Build Generative AI Solutions with Microsoft Foundry AI-102 Practice Questions (Page 6)

Questions 26–30

1. 26application · medium

   A customer service organization is using a prompt flow in Microsoft Foundry to generate responses to customer emails. They have a prompt template that includes placeholders for the customer's name and the issue description. They want to ensure that the template is used consistently across different flow runs. What should they do?

   • AEmbed the template in the application code that calls the flow.
   • BCopy the template into each flow run manually.
   • CUse a different template for each customer email.
   • DStore the prompt template in the flow's files and reference it in the LLM node.
   Reveal Solution Discussion Select an answer first
2. 27foundation · easy

   Before implementing a generative AI solution in Microsoft Foundry, what is the FIRST step in the planning phase?

   • ADeploy a model endpoint
   • BSet up a hub and project
   • CDefine the business problem and success criteria
   • DCreate a prompt flow
   Reveal Solution Discussion Select an answer first
4. 28application · medium

   A marketing agency is using Microsoft Foundry to generate social media posts. They have a prompt flow that uses a prompt template to generate posts. They want to make the template reusable across different campaigns by allowing the campaign name and tone to be passed as variables. What should they do?

   • AUse a different model for each campaign to handle the tone.
   • BCreate a new prompt flow for each campaign with a different template.
   • CHard-code the campaign name and tone in the template for each campaign.
   • DDefine the campaign name and tone as inputs to the prompt template and reference them in the template text.
   Reveal Solution Discussion Select an answer first
5. 29application · medium

   A retail company is building a customer-support chatbot using a prompt flow in Microsoft Foundry. They want to reuse the same system prompt across multiple flows, but they need to customize the greeting based on the user's region. What should they do?

   • AModify the model's system message directly in the deployment settings.
   • BUse a prompt template with a variable for the region and reference it in the flow.
   • CStore the greeting in a configuration file and load it at runtime without using a template.
   • DCreate a separate flow for each region with the greeting hard-coded in each flow.
   Reveal Solution Discussion Select an answer first
6. 30application · medium

   A team has built a prompt flow that uses multiple LLM calls and a Python tool to process intermediate results. They want to test the flow with different input variations before deploying. What is the best way to do this?

   • ARun the flow manually in the Foundry portal with a few sample inputs.
   • BDeploy the flow and test it in production.
   • CCreate a batch run with a dataset of test inputs and review the outputs.
   • DUse the SDK to call the flow with a single input.
   Reveal Solution Discussion Select an answer first