Rovo
Rovo, a new product from Atlassian, is an AI-powered assistant designed to enhance productivity and business collaboration. It provides advanced search, task automation, and centralizes organizational data within the tools teams already use. It makes easier to turn contextual information into actionable insights and informed decisions.
Rovo will be sold as a separate product available in the Enterprise and Premium offerings and will have flexible pricing per user.
The main features of Rovo are outlined below:
Rovo Search allows you to search in one place for information from various Atlassian products as well as external products like Google Drive and SharePoint. It is capable to understand natural language and answer questions by reading information from connected products. You can ask questions as you are talking like, “What is the vacation policy?”.
Also, it allows improving searches by filtering according to the products in which you want to search. This functionality will help you to have quicker access to information and more efficient searches.
Example of use case:
A healthcare company with multiple hospitals and clinics manage a high number of patients and has information stored in various sources, including patient medical records, procedural guides, research papers, etc. The challenge it faces is to efficiently manage and find information across the various sources.
Before ROVO, for employees to find information from each source or specific information whose location is unknown, they need to log into each source and search for the information independently, resulting in multiple searches per source. Now thanks to ROVO, this can be reduced to a single search.
This is achieved through ROVO’s ability to understand natural language and connect to the different information sources of the hospital. Additionally, the search filters make it easier to find what is being sought. Rovo Search drastically reduces the time needed for complex searches, delivering results in just minutes.
ROVO Chat allows having an assistant available at all times, capable of understanding and interpreting what is said. It has access to the company’s information stored in Jira, Confluence, and even some external sources. Stay assured that the access to information is controlled, ensuring users can only see what they are authorized to view in each application.
By having the company context, Rovo Chat is very useful in quickly answering questions such as:
- What tickets did I work on yesterday?
- Which tickets should I work on today?
- Tell me the changes made to this page in the last three months.
- Summarize the content of this page.
- What is the main topic of this page?
- Summarize the comments on this page and tell me which ones are positive.
- Which page has François Bertin been working on lately?
It can even answer questions about how to do something in Jira and Confluence, such as:
- How can I change the project leader of my project?
- How can I modify the workflow of my project?
- Write for me a script in ScriptRunner to perform X action.
Additionally, it can answer general questions similar to those you might ask an AI like ChatGPT.
Example of use case:
A technology company is growing rapidly and therefore needs to onboard and train new employees quickly. This company has a large knowledge base that contains everything an employee needs to know. However, new employees find it quite difficult to locate the information from the large knowledge base.
This is where Rovo Chat comes in to play a crucial role, as it can answer employees’ questions, either provide the information directly or guide them to where the information is located.
Thus ROVO Chat not only speeds up the integration process but also reduces the load on the human resources team.
Rovo Agents are customizable assistants designed to focus on specific tasks. They can handle repetitive duties, provide personalized responses, or specialize in certain areas. Users can interact with these agents via chat and create their own agents based on their needs.
When creating an agent, several parameters are defined:
- Name: The name by which users will see the agent.
- Description: A phrase that describes to users what the agent does.
- Instructions: A space for users to provide context and instructions to the agent on what it should or should not do.
- Conversation Starters: Examples of questions a user might ask the agent.
- Knowledge: A space to specify the sources of information the agent should use. For example, you can restrict the Confluence spaces the agent can use to provide more specific results.
- Actions: A section where users can add extra capabilities to the agent, such as creating, deleting, or assigning issues, adding comments, performing transitions, and many more.
In addition to allowing users to create new agents, Rovo offers several preconfigured agents. Here are three examples:
- Decision Director: Use the DACI framework to help the user to communicate decisions clearly.
- Product Requirements Expert: Create and review product requirement documents to assist product and engineering teams.
- User Manual Writer: Help the users to create personal user manuals to share work preferences with their team.
Additionally, using Forge (Cloud app development platform) alongside Rovo enables the creation of highly customizable and powerful agents capable of performing more complex tasks.
Example of use case:
Building on the earlier use case for a tech company needing to onboard new employees quickly, Rovo Agent allows you to create an agent dedicated to onboarding, with access only to the necessary information. This approach enables the agent to guide users more efficiently and accurately. Additionally, you can tailor the agent’s instructions to customize the onboarding experience based on the employee’s position or the location they’re joining.
