Question 1050
A solution architecture is being reviewed for this business situation: An AI-driven replenishment proposal influences procurement. The observed defect is that recommendation changes spread to all planners without evaluation. Which architectural decision addresses the underlying requirement?
- Plan reversible rollout for supply-chain decision assistance
- Apply output filtering and usage restrictions to model interfaces
- Test Finance-to-Supply-Chain exception handling after partial failure
- Validate finance and operations virtual-entity mappings after change
Explanation
Correct Answer
A:The architecture must reconcile two facts: an AI-driven replenishment proposal influences procurement, yet recommendation changes spread to all planners without evaluation; the selected pattern does so through the mechanism that a pilot cohort and rollback path isolate operational impact.
Incorrect Answers
B:A client can request arbitrary portions of training examples; the alternate design can address that situation because safety controls and authorization checks limit misuse and disclosure, whereas this question needs a solution to recommendation changes spread to all planners without evaluation.
C:In the case where a blocked invoice triggers a hold on related procurement actions, the alternate design depends on failure-injection tests verify retries and compensation between systems; it does not correct the target failure, recommendation changes spread to all planners without evaluation.
D:The technical behavior of this alternative is that integration tests protect virtual-entity schema and source semantics; the user case being assessed is an AI-driven replenishment proposal influences procurement with a different failure: recommendation changes spread to all planners without evaluation.
Question 1051
Security and business owners want to keep this capability: Staff need a contextual workspace assembled around one customer case. They must also prevent the following: A generated page might reveal data fields hidden from its user by backend permissions. Which guardrail should the design incorporate?
- Require solution lifecycle controls before production propagation
- Do not expose fields omitted by server authorization
- Honor source access when syncing or federating content
- Require least-privilege service permissions and action confirmation
Explanation
Correct Answer
B:Because a generated page might reveal data fields hidden from its user by backend permissions, a guardrail must constrain the actual system behavior: generated interface components can be grounded in governed app data and defined actions; protecting a response alone cannot enforce that boundary.
Incorrect Answers
A:Another process could need this guardrail because a shared AI prompt update might spread to live Power Platform apps without change control; the design being assessed supports staff need a contextual workspace assembled around one customer case and has the explicit exposure that a generated page might reveal data fields hidden from its user by backend permissions.
C:This restriction is associated with workers need internal wiki articles from outside Microsoft 365 and addresses the exposure that an external connector might index documents without honoring original ACLs; it does not specifically protect the target workflow against a generated page might reveal data fields hidden from its user by backend permissions.
D:The safeguard for an agent needs to update a governed ticket status deals with a different threat: a tool might update tickets using a shared identity with too many permissions; even after it is enforced, the exposure that a generated page might reveal data fields hidden from its user by backend permissions remains possible in the target workflow where generated interface components can be grounded in governed app data and defined actions.
Question 1052
An enterprise team is making a platform design decision. Business context: Refund, credit, reimbursement and adjustment have different accounting impacts. Present failure: Agent-generated language promises the wrong financial remedy. Which approach resolves the mismatch?
- Normalize account identifiers across CRM and external records
- Control creativity for regulated response templates
- Publish an approved business glossary for customer-service concepts
- Treat regulated financial terms as controlled vocabulary
Explanation
Correct Answer
D:The mechanism that controlled definitions tie each term to approved policies and authorities fits the scenario where refund, credit, reimbursement and adjustment have different accounting impacts; it addresses the defect that agent-generated language promises the wrong financial remedy.
Incorrect Answers
A:The technical behavior of this alternative is that a mapping service resolves stable identities before enrichment; the user case being assessed is refund, credit, reimbursement and adjustment have different accounting impacts with a different failure: agent-generated language promises the wrong financial remedy.
B:In the case where compliance letters must preserve approved language and disclaimers, the alternate design depends on templating and approved clauses limit generated variation; it does not correct the target failure, agent-generated language promises the wrong financial remedy.
C:The technical behavior of this alternative is that business-term definitions align user vocabulary with governed service entities; the user case being assessed is refund, credit, reimbursement and adjustment have different accounting impacts with a different failure: agent-generated language promises the wrong financial remedy.
Question 1053
Business leaders reject an activity-only metric for this process: Anonymous callers can ask generic questions but not change orders. They need to detect the issue where bot attempts protected changes based only on spoken order numbers. Which performance indicator fits?
- Channel-appropriate responses passing usability review
- Protected actions blocked for unauthenticated callers
- Sensitive-field disclosure rate in connector responses
- Handoffs with complete approved case context
- Role-specific tasks completed in supported Copilot surfaces
Explanation
Correct Answer
B:The observed shortcoming is that bot attempts protected changes based only on spoken order numbers; measuring protected actions blocked for unauthenticated callers gives business owners a directly relevant indicator for the process where anonymous callers can ask generic questions but not change orders.
Incorrect Answers
A:The proposed metric concerns the same agent is offered in voice and asynchronous chat, whereas the failure in this case is that bot attempts protected changes based only on spoken order numbers; the indicator needed here is protected actions blocked for unauthenticated callers.
