Rasa-Powered Chatbot: Seamless Confluence & Jira Integration Guide

[email protected] Rai 01 Dec, 2023 • 9 min read


Chatbots have evolved into sophisticated conversational agents driven by artificial intelligence in recent years. This guide delves into constructing an advanced Rasa-powered chatbot specifically tailored to address user queries related to Confluence pages and Jira tickets. Integrating Confluence and Jira brings substantial benefits, streamlining information retrieval and fostering a cohesive work environment. Confluence facilitates collaborative documentation, while Jira is a robust project management tool. By creating a chatbot that seamlessly integrates with these platforms, accessibility is enhanced, and efficiency is optimized for teams collaborating on content and managing projects.

Rasa-Powered Chatbot

Learning Objectives

In this article, you will learn:

  • Rasa Project Setup: Learn to initiate a Rasa project, establishing the groundwork for advanced chatbot development.
  • NLU Intent Definition: Define specific Natural Language Understanding (NLU) intents for Confluence and Jira queries, enhancing the chatbot’s understanding.
  • Custom Action Development: Create Python-based custom actions to interact with Confluence and Jira APIs for expanded functionality.
  • Model Training and Testing: Understand the model training process, ensure the chatbot’s generalization, and employ iterative testing for continuous improvement.

This article was published as a part of the Data Science Blogathon.

Foundational Concepts


Rasa, an open-source conversational AI platform, empowers developers to build robust, context-aware chatbots. Beyond simple rule-based systems, Rasa utilizes machine learning to comprehend and respond to complex user inputs. Its natural language processing capabilities and dialogue management tools make Rasa a versatile solution for creating intelligent conversational agents.


Jira, developed by Atlassian, is a renowned project management and issue-tracking tool. Widely used in agile software development, Jira facilitates efficient collaboration by organizing tasks, tracking issues, and enabling teams to streamline their workflows. Its extensive features, such as customizable workflows and real-time collaboration, contribute to Jira’s popularity among development teams and project managers. Jira’s extensive RESTful APIs allow seamless integration with external tools and applications, facilitating real-time data exchange and automation.


Confluence, again developed by Atlassian, is a collaborative platform facilitating efficient documentation, knowledge sharing, and teamwork within organizations. It is a centralized space for teams to create, share, and collaborate on content, making it an essential tool for project documentation, meeting notes, and general knowledge management. Real-time collaborative editing allows multiple team members to simultaneously work on the same document. With Confluence’s robust search capabilities, finding relevant information is efficient. Confluence integrates seamlessly with other Atlassian products like Jira, creating a unified project management and documentation ecosystem.


Chatbots have become integral to modern digital interactions, providing instant and personalized responses. Fueled by artificial intelligence, chatbots interpret user inputs, understand context, and deliver relevant information or actions. From customer support to process automation, chatbots transform how businesses engage with users, enhancing efficiency and user experience. Chatbots utilize natural language processing to identify user intent, enabling them to respond contextually and accurately. In the context of Rasa, custom actions are Python functions that extend the chatbot’s functionality, allowing it to perform tasks beyond simple intent recognition.


Before we dive into the development process, let’s ensure we have the necessary tools and access:

Python and Virtual Environment

  • Make sure you have Python installed. Create and activate a virtual environment using:
# Command prompt (Windows) or terminal (macOS/Linux)
python -m venv myenv
# On Windows
# On macOS/Linux
source myenv/bin/activate

Install Rasa

  • Install Rasa Open Source using:
# Command prompt (Windows) or terminal (macOS/Linux)
pip install rasa

Confluence and Jira Access:

  • Ensure you can access Confluence and Jira and the necessary permissions to fetch information through their APIs.

Confluence API

  1. Generate Personal Access Token (PAT):
  • Log in to your Confluence instance.
  • Navigate to your profile picture at the bottom left, then choose “Personal settings”.
  • Click on “Create and manage API tokens”.
  • Create a new API token and keep it secure.

2.  Retrieve space key: You’ll need the space key to fetch information about Confluence pages. You can find it in the URL when you navigate to a Confluence space.3.  API endpoint: The Confluence API endpoint will be based on your Confluence instance URL. For example, if your Confluence is hosted at https://your-confluence-instance.com, the API endpoint might look like https://your-confluence-instance.com/rest/api/content.

Jira API

1. Generate API token:

  • Log in to your Jira instance.
  • Navigate to “Account settings” or “Security” (exact location may vary based on your Jira version).
  • Create a new API token.

2. API endpoint:

  • The Jira API endpoint is usually based on your Jira instance URL. For example, if your Jira is hosted at https://your-jira-instance.com, the API endpoint might look like https://your-jira-instance.com/rest/api/latest/issue.

Building the Chatbot

Building the chatbot | Rasa-Powered Chatbot

Create a New Rasa Project

Setting up a Rasa project involves creating a directory structure and initializing the project. This step is foundational to organizing your chatbot development process. The directory structure organizes various aspects of your chatbot, from training data to custom actions. Understanding the purpose of each directory is crucial for maintaining a well-organized and scalable project. This organization enhances project management and provides a clear roadmap for future expansions and improvements.

