third-party source, and Forge gives an extension point to connect to other data sources. This opens many more opportunities to access and use the data.
The current limit for information that can be included in a prompt, known as the context window, is 128k tokens. This is roughly equivalent to around 96k words, or the length of a medium-sized novel. This means we can include up to 128k tokens of information at once for the Agent to process. Exceeding this may lead to timeouts or partial responses.
You can upload your data source to Google Drive or another file-sharing service if your data source is a PDF. The Agent, powered by LLMs with Optical Character Recognition (OCR) capabilities, can then read the image and convert the text from the PDF into a machine-readable format.
Exploring the potential of Agents across industries
AI is already boosting productivity in enterprises, especially in software development and customer support, where LLMs are significantly improving developer efficiency and reducing human-handled contacts by 20-40%.
However, Rovo’s expanding capabilities, with plans to increase third-party data coverage targeting over 80+ external sources, have the potential to drive transformation far beyond these areas, impacting diverse sectors across an organization.
In finance, Rovo Agents could take over routine data entry and reporting tasks, helping to reduce errors and make daily work smoother. Marketing teams may use Rovo to gather and analyze customer insights from various sources, allowing them to create more targeted campaigns. In operations and human resources, Rovo Agents could handle workflows and automate document creation, ensuring consistency and freeing up time for other important tasks.
This is a unique moment for businesses of all sizes and sectors to integrate AI agents like Rovo into their operations. While large enterprises have already begun deploying these tools, Rovo’s expanding capabilities now allow any organization to benefit from increased productivity and smarter workflows.
As Rovo continues to develop, it provides a powerful solution that helps teams across finance, marketing, HR, and more to use AI in ways previously limited to larger companies. By adopting Agents now, businesses can create a work environment where AI and employees support each other, making everyday tasks smoother and more efficient across every role and department.
Published: Nov 26, 2024
