Microsoft AB-100 (Agentic AI Business Solutions Architect) Practice Test 66

Question 650

An investment review asks what measurable result would justify continuing an AI solution where proprietary records are copied to an open test workspace. The original issue is: Training examples become accessible outside intended teams. Which indicator should be tracked?

  1. Cross-role cache disclosure incidents
  2. Unauthorized inference requests successfully blocked
  3. Unauthorized grounding documents included in answers
  4. Unauthorized accesses to model-training datasets

Explanation

Correct Answer

D:

This scorecard needs an outcome related to proprietary records are copied to an open test workspace; unauthorized accesses to model-training datasets. That indicator can be checked against the earlier symptom that training examples become accessible outside intended teams.

Incorrect Answers

A:

For the workflow where shared result caching makes answers fast across roles, cross-role cache disclosure incidents is an actionable measure; it does not establish whether training examples become accessible outside intended teams in the situation under review.

B:

Although unauthorized inference requests successfully blocked is observable, the affected process involves proprietary records are copied to an open test workspace; its acceptance outcome is unauthorized accesses to model-training datasets, not the alternative measurement.

C:

For the workflow where users have different rights to the same knowledge corpus, unauthorized grounding documents included in answers is an actionable measure; it does not establish whether training examples become accessible outside intended teams in the situation under review.

Question 651

After deployment, stakeholders need an interpretable measure for this use case: External credit service responds slowly during regional outages. The pilot showed that a sales workflow hangs awaiting optional enrichment. Which metric captures the needed result?

  1. Escalation rate after repeated unresolved turns
  2. Sales task completion under external API impairment
  3. Sensitive-field disclosure rate in connector responses
  4. Correct identification of governed service intents

Explanation

Correct Answer

B:

The observed shortcoming is that a sales workflow hangs awaiting optional enrichment; measuring sales task completion under external API impairment gives business owners a directly relevant indicator for the process where external credit service responds slowly during regional outages.

Incorrect Answers

A:

A change in escalation rate after repeated unresolved turns is informative for contact Center agents sometimes cannot resolve complex billing intent; it does not establish whether the current process achieves sales task completion under external API impairment.

C:

A change in sensitive-field disclosure rate in connector responses is informative for sales copilots should not expose sensitive margin assumptions; it does not establish whether the current process achieves sales task completion under external API impairment.

D:

A change in correct identification of governed service intents is informative for agents must distinguish an incident, complaint, inquiry and entitlement; it does not establish whether the current process achieves sales task completion under external API impairment.

Question 652

A proposed deployment supports this business case: Refund, credit, reimbursement and adjustment have different accounting impacts. Reviewers also warn that an AI response might promise a credit or concession beyond approved authority. Which restriction mitigates that concrete risk?

  1. Require human sign-off when regulated language is modified
  2. Enforce source-system entitlements in the integration
  3. Reject or escalate uncertain cross-system matches
  4. Escalate unapproved concessions and ambiguous financial commitments

Explanation

Correct Answer

D:

The business setting is refund, credit, reimbursement and adjustment have different accounting impacts, with the exposure that an AI response might promise a credit or concession beyond approved authority; the enforceable protection must correspond to the path where controlled definitions tie each term to approved policies and authorities.

Incorrect Answers

A:

The safeguard for compliance letters must preserve approved language and disclaimers deals with a different threat: a model might alter mandatory legal wording in customer correspondence; even after it is enforced, the exposure that an AI response might promise a credit or concession beyond approved authority remains possible in the target workflow where controlled definitions tie each term to approved policies and authorities.

B:

The proposed control targets sales enrichment might query external records beyond the seller's entitled scope, while the given risk scenario is an AI response might promise a credit or concession beyond approved authority; these exposures arise in different operational contexts.

C:

The alternate guard addresses a guessed cross-system account match might attach credit risk to the wrong customer for two systems share company names but not primary keys; the present threat, an AI response might promise a credit or concession beyond approved authority, needs a different enforcement boundary.

Question 653

An agent-assisted service does not behave as specified: It autonomously contacts a supplier without approval. It is used in this situation: A bot only needs to check warranty and prepare a draft. Which observable diagnostic procedure is best?

  1. Trace action authentication and source mutation evidence
  2. Compare tool inventory with legitimate workflow steps
  3. Compare displayed metrics with independently computed results
  4. Examine schema validation logs for missing or invalid properties

Explanation

Correct Answer

B:

For the situation where a bot only needs to check warranty and prepare a draft, the defect is that it autonomously contacts a supplier without approval; comparing tool inventory with legitimate workflow steps helps verify the cause-and-effect path: limited intent and tool scope prevent unnecessary actions.

Incorrect Answers

A:

Tracing action authentication and source mutation evidence explores a separate connection, where an authorized connector action invokes an operational service; the diagnosed requirement is a bot only needs to check warranty and prepare a draft, and the actual defect is it autonomously contacts a supplier without approval.

