Microsoft AB-100 (Agentic AI Business Solutions Architect) Practice Test 9

Question 80

A business stakeholder describes this operating situation: An internal helper produces attractive but unsupported policy answers. Testing instead shows that satisfaction scores hide factual mistakes. What design approach is most appropriate?

  1. Integrate runtime tracing and alerting into release readiness
  2. Inspect retrieval spans when cited answers omit evidence
  3. Measure tool success, retries and side-effect duplication
  4. Measure grounded factual accuracy on a curated corpus

Explanation

Correct Answer

D:

The defect is that satisfaction scores hide factual mistakes; the expected business case is that an internal helper produces attractive but unsupported policy answers; the design fits because expert-checked evaluation examples quantify evidence fidelity.

Incorrect Answers

A:

The alternative addresses the on-call team cannot explain failed tool calls by a design in which connected observability supplies traces and operational metrics; it does not supply the missing relationship for an internal helper produces attractive but unsupported policy answers, where satisfaction scores hide factual mistakes.

B:

Surface quality scores appear healthy despite wrong documents; the alternate design can address that situation because retrieval telemetry distinguishes evidence search from answer generation, whereas this question needs a solution to satisfaction scores hide factual mistakes.

C:

The technical behavior of this alternative is that tool metrics reveal mutation failures and retry consequences; the user case being assessed is an internal helper produces attractive but unsupported policy answers with a different failure: satisfaction scores hide factual mistakes.

Question 81

Support engineers report this failure: Multiple agents offer overlapping tools with inconsistent permissions. The business context is: Teams rebuild similar assistants without seeing existing assets. What evidence should be collected to isolate the cause?

  1. Inspect write conflicts and source-of-truth definitions
  2. Compare metered usage with adoption and concurrency assumptions
  3. Compare new proposals against registered agents and prompt assets
  4. Inspect baseline equivalence and external trend factors
  5. Trace customer identifiers and change propagation between apps

Explanation

Correct Answer

C:

Multiple agents offer overlapping tools with inconsistent permissions; to attribute the defect rather than guess at a fix, comparing new proposals against registered agents and prompt assets examines the dependency where discovery and ownership metadata support reuse and controlled deprecation.

Incorrect Answers

A:

The proposed inspection is associated with two applications try to overwrite the same customer credit field, where field ownership and integration contracts prevent competing writes; it would not expose the failure mode that multiple agents offer overlapping tools with inconsistent permissions with the required specificity.

B:

The proposed probe concerns an agent scales from a trial to enterprise usage, where workload forecasts include model usage, connectors, storage and observability; this incident instead requires evidence about discovery and ownership metadata support reuse and controlled deprecation and the failure that multiple agents offer overlapping tools with inconsistent permissions.

D:

The evidence this method produces applies to a workflow where management attributes a sales lift to Copilot without a comparison; it cannot by itself establish why multiple agents offer overlapping tools with inconsistent permissions in the current scenario.

E:

For the failure that an opportunity owner sees a stale service escalation status, tracing customer identifiers and change propagation between apps is relevant; this question instead reports that multiple agents offer overlapping tools with inconsistent permissions, so different evidence is needed.

Question 82

A product owner must measure whether an AI-enabled change actually improves this process: The classifier confuses complaints with refund requests. The earlier problem was: A generic instruction cannot express subtle category boundaries. Which outcome indicator is most relevant?

  1. Correct classification rate for ambiguous boundary cases
  2. New-policy answer accuracy after ingestion
  3. Schema-valid response rate
  4. Unsupported regulated answers reaching users

Explanation

Correct Answer

A:

Because representative examples teach decision boundaries without changing underlying facts, the outcome worth following is correct classification rate for ambiguous boundary cases; the metric is tied to the reported issue—a generic instruction cannot express subtle category boundaries—for the scenario where the classifier confuses complaints with refund requests.

Incorrect Answers

B:

The proposed metric concerns answers depend on policies revised every week, whereas the failure in this case is that a generic instruction cannot express subtle category boundaries; the indicator needed here is correct classification rate for ambiguous boundary cases.

