Using Web URLs to Train Your CRM's Conversation AI
What is Web URL Training?
Web URL training is a feature that allows you to enhance your CRM's Conversation AI by using the content from your existing website. The system will automatically scan, or crawl, the web pages you specify and use their text to build a knowledge base. This enables your AI assistant to answer customer questions with information that is directly sourced from your official website content.
Key Benefits
Training your AI with web URLs offers several significant advantages for improving customer interactions.
- Efficient Setup: Quickly import large amounts of content from your website without needing to manually copy and paste text.
- Improved Accuracy: Ensure the AI's answers are consistent and based on the exact information available on your public-facing pages.
- Content Control: Choose from different crawling modes to target specific pages, sections, or your entire website.
- Easy Maintenance: Update or remove trained links as your website content changes over time.
URL Crawling Modes
You can control how broadly the system searches your website by selecting one of three crawling modes.
Exact URL
This mode is ideal for training the AI on a single, specific webpage. The system will only crawl the exact URL you provide. Use this for critical pages like your pricing, a specific service offering, or your FAQ page.
All URLs in this Domain
This option crawls your entire website domain. It will present a list of all discovered pages, allowing you to review and select which ones to use for training. This is best when most of your website's content is relevant to customer inquiries.
All URLs with this Path
This mode is useful for targeting a specific section of your website. By providing a URL path (e.g., /blog or /support), the system will find all pages that share that path. This allows you to train the AI on a group of related pages, like all your blog articles or support documents, without including irrelevant parts of the site.
How to Set Up Web URL Training
Follow these steps to add website content to your AI's knowledge base.
- Navigate to the Bot Training section within your CRM.
- In the provided field, enter the full web address (URL), including the https:// prefix.
- Select your preferred crawling mode: Exact URL, All URLs in this Domain, or All URLs with this Path.
- Click the Get Data button to initiate the crawl.
- The system will display a list of URLs it found. You can then select the specific pages you wish to train the AI on.
Best Practices
To get the best results from web URL training, keep these tips in mind.
- Focus on Quality: Prioritize pages with clear, accurate, and customer-focused information.
- Avoid Irrelevant Content: Do not train the AI on pages that contain information unrelated to typical customer questions.
- Review Before Training: Always check the list of crawled URLs before finalizing the training to ensure only appropriate pages are included.
- Keep Content Fresh: Periodically refresh trained URLs or remove outdated links to maintain the accuracy of your AI's knowledge.