Govern agents scenarios
Consider a fictitious company called Zava, an innovative textile manufacturer. Zava uses AI, agents, and human expertise to become frontier. As Zava adopts AI, teams develop agents across Foundry, Copilot Studio, and partner ecosystems. As the number of agents increase, Zava empowers their IT teams to:
Manage agent lifecycle, including activating agents, approving or rejecting agent requests, assigning owners, and resolving exceptions.
Set up governance policies and controls to ensure compliance, security, and consistency from the start.
Using auditing to ensure traceability and accountability.
Personas
The exercises in this module feature the following personas:
Oscar Ward, IT Administrator at Zava with Global Administrator role
Reed Smith, IT Administrator at Zava with AI administrator role
Hannah Travis, Procurement manager at Zava with an agent maker or developer role
Note
The Agent maker/developer role varies based on where the agent is created. For example, if the agent is created in Foundry, the role is a Foundry maker. If the agent is created in Copilot Studio, the role is a Copilot Studio maker.
Scenario
These personas work together to manage and use agents within the organization.
Oscar Ward assigns Reed Smith the necessary permissions required to manage agents effectively by granting the AI Administrator role.
Reed Smith reviews all the agents in the organization to monitor their activity and take action to ensure compliance, governance, and security.
In this module, Reed performs the following tasks:
Activates agents with proper governance controls in place
Assigns owners to agents without owners
Block agents with security risks or compliance issues
Approve or reject agent requests, assign owners, and resolve alerts and exceptions
Hannah Travis requests new agents and uses them for daily work to improve efficiency and increase productivity.