To sum up:
- Rovo is an easy-to-use tool that facilitates quick information retrieval and acts as a context-aware assistant that enhances the quality of responses.
- Rovo can work in basic and more advanced versions:
- If you’re just using the basic chat or Atlassian’s default agents, no prior setup is required.
- If you want to create a specialized agent, more time will be required to write the necessary instructions. The good news is that you can write these instructions in natural language; you don’t need to use any programming syntax.
- It’s possible to go even further: by using Forge and Rovo, you can develop advanced agents with capabilities. This, however, demands programming skills and effort.
- One of Rovo’s most promising capabilities is its ability to access information from external sources to answer questions. Right now, it connects with Google Drive and SharePoint, and Atlassian is working on adding more options.
JSM Virtual Agent
Similar to other Atlassian Intelligence features, JSM Virtual Agent is accessible on Premium or Enterprise Cloud editions. It provides a pleasant experience for both the agents and those seeking help, offering fast and conversational support. The conversation can take place on the customer portal or on Slack.
Before users create issues for simple questions, Virtual Agent can try to address by providing configurated answers, thus saving agents’ time and allowing them to focus on more critical tasks. In the cases where the Virtual Agent cannot resolve, an issue will be created automatically based on the selected request type and using the first sentence as the summary and the description. Kindly note that when users initiate conversations with greetings, agents can receive multiple context-less issues named simply “Hi!”.
Scenario 1 :
Scenario 2 :
Virtual Agent Configuration
For each potential topic, an intent needs to be configurated. Intents are like a workflow for the questions asked to virtual agents.
The project admin creates new intents and in the project settings: the usage of every intent, the match rates, the resolution rates and the customer satisfaction rates are visible.
There are multiple intent templates, such as Confluence – General, VPN – Troubleshooting, Keyboard request, etc. For every template, there is a list of possible questions, and the answers must be provided during the configuration. Thus, the initial configuration process may require a significant amount of time for this first setup.
It is possible to add new steps to the flow, whether to send message, to offer choices or to ask for information. The next reply will depend on the answer.
For every branch there are two possible endings : either the the customer is satisfied with the provided answer so the conversation is resolved, or an issue is created.
If a customer’s message doesn’t match an intent, the virtual agent can use project’s knowledge base and search the answer online Note that this feature needs to be activated on Virtual Agent settings.
To sum up:
+ Time saver for the agents.
+ Simple UI for the customers – possibility to use it on JSM portal or on Slack
– The initial configuration process is very long because every possible answer must be configurated.
– The agents may find a lot of issues without enough details as Virtual agent uses the first sentence as issue summary and description.
– Atlassian is switching to consumption-based pricing limits for Virtual Agent. The limitation is 1000 assisted conversations per month, more than which will require an extra payment.
“Change tone” prompts
Atlassian Intelligence offers some pre-defined prompts and it is possible to change tone to :
• Casual
• Educational
• Empathetic
• Neutral
• Professional
Example of use case :
The ticket is open for days, a ping-pong is going on between the customer and the agent because the agent can’t get the necessary information from the customer all at once. The agent gets frustrated and starts using a not-so-polite language with the customer in the comments. With the help of the Change tone prompts, the reply updates to be still professional, neutral, or empathetic, which ensure a quality exchange with the customer.
JSM Smarts
In addition to the JSM features such as search bar from Help Center (which provides relevant knowledge base articles and related request types from the requests catalog), JSM Smarts makes life easier for agents by identifying similar requests and similar incidents by analyzing the category, summary and description fields.
Example of use case :
A recently onboarded agent will handle his first issue. As similar requests are mentioned in the issue, the agent can review other requests as an example without wasting time searching for them.
AI Summarize
This Atlassian Intelligence feature summarize the description and the comments of the issue.
Example of use case :
In the absence of your colleague, you have to take over as the backup when an old issue is reopened. Using the Summarize feature will allow you to efficiently grasp the key points without having to go through each individual comment. This way, you can quickly familiarize yourself with the context and the actions previously undertaken.
To read more about Atlassian Intelligence features, particularly for those we didn’t explore here, please visit Atlassian Intelligence features.
