AB-100: Agentic AI Business Solutions Architect › Practice Assessment for Exam AB-100: Agentic AI Business Solutions Architect › Question #26602
Single Choice Moderate

Q

A company has a Microsoft Copilot Studio agent that automatically approves or rejects purchase requests.

The agent produces inconsistent decisions for identical requests, and some requests that are missing required fields are being approved. Historical request data is available, and ongoing validation after deployment is required.

You need to recommend a testing approach that provides measurable validation criteria for evaluating approval decision quality and consistency.

What should you recommend?

ID: #26602 Practice Assessment for Exam AB-100: Agentic AI Business Solutions Architect 8 views
Question Info
#26602Q ID
ModerateDifficulty
Practice Assessment for Exam AB-100: Agentic AI Business Solutions ArchitectTopic
Start Quiz on This Topic
 Your Answer

Choose the Best Option

Click any option to instantly check if you're correct.

  • A Replace the grounding sources and use response citations from representative requests to validate approval correctness. ✔ ✖
  • B Fine-tune a custom model and validate decisions by tracking training accuracy metrics derived from representative request data. ✔ ✖
  • C Increase the response variability during testing, manually review representative requests, and rely on post-deployment user feedback. ✔ ✖
  • D Adjust the agent settings for consistency, validate decisions by using representative request data, and monitor approval outcomes over time. ✔ ✖
Correct Answer

Explanation

Objective:

3.2 Manage the testing of AI-powered business solutions

What This Item Tests:

Create validation criteria of custom AI models

Additional Reading:

What operational factors and settings allow for effective and responsible use of the agent approvals experience?

Key considerations when using RAG

Rationale:

Improving consistency through configuration, validating approval behavior against representative request data, and monitoring approval outcomes over time provides measurable criteria for evaluating the approval quality before and after deployment. Manual review and post-deployment feedback do not scale or provide proactive validation, grounding citations do not measure approval logic correctness, and training accuracy metrics do not reflect real-world approval decisions.

 Continue Practice
Previous Question Next Question
 Share

Share This Question

Challenge a friend or share with your study group.

 More from This Topic

Related MCQ Questions

A company has two production agents: One is built in Microsoft Copilot Studio, and the other is built in Microsoft Foundry. Both agents send telemetry to the same Azure Monitor Application Insights resource. U... A company has a multi-agent solution built in Microsoft Foundry. The solution grounds responses in internal HR documents and uses agent tools to call internal and external systems. Security testing shows that ... A company uses Microsoft Dynamics 365 Sales with a custom model-driven app that sellers use instead of the standard Sales Hub app. The sellers cannot open the Copilot immersive experience from the custom app n... A company uses an agent built in Microsoft Foundry that retrieves internal content and invokes tools to perform actions. Security testing shows that prompt injection attacks can trigger unauthorized tool calls... A company has a Microsoft Copilot Studio agent grounded in a Microsoft SharePoint Online knowledge source. After adding new knowledge documents and updating the agent instructions, responses to the same user q... A company has a customer service agent built in Microsoft Foundry that calls an Azure OpenAI model and a knowledge API. Telemetry and online evaluation scores are sent to Azure Monitor Application Insights. Us... A company has a Microsoft Copilot Studio agent that performs browser-based automation against a legacy web app by using the Computer Use tool. When multiple users trigger the automation simultaneously, some ru... A company has multiple Microsoft Copilot Studio agents that create and update service desk tickets by calling an internal ITSM API. The same ticketing actions must be shared across all agents, and the API sche... A company has a Microsoft Copilot Studio agent that generates a daily incident summary from a user-provided transcript and incident data stored in a Microsoft Dataverse table. The response must always be forma... A company has a Microsoft Copilot Studio agent that uses Azure OpenAI to generate responses and retrieves grounding data from Microsoft Dataverse. Multiple users can modify the agent's configurations, includin...