C:

The alternative tracks schema-valid response rate, arising from the mechanism where clear instructions and structured outputs reduce avoidable ambiguity; its movement alone says little about the failure that a generic instruction cannot express subtle category boundaries.

D:

The proposed metric concerns the system cannot always find evidence for regulated guidance, whereas the failure in this case is that a generic instruction cannot express subtle category boundaries; the indicator needed here is correct classification rate for ambiguous boundary cases.

Question 83

Before production promotion, the team needs evidence for this setting: A Copilot Studio agent needs a reusable interface to governed operations. A prior regression caused the following: Every agent reimplements the same proprietary tool signatures. Which validation is most persuasive?

  1. Call a read-only tool with valid and invalid identity contexts
  2. Ask for a case field and verify the exact authorized record
  3. Submit paraphrases and multi-intent requests and verify correct tool selection
  4. Prototype a unique scheduling dependency with the custom runtime

Explanation

Correct Answer

A:

Calling a read-only tool with valid and invalid identity contexts challenges the path where MCP presents supported tools and resources through a common integration contract for the business setting a Copilot Studio agent needs a reusable interface to governed operations; the specific defect to exclude is that every agent reimplements the same proprietary tool signatures.

Incorrect Answers

B:

In the alternate setting where users ask about governed CRM records rather than prose manuals, the test exercises Dataverse grounding exposes appropriate structured records to the agent; even a pass would not show that every agent reimplements the same proprietary tool signatures has been prevented.

C:

In a solution where employees express several intents and need interactive clarification, submitting paraphrases and multi-intent requests and verifying correct tool selection could catch the defect that a rigid menu fails whenever a request contains two intents; the current release gate must challenge the failure where every agent reimplements the same proprietary tool signatures.

D:

The tested behavior is that owning the engine provides control over models, routing and execution; the necessary evidence here concerns MCP presents supported tools and resources through a common integration contract and the previously observed every agent reimplements the same proprietary tool signatures.

Question 84

A proposed deployment supports this business case: Workers need internal wiki articles from outside Microsoft 365. Reviewers also warn that an external connector might index documents without honoring original ACLs. Which restriction mitigates that concrete risk?

  1. Limit creation privileges and prevent duplicate escalations
  2. Prevent automatic approvals when the credit source is unavailable
  3. Honor source access when syncing or federating content
  4. Require solution lifecycle controls before production propagation

Explanation

Correct Answer

C:

In the context where workers need internal wiki articles from outside Microsoft 365, the exposure is that an external connector might index documents without honoring original ACLs; containing it requires a guardrail in the operation where supported connectors expose external content with discoverability and security metadata.

Incorrect Answers

A:

The proposed control targets repeated agent calls might create duplicate escalation records or excess privileges, while the given risk scenario is an external connector might index documents without honoring original ACLs; these exposures arise in different operational contexts.

B:

This restriction is associated with external credit service responds slowly during regional outages and addresses the exposure that credit-dependent approvals might proceed while the live credit source is unavailable; it does not specifically protect the target workflow against an external connector might index documents without honoring original ACLs.

D:

Where multiple low-code apps need a common extraction prompt, the guard can address a shared AI prompt update might spread to live Power Platform apps without change control; the exposure that an external connector might index documents without honoring original ACLs remains relevant to the present assessment.

Question 85

Troubleshooting starts from a business complaint: Timeout creates an abandoned workflow with no owner. The affected process is: An autonomous agent relies on slow approvals and delayed tool results. Which investigation is most directly tied to the reported defect?

  1. Diff tool traces and evaluation outcomes against baseline
  2. Inspect pending runs, deadlines and compensation events
  3. Reconcile session outcomes to actual completed records
  4. Compare intended task with executed tool names and inputs

Explanation

Correct Answer

B:

Timeout creates an abandoned workflow with no owner; to attribute the defect rather than guess at a fix, inspecting pending runs, deadlines and compensation events examines the dependency where state-machine tests examine retries, expiry and compensation.

Incorrect Answers

A:

In the separate scenario where a hosted agent gains an additional research tool, the same investigation could explain new planning behavior occasionally leaks internal notes; it does not isolate the reported cause in a system where an autonomous agent relies on slow approvals and delayed tool results.