C:Sales copilots should not expose sensitive margin assumptions; its intended measurement is sensitive-field disclosure rate in connector responses; substituting that result here would not capture whether bot attempts protected changes based only on spoken order numbers has been corrected.
D:The proposed metric concerns agents need structured case summaries at handoff, whereas the failure in this case is that bot attempts protected changes based only on spoken order numbers; the indicator needed here is protected actions blocked for unauthenticated callers.
E:The proposed metric concerns employees want curated assistance without owning a separate LLM runtime, whereas the failure in this case is that bot attempts protected changes based only on spoken order numbers; the indicator needed here is protected actions blocked for unauthenticated callers.
Question 1054
A test plan focuses on a previously missed failure: A suggested discount bypasses delegated approval authority. It occurred when service, sales and finance jointly decide high-value concessions. Which check directly validates that behavior?
- Propose a concession above limit and confirm approval enforcement
- Request a high-value refund and verify approval blocks execution
- Run contract tests against both current and previous schema versions
- Test mixed-domain requests and validate specialist selection
Explanation
Correct Answer
A:Proposing a concession above limit and confirming approval enforcement challenges the path where cross-app workflow routes the recommendation to the accountable approver for the business setting service, sales and finance jointly decide high-value concessions; the specific defect to exclude is that a suggested discount bypasses delegated approval authority.
Incorrect Answers
B:For the agent may refund large amounts or deactivate accounts, the alternate test is about whether confirmation and human approval separate advice from execution authority; it cannot prove the outcome needed after the reported regression where a suggested discount bypasses delegated approval authority.
C:A pass on that verification would speak to multiple AI teams consume a changing source API; it would not show that cross-app workflow routes the recommendation to the accountable approver, which is needed to avoid the observed defect a suggested discount bypasses delegated approval authority.
D:The tested behavior is that a coordinating agent selects bounded specialists and merges their outputs; the necessary evidence here concerns cross-app workflow routes the recommendation to the accountable approver and the previously observed a suggested discount bypasses delegated approval authority.
Question 1055
Operations has not established the cause of this defect: Mean response time looks acceptable. The associated setting is: A small portion of calls time out during peak demand. Which inspection should precede remediation?
- Inspect overlapping tool definitions and erroneous call paths
- Diff representative before-and-after activity maps
- Analyze segment-specific score and throughput differences
- Inspect p95 and p99 by dependency and concurrency
Explanation
Correct Answer
D:For the situation where a small portion of calls time out during peak demand, the defect is that mean response time looks acceptable; inspecting p95 and p99 by dependency and concurrency helps verify the cause-and-effect path: tail percentiles surface extreme experiences affecting users.
Incorrect Answers
A:That probe evaluates clear capabilities and restricted sets improve selection reliability in a workflow where planner calls the wrong tool among overlapping capabilities; the reported failure concerns the path for which tail percentiles surface extreme experiences affecting users instead.
B:In the process where a new topic description changes which agent is invoked, the diagnostic procedure could explain regression appears only for compound user requests; this particular incident instead turns on whether tail percentiles surface extreme experiences affecting users and why mean response time looks acceptable.
C:The evidence this method produces applies to a workflow where a tuned model claims higher accuracy but adds latency; it cannot by itself establish why mean response time looks acceptable in the current scenario.
Question 1056
A staging test has to detect the specific defect that pilot satisfaction hides severe low-frequency failures. In the workflow where a helpful assistant can still produce unsafe financial or legal guidance, which exercise verifies the corrective behavior?
- Close a test opportunity and inspect the resulting case
- Rank sample defects under published triage rules
- Test coercive, misleading and out-of-scope requests
- Seed paraphrased failures and verify useful grouping
Explanation
Correct Answer
C:For the required workflow where a helpful assistant can still produce unsafe financial or legal guidance, the test needs to reproduce or prevent the failure where pilot satisfaction hides severe low-frequency failures; testing coercive, misleading and out-of-scope requests exercises the path where risk-based red-team evaluations identify foreseeable harm.
Incorrect Answers
A:The suggested test serves a won opportunity should create a properly owned onboarding case, in which AI recommends an action but downstream case data is incomplete; the present scenario instead concerns the reported problem where pilot satisfaction hides severe low-frequency failures.
B:The verification concerning product owners favor easy fixes over high-consequence failures checks risk-weighted backlog scoring includes severity and exposure against a rare data leak risk remains below cosmetic response issues; its result says little about the observed issue where pilot satisfaction hides severe low-frequency failures.
D:For thousands of transcripts contain recurring misunderstandings, the alternate test is about whether semantic clustering groups related failure patterns for review; it cannot prove the outcome needed after the reported regression where pilot satisfaction hides severe low-frequency failures.
Question 1057
A product owner must measure whether an AI-enabled change actually improves this process: Answers about product policy require traceable evidence. The earlier problem was: Helpful-looking responses contain unverifiable claims. Which outcome indicator is most relevant?
