Introduction
Caldova is preparing to pilot Copilot Cowork in one department. The department lead wants employees to explore what it can do, while the finance team needs spending to stay predictable. Before the pilot starts, the IT admin needs to establish who can access the service, how much they can spend, and who can approve requests for more capacity.
In this module, you'll help Caldova establish those controls. You'll use spending policies to control access, set limits and alerts, manage credit requests, assign admin responsibilities, and extend governance to agents with Microsoft Agent 365.
A familiar moment
Midway through a multi-step Copilot Cowork task, an employee hits a wall: a notice tells them they've used up this month's credits. The task already underway continues and finishes normally—the service blocks only new tasks until an admin raises the limit or the next month starts.
Nothing's broken. This is usage-based AI working exactly as designed, the same way an outside firm pauses new work once a project budget is exhausted, while still finishing what's already in motion. The real test is whether the organization planned for this moment: a sensible limit, an early warning, a clear way to ask for more, and an admin ready to act on it.
What you'll learn
In this module, you learn how to:
- Decide who can use Copilot Cowork and other usage-based AI services with spending policies.
- Set limits and alerts, and know what users see as they approach or reach a limit.
- Manage credit requests and know who approves them.
- Assign the right admin roles and decision rights for cost management.
- Extend that same governance to agents with Agent 365.
Caldova decision
Caldova's pilot department is ready to enable Copilot Cowork. Before anyone uses it, the IT admin and the department lead decide who can access it, how much they can spend, what happens when someone requests more capacity, and which admin role makes each decision.