# Command prompt: Create a new Rasa project
mkdir my_rasa_project
cd my_rasa_project
rasa init

Define NLU Intents for Confluence and Jira

For our chatbot to understand user queries about Confluence pages and Jira tickets, we need to define Natural Language Understanding (NLU) intents. Provide examples of queries related to Confluence and Jira in the NLU training data. The richness of your training data directly impacts the chatbot’s ability to interpret and respond to user queries accurately. For instance, consider adding variations such as “Can you provide information on Confluence?” or “Tell me about Jira ticket statuses.” This enhances the model’s ability to generalize and accommodate different phrasings users might employ.

# YAML file representing the training data for the NLU component of a Rasa chatbot
# data/nlu.yml
version: "2.0"

- intent: query_confluence
  examples: |
    - Tell me about Confluence
    - How can I find information on Confluence?
    - What is a Confluence page?

- intent: query_jira
  examples: |
    - How do I check my Jira tickets?
    - Tell me about Jira ticket status
    - What are my open Jira issues?

Understanding the nuances of user input is paramount. By providing a robust set of examples, you empower your chatbot to handle a wide array of queries effectively. Consider incorporating additional user personas and scenarios to ensure the versatility of your chatbot in real-world usage.

Create Custom Action for Confluence and Jira

Custom actions in Rasa allow us to extend the functionality of our chatbot. Create a custom action (actions.py) to handle queries about Confluence pages and Jira tickets. This action will interact with the respective APIs to fetch information. It’s essential to delve into the intricacies of API interactions, error handling, and response processing to ensure the reliability and resilience of your chatbot.

Expand your custom actions to handle a broader range of scenarios. For example, you can implement a response strategy for cases where Confluence pages or Jira tickets are not found. This proactive approach enhances the user experience and provides more meaningful interactions.

# Python file: Custom action for Confluence and Jira
# actions.py
import requests
from typing import Any, Text, Dict, List
from rasa_sdk import Action, Tracker
from rasa_sdk.executor import CollectingDispatcher

class ActionQueryConfluenceJira(Action):
    def name(self) -> Text:
        return "action_query_confluence_jira"

    def run(
        self, dispatcher: CollectingDispatcher, tracker: Tracker, domain: Dict[Text, Any]
    ) -> List[Dict[Text, Any]]:
        # Replace with your Confluence and Jira details
        confluence_api_url = 'YOUR_CONFLUENCE_API_URL/rest/api/content'
        jira_api_url = 'YOUR_JIRA_API_URL/rest/api/latest/issue'
        confluence_space_key = 'YOUR_CONFLUENCE_SPACE_KEY'
        personal_access_token = 'YOUR_PERSONAL_ACCESS_TOKEN'

        user_query = tracker.latest_message.get('text')
        headers = {'Authorization': f'Bearer {personal_access_token}'}
# Handle Confluence Query
        if tracker.latest_message['intent']['name'] == 'query_confluence':
            params = {'spaceKey': confluence_space_key, 'cql': 'text~"{}"'.format(user_query)}
            response = requests.get(confluence_api_url, headers=headers, params=params)
        if response.status_code == 200:
                results = response.json().get('results', [])       
        if results:
                    result = results[0]
                    title = result.get('title', 'No Title')
                    url = result.get('_links', {}).get('webui', '')
response_message =  "I found information about '{title}'. You can find it [here]({url})."
 response_message = "I couldn't find any information on that topic in Confluence."
 response_message = "I'm sorry, but there was an issue fetching information from Confluence." 
# Handle Jira Query
elif tracker.latest_message['intent']['name'] == 'query_jira':
 params = {'jql': 'text~"{}"'.format(user_query)}
 response = requests.get(jira_api_url, headers=headers, params=params)
if response.status_code == 200: result = response.json()
 issue_key = result.get('key', 'No Key')
 summary = result.get('fields', {}).get('summary', 'No Summary')
 response_message = f"I found information about the Jira issue '{issue_key} - {summary}'."
 response_message = "I'm sorry, but there was an issue fetching information from Jira."
 response_message = "I'm not sure how to handle that query."
return []

Continuously refine and expand these actions to cater to evolving user needs and new integration possibilities.

Configure Rasa

Configuring your Rasa project is pivotal in determining the chatbot’s behavior. The config.yml file contains settings for training the NLU and dialogue management models. Explore different configurations and experiment to optimize your chatbot’s performance based on your unique requirements.

The configuration file serves as the control center for your chatbot’s behavior. Delve into the various parameters, experiment with different machine learning pipelines, and fine-tune settings to achieve the desired balance between accuracy and efficiency.

# config.yml
language: "en"
  - name: "WhitespaceTokenizer"
  - name: "RegexFeaturizer"
  - name: "CRFEntityExtractor"
  - name: "CountVectorsFeaturizer"
  - name: "EmbeddingIntentClassifier"

Train the Rasa Model

Training the Rasa model is crucial in preparing your chatbot for real-world scenarios. The training process involves exposing the model to your training data, enabling it to learn and generalize from examples. Regularly revisit and update your training data to improve the model’s accuracy over time.