- Authorized structured-record answer accuracy
- Successful prompt-injection override rate
- Percentage of material analytical claims with checked citations
- Share of material claims supported by retrievable citations
Explanation
Correct Answer
D:For the use case where answers about product policy require traceable evidence, the indicator share of material claims supported by retrievable citations is worth baselining against the earlier issue where helpful-looking responses contain unverifiable claims.
Incorrect Answers
A:Although authorized structured-record answer accuracy is observable, the affected process involves answers about product policy require traceable evidence; its acceptance outcome is share of material claims supported by retrievable citations, not the alternative measurement.
B:Third-party documents include sentences telling the model to ignore rules; its intended measurement is successful prompt-injection override rate; substituting that result here would not capture whether helpful-looking responses contain unverifiable claims has been corrected.
C:For the situation where analysts compare unstructured reports but must retain human interpretation, percentage of material analytical claims with checked citations is a sensible indicator; the problem that helpful-looking responses contain unverifiable claims requires measurement of share of material claims supported by retrievable citations.
Question 1058
A product owner must measure whether an AI-enabled change actually improves this process: A canvas app passes customer records to a model action. The earlier problem was: Hidden fields from the UI appear in the prompt payload. Which outcome indicator is most relevant?
- Accepted structured suggestions with correct record linkage
- Successful traceable escalations initiated from assistance
- Prompt action responses conforming to output contract
- Unauthorized data fields sent to model actions
Explanation
Correct Answer
D:This scorecard needs an outcome related to a canvas app passes customer records to a model action; unauthorized data fields sent to model actions. That indicator can be checked against the earlier symptom that hidden fields from the UI appear in the prompt payload.
Incorrect Answers
A:Accepted structured suggestions with correct record linkage; that indicator describes a distinct workflow where inspectors submit a short description and need suggested categorization; it does not quantify unauthorized data fields sent to model actions, which is needed for the stated case.
B:The proposed metric concerns a representative must initiate an exception workflow from the case, whereas the failure in this case is that hidden fields from the UI appear in the prompt payload; the indicator needed here is unauthorized data fields sent to model actions.
C:For the workflow where an agent must turn service notes into a fixed JSON summary, prompt action responses conforming to output contract is an actionable measure; it does not establish whether hidden fields from the UI appear in the prompt payload in the situation under review.
Question 1059
An enterprise team is making a platform design decision. Business context: Employees install external agents with unknown data pathways. Present failure: An unreviewed plugin exports sensitive tenant information. Which approach resolves the mismatch?
- Define policies for third-party agent onboarding
- Verify downstream vendor terms and data handling before integration
- Define trace and conversation retention schedules by sensitivity
- Design a documented agent retirement and offboarding process
Explanation
Correct Answer
A:Without the intended behavior, an unreviewed plugin exports sensitive tenant information; the proposed architecture establishes the specific design principle that procurement and security review evaluate permissions and data flows for the process where employees install external agents with unknown data pathways.
Incorrect Answers
B:There is a valid place for the mechanism that contract and privacy assessment cover service location and subprocessor use, namely a third-party AI tool changes where submitted documents are processed; it does not remove the specific defect that an unreviewed plugin exports sensitive tenant information from this case.
C:In the case where debugging logs include prompts containing customer data, the alternate design depends on retention schedules and minimization control telemetry exposure; it does not correct the target failure, an unreviewed plugin exports sensitive tenant information.
D:The technical behavior of this alternative is that decommissioning revokes credentials, connectors and publishing rights; the user case being assessed is employees install external agents with unknown data pathways with a different failure: an unreviewed plugin exports sensitive tenant information.
Question 1060
Risk reviewers need a control, not merely a monitoring dashboard. The use case is: Operations must distinguish healthy traffic from quality incidents. Potential harm: Alerting might fire on raw request volume while missing user-impacting failures. Which protection fits?
- Alert on meaningful user failures, not just raw token volume
- Fail closed for unsupported sensitive answers
- Fail closed on authorization and schema-validation failures
- Prioritize incidents with safety or privacy impact
Explanation
Correct Answer
A:In the scenario where operations must distinguish healthy traffic from quality incidents, the explicit risk is that alerting might fire on raw request volume while missing user-impacting failures; enforcement must follow the mechanism by which service-level objectives and alert thresholds turn telemetry into response signals.
Incorrect Answers
B:The proposed control targets missing retrieval evidence might still generate definitive sensitive advice, while the given risk scenario is alerting might fire on raw request volume while missing user-impacting failures; these exposures arise in different operational contexts.
C:Another process could need this guardrail because malformed calls might pass a happy-path suite yet mutate records; the design being assessed supports operations must distinguish healthy traffic from quality incidents and has the explicit exposure that alerting might fire on raw request volume while missing user-impacting failures.
D:That protection is tied to the mechanism that structured defect taxonomy assigns the right remediation owner; it does not prevent the distinct risk where alerting might fire on raw request volume while missing user-impacting failures in the system under review.