Exercise 03: Copilot Studio with MCP
Case study overview
The HR team at Zava wants a way to easily search and update candidate information and thinks a conversational agent will be the best way to do this. You’ll extend an agent made with Microsoft Copilot Studio using an MCP (Model Context Protocol) server. Specifically, you’re going to consume an existing MCP server that provides tools for managing a hypothetical list of candidates for a job role. The MCP server will offer functionalities to:
- List all candidates.
- Search for candidates by criteria.
- Add new candidates.
- Update existing candidate information.
- Remove candidates.
Adele Vance, a store manager embedded in Zava’s AI COE, will help configure and pilot the new candidate-management agent with the HR team.
Objectives
After you complete this exercise, you’ll be able to:
- Identify when you must use an MCP server over a prebuilt Copilot Studio connector.
- Extend an agent made with Microsoft Copilot Studio using a Model Context Protocol (MCP) server.
- Configure features in Copilot Studio to develop a working agent using MCP or a more complex connector that is not prebuilt into CPS.
- Publish an agent built in Copilot Studio to the customer’s M365 catalog.
Duration
Estimated time: 15-20 minutes