Consider implementing a robust versioning system for your training data. This ensures traceability and facilitates the identification of specific datasets that contribute to improvements or challenges in your chatbot’s performance.

# Command prompt (Windows) or terminal (macOS/Linux)
rasa train

Run Rasa Action Server

The Rasa action server handles the execution of custom actions. Running a dedicated action server ensures the responsiveness of your chatbot, allowing it to perform complex tasks, such as fetching real-time data from Confluence and Jira.

Explore the scalability of your action server as your chatbot gains momentum. Consider load balancing and redundancy options to ensure a seamless user experience even during peak usage.

# Command prompt: Run the Rasa action server
rasa run actions

Interact with Your Chatbot

With the Rasa action server running, open a new terminal and start the Rasa shell. This interactive shell facilitates communication with your chatbot, enabling you to test various scenarios and assess its responses to different queries. Iterative testing is essential for identifying areas for improvement and refinement.

rasa shell

Sample output:

User: What's the status of our current project?
Chatbot: The current project is progressing well. We have completed 80% of the tasks.
User: Can you provide information on our project documentation in Confluence?
Chatbot: The project documentation is available on Confluence. 
You can find detailed information, [here](link-to-confluence-page).
User: Create a new Jira issue for the upcoming sprint planning.
Chatbot: A new Jira issue for the upcoming sprint planning has been created successfully. 
The issue Id is PROJ-123. 
Anything else you'd like to ask?

As you refine your chatbot, consider exploring advanced features such as entity recognition for extracting specific information from user queries.

Rasa-Powered Chatbot

Use Cases

The integration of Confluence and Jira into a Rasa-powered chatbot unfolds a spectrum of compelling use cases, redefining collaboration and project management:

1. Project management: Chatbots integrated with Jira can provide real-time updates on project status, issue tracking, and sprint progress. You can inquire about specific Jira issues, upcoming deadlines, and team assignments.

2. Knowledge base access: Confluence-integrated chatbots enable quick access to documentation, FAQs, and project-related knowledge.

3. Automated Reporting: Chatbots can generate automated reports on project milestones, task completion, and overall team productivity. Improves efficiency by reducing manual efforts and streamlining task management processes.

Risks and Considerations

Ensure secure handling of personal access tokens and API keys to prevent unauthorized access to Confluence and Jira data. Be cautious about the type of information shared via the chatbot, especially when it involves sensitive project details or user data. Users may need guidance on how to interact with the chatbot effectively. Provide clear instructions and examples to enhance user experience. Implement robust error-handling mechanisms in your custom actions to gracefully manage situations where Confluence pages or Jira issues are not found.


In conclusion, this comprehensive guide navigates the journey of constructing an advanced Rasa-powered chatbot tailored to address user queries related to Confluence pages and Jira tickets. The article illuminates the substantial benefits of streamlining information retrieval and fostering a cohesive work environment by exploring the integration of Confluence and Jira into the chatbot. From setting up a Rasa project and defining NLU intents to developing custom actions for API interactions, you gain a holistic understanding of the chatbot’s creation process. The guide emphasizes configuring Rasa for optimal performance, training the model with diverse examples, and iterative testing for continuous refinement. This approach not only enhances accessibility for collaborative content creation and project management but also lays the foundation for further exploration and customization in the evolving landscape of conversational AI.

Key Takeaways

  • Gain practical insights into building an advanced chatbot using Rasa, focusing on tailored applications for Confluence and Jira integration.
  • Understand the benefits of integrating Confluence and Jira, streamlining information retrieval, and fostering a cohesive and efficient work environment.
  • Explore configuration techniques to optimize the Rasa project for improved Natural Language Understanding (NLU) and dialogue management.
  • Embrace iterative testing and model training for continuous refinement, ensuring the chatbot’s adaptability and effectiveness over time.

Frequently Asked Questions

Q1: Can I use this guide if I’m new to chatbot development?

A: Yes, this guide is structured to provide step-by-step instructions, making it accessible to beginners and those familiar with chatbot development.

Q2: Do I need admin access to integrate the chatbot with Confluence and Jira?

A: Admin access is not necessary for basic integration. However, certain advanced features may require specific permissions for API access.

Q3: Can I adapt the chatbot for other platforms besides Confluence and Jira?

A: Yes, the principles discussed can be adapted for other platforms with similar APIs, allowing for flexibility in integrating the chatbot into various environments.

Q4: What programming skills do I need for custom action development?

A: Basic Python skills are beneficial. The guide provides code snippets and explanations to help users understand and adapt custom actions for their needs.

Q5: How can I gather user feedback for continuous improvement?

A: Encourage users to provide feedback during testing. Additionally, explore Rasa’s capabilities for user feedback collection, aiding in the iterative refinement of the chatbot.

The media shown in this article is not owned by Analytics Vidhya and is used at the Author’s discretion. 

Frequently Asked Questions

Lorem ipsum dolor sit amet, consectetur adipiscing elit,

Responses From Readers

  • [tta_listen_btn class="listen"]