Currently, the Similar Cases capability is available in the English language only.

Yes, Similar Cases can be enabled for some selected users.

At least 25K cases in the English language are required for AI-Model training. You can decide which cases need to be considered for AI-model training basis some parameters such as case status or any custom field. By default, cases that are in "Closed", "Resolved" status in the last 1 year are considered.

You need to wait until the required number of cases gets accumulated in your environment.

The timeline to get Similar Cases implemented is ~3 weeks.

Yes, there is a standard reporting dashboard to track the adoption and feedback around Similar Cases.

The following key functionalities can be accessed in Similar Cases.

1. Smart Assist Search Bar - Care agents can manually search for similar cases.

2. Feedback Buttons - Agents can provide feedback on cases predicted using the thumbs up/down buttons.

The brands can choose which cases to consider for AI model training.

For example:

1. Cases in a particular status (Closed, Resolved).

2. Cases with a particular custom field value.

3. Cases from a particular time period, etc.

Since the AI model is based on the unsupervised learning approach there is scope for improvement, and constant feedback can help make the model more accurate.

Feedback for similar cases can be given via the thumbs up/down button on the case cards which can be selected as follows:

Thumbs up when a particular recommendation is correct and relevant to the current case.

Thumbs down when no recommendations are relevant in context with the current case.

Once there is a sufficient number of feedback responses received, we will internally do a quality check on the feedback before re-training the model. So ideally, we can re-train a model once in a quarter, provided we have sufficient feedback to incorporate (to be decided by the product team).

Also, the feedback can be given by any user who has the permission to view similar cases suggestions.

Model retraining will be taken on a case-to-case basis depending upon the type and quantity of feedback given by the brand. The product team will discuss with the brand and decide whether model retraining is necessary or not.