{"article":{"id":20525671750029,"url":"https://betterworks.zendesk.com/api/v2/help_center/en-us/articles/20525671750029.json","html_url":"https://support.betterworks.com/hc/en-us/articles/20525671750029-Conversation-AI-Assist","author_id":484570113,"comments_disabled":true,"draft":false,"promoted":false,"position":2,"vote_sum":0,"vote_count":0,"upvote_count":0,"downvote_count":0,"section_id":360002670632,"created_at":"2023-10-19T12:44:18Z","updated_at":"2025-02-13T19:38:29Z","name":"Conversation AI Assist","title":"Conversation AI Assist","source_locale":"en-us","locale":"en-us","outdated":false,"outdated_locales":[],"edited_at":"2025-02-13T19:38:12Z","user_segment_id":null,"permission_group_id":1692132,"content_tag_ids":[],"label_names":[],"body":"

Contents

\n\n

Overview

\n

Note: Our AI features use the Betterworks Private LLM. It is our own self-hosted LLM contained in the Betterworks AWS server. This keeps our customer data private and secure without the risk of it going over API to a third party. Organizations are not enrolled automatically. To enroll, please contact your Program Architect or Customer Success Manager.

\n

The Conversation AI Assist feature is designed to assist users write open-text answers to conversation questions. While typing or after completing their responses, users can click an arrow to expand the AI Assist panel and access the feature.

\n

This feature uses artificial intelligence to review and refine user responses, ensuring grammatical correctness, adopting a professional or personal tone, and eliminating biases. The AI suggestion also includes an explanation for the suggested changes.

\n

Users can accept and apply the AI suggestion, ask the AI to generate another suggestion immediately or modify their original response before asking the AI to generate another suggestion. 

\n

Usage

\n

Step 1: The user is prompted to provide an open-text answer in the conversation template.

\n

Note: A tooltip will indicate that at least 20 characters are required for AI assistance.

\n

\n

Step 2: The user types their response.

\n

If the user would like AI assistance, they can expand the AI Assist panel and access the feature.

\n

\n

Step 3: The user will receive a suggested response from the AI Assistant.

\n

Note: The AI Assistant does not fully support bulleted formats or multiple separated paragraphs. As a result, the tool will offer suggestions in a single-paragraph format.

\n

The tool tip explains that the suggestion focuses on clarity, impartiality, and professionalism.

\n

\n

 

\n

\n

Users are given further details about the suggestion and an option to indicate if it is or isn't helpful (thumbs up or thumbs down). We encourage users to do so as it will help improve future suggestions.

\n

Step 4: The user can apply the text or regenerate the output and choose to apply the new text.

\n

Users can accept the AI suggestion, ask the AI to generate another suggestion immediately, or modify their original response before asking the AI to generate another suggestion. 

\n

\n

If the user asks AI to generate another suggestion, the user will be asked if the new suggestion is better, worse, or the same.

\n

\n

Step 5: Review and Share.

\n

Downloadable Resources

\n\n

FAQs

\n

Are our employees' data used to train AI models? 

\n

No, your employees' data is never used to train or fine-tune our generative AI model. We understand the importance of data privacy and take steps to ensure your employees; data is used solely for your specific use case. Information given using the thumbs up and thumbs down icons does not train the model, it just informs our team if the output is helpful or not. It will help us improve the prompts that generate outputs.

"}}