C:

A dashboard calls the bot successful because messages increased; examining that other path does not determine whether state-machine tests examine retries, expiry and compensation, which is what distinguishes the issue in the stated use case.

D:

In the separate scenario where the planner calls an order-cancel tool for status requests, the same investigation could explain final answer text conceals the erroneous tool choice; it does not isolate the reported cause in a system where an autonomous agent relies on slow approvals and delayed tool results.

Question 86

An enterprise team is making a platform design decision. Business context: Planner calls the wrong tool among overlapping capabilities. Present failure: The team proposes adding another similar tool. Which approach resolves the mismatch?

  1. Apply document-level access trimming during retrieval
  2. Cluster similar agent failures using embeddings or AI-assisted analysis
  3. Tune descriptions and tool constraints before adding more tools
  4. Classify feedback by factual, workflow, usability and policy defect

Explanation

Correct Answer

C:

For the need that planner calls the wrong tool among overlapping capabilities, the relevant mechanism is that clear capabilities and restricted sets improve selection reliability; that architectural dependency addresses the observed failure where the team proposes adding another similar tool.

Incorrect Answers

A:

For another type of architecture, retrieval applies effective source entitlements to candidate documents is useful; this design must instead support planner calls the wrong tool among overlapping capabilities despite the problem where the team proposes adding another similar tool.

B:

This design would make sense if thousands of transcripts contain recurring misunderstandings, particularly where analysts overlook a widespread intent failure among unique phrasings; the requested architecture instead needs the dependency that clear capabilities and restricted sets improve selection reliability because the team proposes adding another similar tool.

D:

For complaints use vague labels such as agent is wrong, this other approach relies on the fact that structured defect taxonomy assigns the right remediation owner; the issue here is the team proposes adding another similar tool; that mechanism does not establish the required clear capabilities and restricted sets improve selection reliability.

Question 87

One enterprise rollout has 2 separately testable requirements. (1) Business use: Retrieved policy fragments lose context when documents are split. Known defect: Answers cite isolated sentences without section qualifiers. (2) Business use: Product catalog queries mix synonyms and exact serial codes. Known defect: Pure semantic search misses precise part identifiers. Which 2 architectural choices directly satisfy these stated cases? Choose TWO.

  1. Choose document intelligence extraction for complex business forms
  2. Use an agent flow for repeatable ordered business steps
  3. Chunk long documents with stable metadata and source links
  4. Use hybrid semantic and lexical retrieval for specialist terms

Explanation

Correct Answers

C:

For the independent requirement that retrieved policy fragments lose context when documents are split, semantically coherent chunks with provenance preserve context during retrieval; it addresses the defect answers cite isolated sentences without section qualifiers, while product catalog queries mix synonyms and exact serial codes needs its own architecture.

D:

For the independent requirement that product catalog queries mix synonyms and exact serial codes, hybrid ranking combines lexical precision and semantic relevance; it addresses the defect pure semantic search misses precise part identifiers, while retrieved policy fragments lose context when documents are split needs its own architecture.

Incorrect Answers

A:

This design is associated with the different failure where template parsing omits line items on unfamiliar forms in a workflow where vendors send variable-layout invoices and shipping documents; it does not directly solve the listed conditions: answers cite isolated sentences without section qualifiers; pure semantic search misses precise part identifiers.

B:

This design is associated with the different failure where an unconstrained planner sends a notification before authorization in a workflow where approval processing requires sequential verification and notification; it does not directly solve the listed conditions: answers cite isolated sentences without section qualifiers; pure semantic search misses precise part identifiers.

Question 88

The fault affects an AI-powered workflow in which management attributes a sales lift to Copilot without a comparison. Specifically, seasonality produces the same change in unassisted teams. Which diagnostic investigation best distinguishes the relevant failure path?

  1. Inspect baseline equivalence and external trend factors
  2. Identify workflow steps with fixed rules versus variable language
  3. Compare baseline model errors by language and domain
  4. Examine tool calls by transaction impact and approval requirement

Explanation

Correct Answer

A:

Seasonality produces the same change in unassisted teams; to attribute the defect rather than guess at a fix, inspecting baseline equivalence and external trend factors examines the dependency where matched comparisons separate solution impact from unrelated shifts.

