Boost Virtual Agent with AI Search in ServiceNow

The ServiceNow Virtual Agent is a powerful tool for automating interactions and providing self-service support. However, users may encounter frustrating experiences if the Virtual Agent fails to understand their requests. This article explores how to leverage AI Search and other search tools within the Virtual Agent to deliver more accurate and relevant results, ultimately improving user satisfaction and reducing support costs. By implementing these strategies, you can ensure that your Virtual Agent effectively addresses user queries and provides seamless access to information.

AI Search is a modern search engine within ServiceNow that utilizes machine learning to deliver highly relevant search results. Unlike traditional keyword-based search, AI Search understands the context and intent behind user queries, providing more accurate and personalized results. Genius Results, powered by Natural Language Understanding (NLU), take this a step further by presenting the single most relevant item in a rich card format, allowing users to quickly access the information they need. This feature is particularly useful for common requests and frequently asked questions.

AI Search is typically provisioned automatically in ServiceNow instances. If it's not enabled, you can request it through the Conversational Interfaces Home page. Once provisioned, the AI Search Status page should display "AI Search is Ready!". To use AI Search as the default fallback in your Virtual Agent, ensure that the "AI Search – Fallback" topic is added to your Default Chat Experience setup. You can also customize the Search Application and Search UI (EVAM) Configuration in the Search Mapping settings to tailor the search experience. For Genius Results, synchronize your Vocabulary Sources related to AI Search and activate Genius Results for "Catalog" and "Q&A" (Knowledge) in your AI Search profile.

Contextual Search serves as a valuable fallback option when AI Search is unavailable or fails to return relevant results. The "Search Fallback topic" utilizes the contextual search topic block, which can also be incorporated into other topics. This approach ensures that users can still access information from the Knowledge Base and Service Catalog even if their initial query is not directly addressed by a specific topic. You can configure the search context of the Search Fallback Topic by duplicating it and modifying the context field within the Contextual Search topic block. The default search context is "Incident Deflection," but you can create custom contexts with different Search Sources to meet your specific needs.

The FAQ Builder topic block provides a simple and efficient way to address frequently asked questions within the Virtual Agent. When a user's question matches an FAQ record in the Knowledge Base, the Virtual Agent will display the corresponding answer. To enable FAQ Builder, you need the Knowledge Management – Advanced Installer plugin and must activate the "FAQ" article template in Knowledge > Administration > Templates. Ensure that you have published FAQ Knowledge records and then incorporate the FAQ Conversation Builder topic block into your Virtual Agent topics. This topic block requires an input query, which can be a text input, allowing users to ask their questions in natural language.

In some cases, the information users need may reside outside of ServiceNow. The Virtual Agent can be configured to integrate with external sources, such as search engines or product support sites, to provide more comprehensive results. AI Search can also integrate with customer-owned external sites like corporate SharePoint. To configure external sources, you can create dynamic links where a portion of the URL is scripted or based on the user's input. This allows you to generate a URL link that users can click to access external search results based on their search terms. For example, you can create a link to the Microsoft Support site that automatically searches for the user's query.

If you encounter issues with Virtual Agent and AI Search, start by testing your search terms in the AI Search Preview tool to verify the expected results. Ensure that the "AI Search - Fallback" topic uses the updated "AI Search" topic block. Admins can also adjust the number of AI search results displayed by modifying the `com.glide.cs.ai_search.max_regular_result` system property. If you are expecting certain search results and they are not appearing, double-check the configuration of your Search Sources and Search Profiles. Also, verify that your Knowledge Base and Service Catalog are up-to-date and contain the relevant information.

To ensure the effectiveness of your Virtual Agent search capabilities, it's crucial to maintain a high-quality Knowledge Base and Service Catalog. Regularly update your articles and catalog items with the latest information. Avoid creating topics solely for displaying knowledge articles; instead, rely on search fallback to surface relevant information. Keep your ITSM/HRSD Virtual Agent Conversations store app updated to benefit from the latest improvements and features. By following these best practices, you can ensure that your Virtual Agent consistently delivers accurate and helpful search results.

By strategically implementing AI Search, Contextual Search, FAQ Builder, and external source integrations, you can significantly enhance the search capabilities of your ServiceNow Virtual Agent. Prioritize keeping your Knowledge Base and Service Catalog current, and leverage the latest features available through store app updates. These efforts will lead to improved user satisfaction, reduced support costs, and a more efficient self-service experience. Regularly monitor search performance and user feedback to identify areas for further optimization and ensure that your Virtual Agent continues to meet the evolving needs of your users.