Skip to main content
  1. Public practice questions
  2. AB-620 Practice Exams | AI Agent Builder Associate
  3. AB-620 Practice Exam 1

Question 32 — AB-620 Practice Exam 1

For each statement below, indicate whether the answer is True or False

Response matrix
ItemTrue False
The Test your agent panel in Copilot Studio can be used to validate whether requests representing each connected agent's intended responsibility are routed appropriately. □□
A connected-agent routing test is complete when one representative prompt is tested, even when the orchestrator contains several connected agents with different scopes. □□
When inaccurate routing is caused by overlapping connected-agent descriptions, a valid diagnostic cycle is to test the request, inspect the activity map, refine the descriptions, and then retest. □□
A staged rollout requires disconnecting a connected agent until the organization is ready to make that agent available to users. □□

Correct answers

  1. The Test your agent panel in Copilot Studio can be used to validate whether requests representing each connected agent's intended responsibility are routed appropriately. → True
  2. A connected-agent routing test is complete when one representative prompt is tested, even when the orchestrator contains several connected agents with different scopes. → False
  3. When inaccurate routing is caused by overlapping connected-agent descriptions, a valid diagnostic cycle is to test the request, inspect the activity map, refine the descriptions, and then retest. → True
  4. A staged rollout requires disconnecting a connected agent until the organization is ready to make that agent available to users. → False

Explanations

True: The Test your agent panel provides the testing interface for exercising multi-agent routing. Prompts representing the intended responsibilities of the available connected agents help verify that generative orchestration delegates requests appropriately.

False: An orchestrator with multiple connected agents should be tested with requests representing the different agent scopes. Testing only one agent's request does not validate routing across the complete multi-agent solution.

True: Testing identifies the behavior, the activity map exposes the routing decision, description refinement corrects ambiguous routing signals, and retesting verifies whether the change produced the intended result.

False: A connected agent can be disabled for a staged rollout. Disabling prevents routing while retaining the saved configuration, allowing the agent to be enabled later without rebuilding the connection.

The Test your agent panel in Copilot Studio can be used to validate whether requests representing each connected agent's intended responsibility are routed appropriately.

A connected-agent routing test is complete when one representative prompt is tested, even when the orchestrator contains several connected agents with different scopes.

When inaccurate routing is caused by overlapping connected-agent descriptions, a valid diagnostic cycle is to test the request, inspect the activity map, refine the descriptions, and then retest.

A staged rollout requires disconnecting a connected agent until the organization is ready to make that agent available to users.