C:

In the process where a page must show a reproducible service-level metric, the diagnostic procedure could explain lLM-generated arithmetic changes across otherwise identical prompts; this particular incident instead turns on whether limited intent and tool scope prevent unnecessary actions and why it autonomously contacts a supplier without approval.

D:

That probe evaluates structured outputs define required fields and allowable values in a workflow where a summarizer sends output into a ticket ingestion service; the reported failure concerns the path for which limited intent and tool scope prevent unnecessary actions instead.

Question 654

An agent rollout is intended for this situation: Operations need to correlate prompts, model calls and tools in a Foundry workflow. The known problem is that a failing agent shows only a generic HTTP error. Which success measure connects to that specific outcome?

  1. Failed agent requests with diagnosable trace paths
  2. User-impacting incidents detected within response target
  3. Critical journeys passing mandatory acceptance tests
  4. Incorrect tool selections identified before side effects

Explanation

Correct Answer

A:

For the use case where operations need to correlate prompts, model calls and tools in a Foundry workflow, the indicator failed agent requests with diagnosable trace paths is worth baselining against the earlier issue where a failing agent shows only a generic HTTP error.

Incorrect Answers

B:

Although user-impacting incidents detected within response target is observable, the affected process involves operations need to correlate prompts, model calls and tools in a Foundry workflow; its acceptance outcome is failed agent requests with diagnosable trace paths, not the alternative measurement.

C:

A change in critical journeys passing mandatory acceptance tests is informative for agent supports refunds, order status and generic FAQs; it does not establish whether the current process achieves failed agent requests with diagnosable trace paths.

D:

The proposed metric concerns the planner calls an order-cancel tool for status requests, whereas the failure in this case is that a failing agent shows only a generic HTTP error; the indicator needed here is failed agent requests with diagnosable trace paths.

Question 655

The rollout supports this workflow: Agent looks up records on behalf of employees with different rights. Leaders want a metric that responds to the reported shortcoming that shared admin credentials return restricted cases to ordinary users. Which measure best reflects success?

  1. Cross-role unauthorized record access rate
  2. Unauthorized finance postings in release tests
  3. Unauthorized write operations executed by read-only agents
  4. Unauthorized inference requests successfully blocked

Explanation

Correct Answer

A:

For the use case where agent looks up records on behalf of employees with different rights, the indicator cross-role unauthorized record access rate is worth baselining against the earlier issue where shared admin credentials return restricted cases to ordinary users.

Incorrect Answers

B:

The proposed metric concerns agents suggest payment adjustments and inventory holds, whereas the failure in this case is that shared admin credentials return restricted cases to ordinary users; the indicator needed here is cross-role unauthorized record access rate.

C:

For the workflow where an agent should summarize account data but not alter it, unauthorized write operations executed by read-only agents is an actionable measure; it does not establish whether shared admin credentials return restricted cases to ordinary users in the situation under review.

D:

A sensitive model endpoint is reachable using a broadly shared key; its intended measurement is unauthorized inference requests successfully blocked; substituting that result here would not capture whether shared admin credentials return restricted cases to ordinary users has been corrected.

Question 656

The rollout supports this workflow: A customer handoff depends on shared record and term mappings. Leaders want a metric that responds to the reported shortcoming that independent releases cause contradictory customer statuses. Which measure best reflects success?

  1. Unauthorized production customization events
  2. Regional resolution and escalation change after deployment
  3. Complete onboarding cases with accurate Sales context
  4. Customer handoffs free from integration mismatches

Explanation

Correct Answer

D:

The observed shortcoming is that independent releases cause contradictory customer statuses; measuring customer handoffs free from integration mismatches gives business owners a directly relevant indicator for the process where a customer handoff depends on shared record and term mappings.

Incorrect Answers

A:

A change in unauthorized production customization events is informative for production components should not be edited casually; it does not establish whether the current process achieves customer handoffs free from integration mismatches.

B:

For the workflow where a conversational change affects only one support geography, regional resolution and escalation change after deployment is an actionable measure; it does not establish whether independent releases cause contradictory customer statuses in the situation under review.

C:

The proposed metric concerns a won opportunity should create a properly owned onboarding case, whereas the failure in this case is that independent releases cause contradictory customer statuses; the indicator needed here is customer handoffs free from integration mismatches.

Question 657

A product owner must measure whether an AI-enabled change actually improves this process: Two topics claim nearly identical scheduling intents. The earlier problem was: Agent invokes the wrong booking action for a reschedule. Which outcome indicator is most relevant?

  1. Escalation rate after repeated unresolved turns
  2. Drafts meeting structure and source-accuracy criteria
  3. Dialogues with all mandatory compliance steps
  4. Correctly sequenced workflow executions
  5. Correct topic selection for closely related user intents

Explanation

Correct Answer

E:

For the business use where two topics claim nearly identical scheduling intents, the result to quantify is correct topic selection for closely related user intents; it connects measurement to the mechanism where distinct descriptions and inputs improve topic selection, not merely to agent traffic.

