Anvil: Revolutionizing Team Collaboration with Sentiment Analysis

Inspiration

The inspiration for Anvil came from the realization that team communication, especially in large projects, often involves a significant amount of unstructured data (comments in Jira and Bitbucket). Despite the abundance of information, it can be challenging for team leaders to extract actionable insights that reflect the true sentiment of the team. We wanted to create an app that could analyze team comments, gauge emotional tones, and provide valuable insights to improve team collaboration and decision-making.

What it does

Anvil is a Forge-powered app designed to analyze team comments in Jira and Bitbucket using advanced sentiment analysis. It turns unstructured feedback into meaningful insights, enabling teams and leadership to stay aligned and proactive. By summarizing key points and emotions from comments, Anvil helps to highlight trends, identify potential issues, and drive better communication. It integrates with Power BI and Tableau, allowing for detailed, interactive dashboards that visualize sentiment over time, project health, and team morale.

How we built it

We built Anvil using Atlassian Forge for seamless integration with Jira, Bitbucket, and other Atlassian products. The app’s backend uses OpenAI’s GPT-4 model for sentiment analysis and comment summarization. We also integrated Power BI and Tableau** using their respective APIs to push processed data into interactive dashboards for leadership and teams. The app works by fetching comments from Jira and Bitbucket, analyzing the sentiment and summarizing the content, and then pushing the results to Power BI or Tableau for visualization.

1. Data Extraction: We used the Jira and Bitbucket APIs to fetch comments from issues and pull requests.
2. Sentiment Analysis: Leveraged OpenAI’s GPT-4 for sentiment analysis and to generate summaries of the comments.
3. Integration: Anvil pushes the sentiment data into Power BI and Tableau using their REST APIs to create visual reports and dashboards.

Challenges we ran into

• API Rate Limiting: Both Jira and Bitbucket have rate limits for API requests, which made it challenging to fetch a large number of comments in a short time. We had to carefully manage API calls and implement caching to mitigate this issue.
• Accuracy of Sentiment Analysis: While GPT-4 does an excellent job at sentiment analysis, we faced challenges in fine-tuning the model to be highly accurate for specific contexts, like technical jargon or ambiguous comments.
• Data Integration: Ensuring smooth data transfer between Anvil, Power BI, and Tableau was tricky. We had to handle various data formats and ensure that the API requests were made correctly, especially for large data sets.

Accomplishments that we're proud of

• Real-time Data Sync: Implementing webhooks for real-time data updates from Jira and Bitbucket was a significant achievement. It ensures that any new comments are processed and visualized immediately.
• Advanced Sentiment Analysis: The integration of GPT-4 for sentiment analysis is a standout feature of Anvil, providing in-depth emotional insights that go beyond simple positive/negative feedback.
• Seamless Integration with Power BI and Tableau: Building a bridge to easily push data to Power BI and Tableau has created powerful, interactive dashboards that give users actionable insights about team morale and project health.

What we learned

• API Integration: We learned a lot about integrating third-party APIs, especially working with Jira and Bitbucket APIs for data extraction and ensuring smooth communication with the Power BI and Tableau APIs.
• Sentiment Analysis: We gained deeper knowledge of how NLP models, particularly GPT-4, can be applied to real-world use cases like sentiment analysis, and how to fine-tune them for specific needs.
• Forge Platform: Using the Atlassian Forge platform gave us a great understanding of how to build scalable apps with easy integration into Atlassian products.

What's next for Anvil

• Enhanced Emotional Categorization: We plan to extend the sentiment analysis capabilities to categorize emotions more granularly (e.g., frustration, excitement) and use these insights to create tailored action plans for teams.
• Broader Tool Integration: We aim to integrate Anvil with other popular collaboration tools like Slack and Microsoft Teams to send real-time sentiment alerts and notifications.
• AI-powered Recommendations: Based on sentiment trends, we want to develop AI-powered recommendations for improving team communication, addressing concerns, and boosting morale.
• Mobile Support: We plan to develop a mobile-friendly version of Anvil to allow teams to access sentiment insights and dashboards on the go.

We believe that Anvil has the potential to revolutionize the way teams communicate and collaborate, helping them stay proactive and aligned at all times.

Built With

• allowing-for-advanced-visualizations-and-reporting.-**apis**:-**jira-api**:-used-to-fetch-issue-comments-from-jira
• and-confluence.-**node.js**:-employed-to-build-the-backend-logic
• and-run-server-side-processes-within-the-forge-app.-**cloud-services**:-**openai**:-leveraged-the-gpt-4-api-for-sentiment-analysis-and-text-summarization
• and-team-morale.-**tableau**:-used-tableau's-rest-api-to-push-processed-sentiment-data-into-tableau
• api
• atlassian
• bi
• bitbucket
• csv
• displaying-sentiment-trends
• enabling-smooth-integration-with-jira
• forge
• handle-api-calls
• javascript
• jira
• json
• leveraging-the-power-of-the-forge-platform.-**python**:-utilized-for-backend-processing
• node.js
• oauth
• openai
• particularly-for-sentiment-analysis-and-handling-api-integrations-with-power-bi-and-tableau.-**platforms**:-**atlassian-forge**:-the-platform-used-to-build-and-deploy-the-app-seamlessly-into-the-atlassian-ecosystem
• postgresql
• power
• project-health
• providing-in-depth-emotional-insights-and-context-aware-summaries.-**power-bi**:-integrated-with-power-bi-to-create-interactive-dashboards
• python
• rest
• tableau
• webhooks