Design AI and Agents for Business Solutions AB-100 Practice Questions (Page 11)

Questions 51–55

1. 51foundation · easy

   What is the purpose of designing data processing workflows for AI models and grounding?

   • ATo reduce the size of the AI model
   • BTo speed up the training of AI models
   • CTo eliminate the need for human oversight
   • DTo ensure that AI models have access to relevant and high-quality data for accurate responses
   Reveal Solution Discussion Select an answer first
2. 52foundation · easy

   What is the purpose of designing prompt actions within Copilot Studio?

   • ATo secure the agent from unauthorized access
   • BTo automatically generate responses without user input
   • CTo enhance user engagement by providing interactive prompts that guide the conversation
   • DTo replace the need for topics
   Reveal Solution Discussion Select an answer first
4. 53application · medium

   A company wants to automate the process of invoice approval. The agent should collect invoice details, validate them against a business rule, and then route the invoice to the appropriate approver. The process is well-defined and does not require creative decision-making. What type of agent should the architect design?

   • AA task agent that uses a guided conversation flow with validation steps and conditional routing.
   • BA fallback topic that forwards all invoices to a human for manual approval.
   • CA prompt and response agent that uses generative AI to interpret the invoice and decide the approver.
   • DAn autonomous agent that independently decides the approval path based on past behavior.
   Reveal Solution Discussion Select an answer first
5. 54expert · hard

   A multinational bank is designing an intelligent application that uses AI to process loan applications. The solution must meet strict data residency requirements, handle sensitive customer data, and be resilient to regional outages. The architect is applying the Power Platform Well-Architected Framework. Which design decision best aligns with the framework's pillars?

   • ADeploy the solution in a single region to simplify data residency and reduce cost.
   • BStore all data in a central location and use AI to anonymize it before processing.
   • CUse a single region but implement a manual failover process to another region.
   • DUse multiple regions with data residency enforced by geo-located data sources and design for failover.
   Reveal Solution Discussion Select an answer first
6. 55expert · hard

   A contact center wants to deploy an agent that can handle both chat and voice interactions. The agent must be able to transfer a conversation from chat to voice without losing context, and also escalate to a human with full history. The team has limited development resources. What should the architect design?

   • ADeploy a single Copilot Studio agent on both channels and use session variables to maintain context.
   • BUse a third-party integration to connect chat and voice, but do not use Copilot Studio.
   • CDeploy separate agents for chat and voice, and manually copy context between them.
   • DUse a single agent but only support chat, with voice handled by a human.
   Reveal Solution Discussion Select an answer first