In this second episode of our four-part series, we start building the proof-of-concept solution introduced in Episode 1 by constructing the shared framework that all components will depend on. This episode focuses on two custom Now Assist Skills that serve as common tools across the solution, and the agentic workflow that acts as the entry point for the entire process. These foundational components are what allow end users to interact with the solution through natural language, without needing to know the exact CMDB class or attribute names they are working with. In this episode we cover: - A review of the use case and solution architecture - Building a custom Now Assist Skill to identify the correct CMDB Class based on user input - Building a custom Now Assist Skill to identify the correct CMDB Attribute based on user input - Building the Agentic Workflow that guides users to the appropriate AI Agent - A live demo of configuring these components 0:00 - Introduction 1:10 - Review 4:58 - Common Tools 6:33 - Agentic Workflow 7:42 - Demo creation of custom Now Assist skill for finding CMDB Class 17:02 - Demo creation of custom Now Assist skill for finding CMDB Attribute 21:51 - Demo creation of Agentic Workflow 25:18 - Recap and Preview Supporting blog post with code snippets and additional resources: https://www.servicenow.com/community/cmdb-blog/closing-the-cmdb-data-gap-with-purpose-built-ai-agents/ba-p/3450078