Exercise 01: Create agents by using Copilot Studio
Introduction
Zava Retail Group’s AI COE is under pressure to deliver visible, high-impact agent experiences quickly, while proving those agents are grounded in enterprise data and behave consistently across teams. For early wins like executive reporting and operational support, the foundation is always the same: an agent that is structured, maintainable, and aligned to real business workflows, not ad-hoc prompting.
In this exercise, you’ll build that foundation in Copilot Studio. You’ll create an agent, connect it to approved SharePoint knowledge sources, and design modular topics that break complex workflows into smaller, reusable components. By the end, you’ll have a well-structured agent pattern that Zava can reuse as it scales to more business-critical scenarios.
This exercise includes the following tasks:
- Create and enrich a Copilot Studio agent by defining business goals and structuring them into clear, reusable modules.
- Build a refund‑initiator modular topic that greets the customer, extracts refund details, and validates the transaction.
- Build a refund‑submission modular topic that confirms cancellation, collects issue details, saves the record, and returns a confirmation ID.