Microsoft AB-100 Practice Questions

The latest changes and updates from the administration for this exam.

verified

Latest Update: Jun 14 2026

All questions are working fine.

All Questions (358)bookmarkBookmarked (0)

All358Show all accessible questions in this exam.Missed358Questions you haven't attempted yet.Incorrect0Questions you answered incorrectly on your last attempt.

Question 131

Your organization has deployed Dynamics 365 Contact Center with voice, live chat, and SMS channels connected through Azure Communication Services. Management requires incoming requests from all channels to be assigned to service representatives based on skills, priority, and capacity. Which routing design should the architect recommend?

Use unified routing, which applies AI models and rules to assign work items from all channels to the best-suited representative

Configure separate routing engines for each channel type with independently managed queues and no cross-channel workforce utilization visibility

Set up round-robin queue assignment across all channels without evaluating representative skills, current availability, or active work capacity

Implement manual supervisory distribution where managers monitor each individual channel queue and assign conversations to representatives one by one

Question 132

Your team notices inflated conversation metrics when analyzing Copilot Studio agent telemetry in Application Insights. You discover that test conversations from the authoring canvas are included in the production data. Which KQL approach correctly filters out test conversations?

Filter the requests table where client_Type does not equal 'TestClient' to exclude all test traffic

Extend the query using customDimensions['designMode'] and add a where clause filtering for 'False'

Query only the traces table instead of customEvents because traces automatically excludes test data

Apply a time-based where clause to exclude records logged during off-hours when testing typically occurs

Question 133

Your organization is tuning retrieval accuracy for an AI agent grounded by Azure AI Search. You configure a scoring profile with only weighted text fields and enable semantic ranking with the 'boostedRerankerScore' ranking order. After testing, you find that @search.rerankerBoostedScore shows no improvement over the standard reranker score. What is the MOST likely cause?

The search index must be completely rebuilt from scratch after adding a scoring profile to apply it alongside semantic ranking

Scoring profile boosting at the semantic ranking phase requires function-based boosting such as freshness or magnitude, not just weighted text fields

The semantic ranker feature has not been enabled on the search service through the Premium features settings in the Azure portal

The BM25 similarity algorithm must be replaced with the Classic similarity algorithm before scoring profiles can integrate with semantic ranking

Question 134

Your organization uses Azure Pipelines with Power Platform Build Tools to deploy Copilot Studio agent solutions across development, test, and production. Each environment connects to different SharePoint knowledge sources and Dataverse instances. How should you manage environment-specific configuration during automated deployments?

Manually reconfigure each target environment after the pipeline deployment completes by editing all connections through the Copilot Studio portal

Create separate copies of the agent solution for each environment with unique connections embedded and maintain them in independent Git repositories

Hardcode all connection strings and SharePoint URLs directly inside agent topic configurations before exporting the solution from the development environment

Use deployment settings JSON files stored in source control with environment-specific values passed during the import task

Question 135

An architect is designing test cases in Azure Test Plans for a workflow where Dynamics 365 Sales Copilot generates an opportunity summary that is referenced by Copilot in Dynamics 365 Customer Service during case creation. Multiple test cases across both applications require the same validation steps to verify AI-generated summary accuracy. What should the architect recommend?

Copy the summary validation steps individually into each test case so that every test case contains complete step documentation

Remove the overlapping validation steps from Customer Service test cases since they duplicate the steps already tested in Sales

Create Shared Steps for the common summary validation sequence and reference them across test cases in both suites

Combine all validation steps for both applications into a single comprehensive test case that is executed in one sequential run

Question 136

You are conducting a total cost of ownership analysis for a RAG-based solution using Azure AI Search. The finance team asks which factors drive the monthly Azure AI Search bill. Which statement accurately describes the Azure AI Search billing model?

Variable cost calculated per query executed with higher rates applied for vector search queries than keyword queries

Consumption-based billing where you pay per document indexed and per API call made to the search service endpoint

Fixed cost based on search units (replicas multiplied by partitions) with optional charges for premium features

Storage-only billing model where you pay per GB of indexed data with no separate charge for underlying compute resources

Question 137

Contoso's sales leadership wants to understand how the sales team uses Copilot for Sales across organizational groups and identify which Copilot for Sales actions contribute to key sales tasks. Which approach should the architect recommend?

Export the Microsoft 365 usage report from the admin center and manually build custom Power BI visualizations from the data

Query Dynamics 365 Sales Dataverse tables directly using custom FetchXML queries and build a reporting solution from CRM activity records

Configure Copilot Studio analytics to track each Copilot for Sales interaction and export session transcripts into a separate data warehouse

Use the Copilot for Sales adoption Power BI template in the Viva Insights Analyst Workbench

Question 138

You are writing a KQL query to analyze generative answer performance in Copilot Studio agent telemetry. You need to extract the topic name, result, and conversation ID from the logged events. These properties are stored within a dynamic-type field. Which KQL operator combination correctly extracts these properties?

Use extend with todynamic() on customDimensions and tostring() to extract each property into separate columns

Use the parse operator on the name column to extract topic name, result, and conversation ID as separate text fields

Use the summarize operator with make_list() on the customDimensions column to aggregate all properties into a single array per session

Use the project operator to directly reference customDimensions.TopicName without any type conversion or extend statements

Question 139

Your Copilot Studio agent occasionally blocks generative AI responses due to responsible AI content filtering. You need to investigate these incidents using Application Insights telemetry. Which KQL query approach is CORRECT for identifying these events?

Query the customEvents table and filter where customDimensions contains ContentFiltered

Query the exceptions table and filter where the exceptionType field equals ResponsibleAIViolation

Query the requests table and filter where the resultCode field equals HTTP status code 451

Query the traces table and filter where the severityLevel equals Critical and message contains blocked

Question 140

Contoso is deploying Microsoft 365 Copilot and custom agents built in Copilot Studio. The compliance team requires visibility into sensitive data shared in AI prompts and responses, along with the ability to enforce data loss prevention policies for AI interactions. Which Microsoft Purview capability should the architect recommend as the centralized management solution?

Microsoft Purview eDiscovery Premium configured with content search queries targeting AI interaction records

Microsoft Purview Data Map scanning with L3 classification rules applied to all connected data sources

Microsoft Defender for Cloud Apps integrated with custom API connectors to monitor all AI application traffic

Data Security Posture Management (DSPM) for AI with its centralized dashboard and one-click policies

Update History

Update History