Incorrect Answers

A:

The proposed metric concerns contact Center agents sometimes cannot resolve complex billing intent, whereas the failure in this case is that agent invokes the wrong booking action for a reschedule; the indicator needed here is correct topic selection for closely related user intents.

B:

A change in drafts meeting structure and source-accuracy criteria is informative for marketing specialists need tailored descriptions from existing briefings; it does not establish whether the current process achieves correct topic selection for closely related user intents.

C:

Dialogues with all mandatory compliance steps; that indicator describes a distinct workflow where a regulated process requires fixed questions in exact sequence; it does not quantify correct topic selection for closely related user intents, which is needed for the stated case.

D:

The alternative tracks correctly sequenced workflow executions, arising from the mechanism where defined flow steps make sequencing and error paths explicit; its movement alone says little about the failure that agent invokes the wrong booking action for a reschedule.

Question 658

The architect wants evidence before altering the deployment. In a workflow where operations teams review multiple proposed next actions daily, recommendations disappear in conversation history. What should the team examine first?

  1. Compare generated recommendations with recorded dispositions
  2. Compare advertised tools with backend permissions
  3. Review target host support and publishing settings
  4. Review plans for unauthorized leaps or tool chaining errors

Explanation

Correct Answer

A:

For the case where operations teams review multiple proposed next actions daily, the diagnostic path must explain the failure that recommendations disappear in conversation history; comparing generated recommendations with recorded dispositions targets the mechanism by which a persistent agent activity feed supports review, assignment and disposition.

Incorrect Answers

B:

Several agents need the same inventory query operation; its matching diagnostic check is reasonable for each agent builds a different unaudited HTTP wrapper; the present investigation needs evidence for the different failure where recommendations disappear in conversation history.

C:

That probe evaluates host-specific support, packaging and admin approval shape deployment in a workflow where agent experience must appear in selected Microsoft 365 contexts; the reported failure concerns the path for which a persistent agent activity feed supports review, assignment and disposition instead.

D:

The proposed probe concerns procurement exceptions require comparing several policies and records, where reasoning can plan through evidence and tools within enforced boundaries; this incident instead requires evidence about a persistent agent activity feed supports review, assignment and disposition and the failure that recommendations disappear in conversation history.

Question 659

The architecture board has evidence that security cannot identify which agents use protected connectors. The affected situation is: Business units create agents with no central visibility. What change most directly establishes the missing behavior?

  1. Maintain an enterprise inventory of production agents and owners
  2. Preserve model-compatible preprocessing and postprocessing contracts
  3. Validate connector security and consent in target environments
  4. Design a documented agent retirement and offboarding process

Explanation

Correct Answer

A:

Business units create agents with no central visibility; its failing behavior is that security cannot identify which agents use protected connectors; applying the underlying design principle—inventory metadata records owner, risk tier, tools and lifecycle state—connects the requirement to an implementable control.

Incorrect Answers

B:

Bundled preprocessing, model and label maps maintain contract consistency; its main purpose is a retraining release changes the tokenizer and label order, not correction of the reported defect where security cannot identify which agents use protected connectors in the target process.

C:

When an imported agent can authenticate in dev but not production, production consent, identity and connection setup are distinct release checks; in the stated workflow, however, security cannot identify which agents use protected connectors; the relevant architectural behavior is that inventory metadata records owner, risk tier, tools and lifecycle state.

D:

There is a valid place for the mechanism that decommissioning revokes credentials, connectors and publishing rights, namely unused agents retain tokens and active integrations; it does not remove the specific defect that security cannot identify which agents use protected connectors from this case.

Question 660

Operations has not established the cause of this defect: Half-complete transactions leave operational state inconsistent. The associated setting is: An agent creates an order but reservation of inventory fails. Which inspection should precede remediation?

  1. Verify virtual entity enablement and row-level access
  2. Compare cross-service commit histories and compensation events
  3. Compare advertised tools with backend permissions
  4. Trace agent plan for extra tool invocations

Explanation

Correct Answer

B:

Where half-complete transactions leave operational state inconsistent, the useful evidence comes from comparing cross-service commit histories and compensation events; its focus matches the architectural mechanism that compensating operations reconcile earlier successful steps in the process where an agent creates an order but reservation of inventory fails.

Incorrect Answers

A:

Finance users need current operational data in contextual help; its matching diagnostic check is reasonable for a static knowledge upload gives stale purchase-order states; the present investigation needs evidence for the different failure where half-complete transactions leave operational state inconsistent.

C:

Each agent builds a different unaudited HTTP wrapper; examining that other path does not determine whether compensating operations reconcile earlier successful steps, which is what distinguishes the issue in the stated use case.

D:

The proposed probe concerns an employee requests a standard equipment return initiation, where single-purpose tools and typed parameters bound the task; this incident instead requires evidence about compensating operations reconcile earlier successful steps and the failure that half-complete transactions leave operational state inconsistent.

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