ChatGPT Integration for User Story Estimation - Backend

[Dan1elBox]

IMPORTANT NOTE: This issue/feature request has not yet received approval from the project owner and should not be worked on at this time!!!

Feature Request:
I suggest integrating ChatGPT into our Planning Poker app to enhance user story estimation.
This feature will allow users the option to activate the ChatGPT Integration before a session, enabling ChatGPT to assist in estimating selected user stories.
Prior to activation, a clear and explicit warning should be presented to prevent the inadvertent sharing of sensitive project data with third parties(OpenAI..).
This integration, combined with selective activation and data sensitivity warnings, will provide users with a more tailored and accurate user story estimation experience while leveraging ChatGPT's capabilities effectively.

To integrate ChatGPT into our Spring Boot backend, we will probably need the following components:

• API Key or Token: You will generally require an API key or authentication token to access the ChatGPT API.
• REST Controller: Develop a REST controller that receives requests from our Vue frontend and forwards them to the service class.
• Service Class: Create a service class responsible for communicating with the ChatGPT API. This service class should send requests to the API and handle responses, possibly including data processing logic.
• Security: Consider security aspects. In case the feature has been disabled due to project-sensitive data, it must be ensured that no data is sent to ChatGPT.
• Documentation: Document how the integration works and guide developers/users on utilizing the ChatGPT feature within our web application.
• Testing: Write tests to ensure proper functionality of the integration and to monitor for future changes.

Consider whether it makes sense to simulate ChatGPT as a separate user for the ChatGPT integration in order to utilize the WebSocket mechanism.

The most critical and error-prone aspect is likely to be reliably extracting the correct values from the ChatGPT response in the service class and then converting them into the corresponding planning poker card within the previously selected card set.

Tip: It would be beneficial to provide ChatGPT with the chosen card set and its minimum and maximum values for this planning poker feature, enabling it to contextually associate the stories correctly.

(This feature is quite extensive and could potentially be split into smaller components if necessary)