How to train BoldDesk AI to Handle Customer Queries Effectively
This article explains how to train BoldDesk’s AI Copilot using various knowledge sources to improve response accuracy and relevance. It covers integrating structured data like Knowledge Base articles, predefined Q&A pairs, authorized web pages, and uploaded PDF files. By organizing and enriching these sources, businesses can ensure their AI delivers consistent and intelligent support across customer interactions.
Train AI with Knowledge Sources
BoldDesk AI Copilot uses several types of data to generate responses:
1. Knowledge Base (KB)
Articles from your KB are the primary source. Ensure they are well-organized and topic-specific. Please refer to this article to learn more about adding KB as a data source for training your AI.
2. Questions and Answers
Predefined responses to common queries can be added to guide the AI. Please refer to this article for more insights on custom answers for AI.
These answers are not visible to end users but are used by the AI Copilot to generate responses.
3. Authorized Web Pages
You can add publicly accessible URLs. The AI will crawl these pages and their child pages within the same domain. Please check this article for more information on using web pages as a source of information for AI.
4. Uploaded Files
PDF documents and other files can be uploaded to enrich the AI’s understanding. Please check this article to learn more about uploading files as a data source for AI.
These knowledge sources can be accessed in the AI module as shown below.
- The system currently supports uploading files in PDF format only. Please ensure that your documents are clear, well-structured, and relevant to enhance AI processing and response accuracy.
- Only files that are publicly accessible will be processed. Private files will be excluded from the end user’s update.
