Chatref reads a website from a URL and answers questions with sources linked back to the page. The system trains itself on website content without requiring manual copy-pasting of pages. Users can test the chat interface privately before deployment. The tone and voice of responses can be configured to match brand guidelines. The interface includes suggested questions that customers can select instead of typing. Pricing inquiries and product questions receive automated responses with detailed information. Knowledge can be expanded by adding website URLs, files, or text, and the system retrains automatically when new content is added. All customer conversations are collected in a unified inbox where support teams can review questions and intervene when needed. Conversations can be tagged by topic, with automatic tagging available to identify common question categories. Each conversation includes an automated summary for quick review. Analytics show which questions Chatref answered successfully, which it could not answer, and overall performance metrics. The system can be deployed on any website.
