Configuring Grounded Generative Answers with Azure AI Search

AB-620 Last updated

Configure generative answers by using Azure AI Search with Foundry

Short answer

C Require grounding for generative answers and use Azure AI Search as the retrieval source.

Question

A company is configuring generative answers for an agent in Copilot Studio agent. The agent must use Azure AI Search to retrieve enterprise content, and all responses must be generated by a Microsoft Foundry model. The agent sometimes generates answers that are not grounded in Azure AI Search content. The company requires that the agent only generate responses when relevant indexed content is retrieved. You need to configure the agent so that its generative answer behavior meets the organization's requirements. What should you do?

  1. Use Azure AI Search as a tailback source and enable answer expansion.
  2. Upload documents into the agent and disable fallback responses.
  3. Require grounding for generative answers and use Azure AI Search as the retrieval source. Correct Answer
  4. Enable citations and configure a public model endpoint.
  5. Enable conversational boosting and select a Foundry-approved model.

Key Takeaways

The question tests the specific setting 'Require grounding' which forces the agent to rely exclusively on indexed content, preventing it from using its internal knowledge when no relevant documents are found.

This page explains how to ensure Copilot Studio agents only generate responses based on retrieved enterprise content by requiring grounding. It details the configuration of Azure AI Search as the retrieval source to prevent ungrounded hallucinations.

Common mistake: Candidates often select options involving answer expansion or public models, failing to realize that without explicit grounding requirements, the agent will default to generating answers even if search results are irrelevant or missing.

Expert Analysis

Why the Answer Is Correct

The requirement is strict: the agent must ONLY generate responses when relevant indexed content is retrieved. Option C explicitly states 'Require grounding for generative answers', which is the precise mechanism in Copilot Studio that enforces this constraint. By combining this with 'use Azure AI Search as the retrieval source', the agent is configured to look for evidence before speaking. If no relevant content is found, grounding prevents the model from making up an answer, thus meeting the organization's compliance needs.

Why the Other Options Are Wrong

Option A suggests answer expansion, which typically broadens the search but does not strictly enforce that a response only happens if content is found; it might still allow generation based on expanded but potentially weak signals. Option B involves uploading documents directly, which bypasses the dynamic retrieval nature of Azure AI Search and doesn't address the grounding logic for external sources. Option D enables citations (good for transparency) but does not prevent the agent from answering without grounded data. Option E focuses on conversational boosting and model selection, which affects relevance ranking and capability but not the fundamental constraint of grounding responses to retrieved facts.

Community Comment Notes

Community consensus strongly favors Option C, with multiple users citing the official Microsoft guidance on Retrieval-Augmented Generation (RAG). As noted by user BolDeFruits, the source documentation confirms that requiring grounding is the correct approach for ensuring answers are based on indexed content. Users Joselito79 and faisalronnie also verified that this setting prevents ungrounded responses, aligning perfectly with the scenario's constraints.

Official Reference

https://learn.microsoft.com/en-us/microsoft-copilot-studio/guidance/retrieval-augmented-generation

Exam Strategy

When configuring agents, always distinguish between 'enabling features' (like citations) and 'enforcing constraints' (like grounding). If the question demands that the agent NOT answer unless specific conditions are met, look for 'Require grounding' or similar restrictive settings rather than enhancement options.

Frequently Asked Questions

What happens if grounding is required but no content is retrieved?

The agent will not generate a response based on its internal training data, effectively refusing to answer to prevent hallucination.

Is Azure AI Search mandatory for grounding?

No, you can ground against other data sources like SharePoint or Dataverse, but the question specifically mandates Azure AI Search.

More AB-620 FAQ →

Discussion 4 contributions

Compiled from public community threads and reader submissions, lightly edited for clarity. Handles and vote tallies are shown as recorded.

From public community threads 4

  1. faisalronnie 👍 3 Answered C
    Require grounding for generative answers and use Azure AI Search as the retrieval source.
  2. Joselito79 👍 2 Answered C
    C. Require grounding for generative answers and use Azure AI Search as the retrieval source. ✅ This ensures the agent only generates responses when relevant content is retrieved from Azure AI Search, preventing ungrounded responses and satisfying the enterprise-content requirement.
  3. BolDeFruits 👍 2 Answered C
    Proper answer is C Source: https://learn.microsoft.com/en-us/microsoft-copilot-studio/guidance/retrieval-augmented-generation
  4. Snoopy70 👍 1 Answered C
    Require grounding for generative answers and use Azure AI Search as the retrieval source

Reader comments & corrections

No reader corrections yet — spotted an error or have a note? Share it below.

Log in to comment, report an error, or add a note about this question.

Submitted for moderation before publishing. Keep it helpful and respectful.

Copilot Studio Generative Answer Disclaimer Strategy

S-Grade

Evaluate Copilot Agent Configuration Compliance Status

S-Grade

Export Copilot Studio Agent for Production ALM

S-Grade

← Back to AB-620 Study Guide