[accuracy] More control over the message samples used for training (e.g after tag deletion)

Case: after tag deletion

Situation:

Currently, when an admin deletes an AI tag from an AI tagging rule, the associated message samples used for training are also removed. If the admin wishes to add the tag back, they must go through the process of retraining the AI tagging system with new message samples. This can be a positive or negative outcome based on the admin desire.

• Positive is they wanted to redo the training to improve/change the accuracy

• Negative if they made a mistake and just wanted to have the tag back because in that case, it's time-consuming

Request:

Provide an option that allows admins to retain the message samples used for training when an AI tag is added to a tagging rule and has already been trained. Additionally, admins should have the flexibility to decide whether to reuse existing message samples or retrain the AI from scratch, depending on their needs. This would apply both when an AI tag is deleted and when admins opt to retrain the AI for other reasons.

Impact and Benefit:

Implementing this feature would reduce the time and effort required to manage AI tagging rules. Admins would have greater control over training data, improving workflow efficiency by reusing valuable message samples without the need for constant retraining. This flexibility would enable faster iterations on AI tagging and improve overall user satisfaction with the AI feature.