Incorrect Answers

B:

For the failure that a single agent unpredictably handles policy rules and narrative summarization, identifying workflow steps with fixed rules versus variable language is relevant; this question instead reports that seasonality produces the same change in unassisted teams, so different evidence is needed.

C:

A unique domain vocabulary produces persistent measured errors; its matching diagnostic check is reasonable for teams propose custom training without any baseline; the present investigation needs evidence for the different failure where seasonality produces the same change in unassisted teams.

D:

In the process where teams propose that an agent perform both low-risk lookups and financial postings, the diagnostic procedure could explain the same agent can read an invoice and commit a payment unchecked; this particular incident instead turns on whether matched comparisons separate solution impact from unrelated shifts and why seasonality produces the same change in unassisted teams.

Question 89

A risk analysis highlights a threat in this workflow: Operational records update faster than the agent's indexed copy. Exposure: Stale indexed values could lead an agent to issue urgent but incorrect commitments. Which enforceable safeguard addresses it at the relevant boundary?

  1. Roll back route changes if quality guardrails regress
  2. Route urgent live values to a transactional API when the lag threshold is exceeded
  3. Block unsupported sensitive answers and offer escalation
  4. Exclude use cases lacking a safe data-access path from early rollout

Explanation

Correct Answer

B:

When stale indexed values could lead an agent to issue urgent but incorrect commitments, the use case operational records update faster than the agent's indexed copy requires a guardrail aligned with the behavior that incremental refresh and source-to-index lag monitoring reduce staleness.

Incorrect Answers

A:

Where real task mix differs from prelaunch performance tests, the guard can address router optimization could cut cost while materially worsening critical outcomes; the exposure that stale indexed values could lead an agent to issue urgent but incorrect commitments remains relevant to the present assessment.

C:

Where answers about product policy require traceable evidence, the guard can address the agent might deliver sensitive guidance when no trusted citation exists; the exposure that stale indexed values could lead an agent to issue urgent but incorrect commitments remains relevant to the present assessment.

D:

The alternate guard addresses early adopters might receive access to an ungoverned or unreliable data source for candidate processes are attractive but permissions and data quality are weak; the present threat, stale indexed values could lead an agent to issue urgent but incorrect commitments, needs a different enforcement boundary.

Question 90

A test plan focuses on a previously missed failure: A general large model exceeds per-document unit economics. It occurred when high-volume narrow extraction needs predictable latency and cost. Which check directly validates that behavior?

  1. Test the custom action with existing identity and audit policy
  2. Recalculate the case under higher review time and lower adoption
  3. Demonstrate a custom multi-step workflow and failure recovery
  4. Simulate endpoint failure and verify continuity without policy bypass
  5. Run a representative extraction holdout at production concurrency

Explanation

Correct Answer

E:

A general large model exceeds per-document unit economics; a release test must establish that this no longer occurs; running a representative extraction holdout at production concurrency measures the specific behavior for which smaller models can be evaluated for narrow tasks with lower cost.

Incorrect Answers

A:

In the alternate setting where current users need one governed custom action inside their workflow, the test exercises extension retains familiar user experience and adds bounded custom behavior; even a pass would not show that a general large model exceeds per-document unit economics has been prevented.

B:

This check exercises a different system behavior: net benefit compares realized outcomes with implementation and recurring costs; a pass there cannot establish that a general large model exceeds per-document unit economics no longer occurs in the required workflow.

C:

A pass on that verification would speak to the solution demands its own model selection and hosting policies; it would not show that smaller models can be evaluated for narrow tasks with lower cost, which is needed to avoid the observed defect a general large model exceeds per-document unit economics.

D:

The verification concerning peak load can saturate a primary model endpoint checks health-aware fallback uses a compatible approved model when needed against requests fail despite another approved deployment being available; its result says little about the observed issue where a general large model exceeds per-document unit economics.

Agentic AI Business Solutions Architect (AB-100) Practice Tests