Question 540
A risk analysis highlights a threat in this workflow: Multiple systems use inconsistent customer identifiers and duplicate cases. Exposure: Similar customer identities could be merged without sufficient matching evidence. Which enforceable safeguard addresses it at the relevant boundary?
- Roll back route changes if quality guardrails regress
- Retire obsolete shared components and notify dependent teams
- Enforce business-unit and row-level permissions at the data layer
- Perform authorization and invariant checks in the backend
- Reject ambiguous identity matches rather than guessing
Explanation
Correct Answer
E:In the scenario where multiple systems use inconsistent customer identifiers and duplicate cases, the explicit risk is that similar customer identities could be merged without sufficient matching evidence; enforcement must follow the mechanism by which canonical keys and deduplication prevent incorrect cross-record joins.
Incorrect Answers
A:Real task mix differs from prelaunch performance tests; its guardrail addresses the possibility that router optimization could cut cost while materially worsening critical outcomes; the present architecture must constrain the different threat that similar customer identities could be merged without sufficient matching evidence.
B:Another process could need this guardrail because unmaintained reusable components could continue being used after deprecation; the design being assessed supports multiple systems use inconsistent customer identifiers and duplicate cases and has the explicit exposure that similar customer identities could be merged without sufficient matching evidence.
C:Where agents and apps need shared definitions of customers and work orders, the guard can address shared data tables might expose rows to a different business unit; the exposure that similar customer identities could be merged without sufficient matching evidence remains relevant to the present assessment.
D:For the use case where the model submits malformed identifiers to an operational API, the control addresses the threat that a well-formed tool request could still violate authorization or business invariants; the specified agent must be protected against similar customer identities could be merged without sufficient matching evidence.
Question 541
The architecture board has evidence that the pilot has no team for monitoring, incidents, or iteration. The affected situation is: An agent will evolve continuously after its initial project delivery. What change most directly establishes the missing behavior?
- Establish a product-oriented operating model for AI services
- Sequence AI adoption by data and security maturity
- Establish a cross-functional AI Center of Excellence playbook
- Decompose end-to-end workflows into agent-appropriate tasks
Explanation
Correct Answer
A:The architecture must reconcile two facts: an agent will evolve continuously after its initial project delivery, yet the pilot has no team for monitoring, incidents, or iteration; the selected pattern does so through the mechanism that long-lived product ownership funds ongoing evaluation and operations.
Incorrect Answers
B:For another type of architecture, readiness assessment places safe high-value workloads ahead of risky ones is useful; this design must instead support an agent will evolve continuously after its initial project delivery despite the problem where the pilot has no team for monitoring, incidents, or iteration.
C:One division releases agents without risk review; the alternate design can address that situation because a CoE publishes patterns, reusable controls, standards and escalation paths, whereas this question needs a solution to the pilot has no team for monitoring, incidents, or iteration.
D:For a business process mixes deterministic calculations and open-ended judgments, this other approach relies on the fact that task decomposition separates deterministic services from bounded agent reasoning; the issue here is the pilot has no team for monitoring, incidents, or iteration; that mechanism does not establish the required long-lived product ownership funds ongoing evaluation and operations.
Question 542
Security and business owners want to keep this capability: Many tenants have identically named projects. They must also prevent the following: Search might surface a record from another tenant with an identical project title. Which guardrail should the design incorporate?
- Enforce tenant partitioning in storage and retrieval
- Require reviewer approval before outbound communication
- Keep approved alternate processing for endpoint outages
- Prevent deployment of unapproved model/data combinations
Explanation
Correct Answer
A:Because search might surface a record from another tenant with an identical project title, a guardrail must constrain the actual system behavior: tenant and entity filters constrain candidate records before generation; protecting a response alone cannot enforce that boundary.
Incorrect Answers
B:Service representatives draft responses from authorized knowledge; its guardrail addresses the possibility that aI-drafted customer responses might be sent before the responsible reviewer approves; the present architecture must constrain the different threat that search might surface a record from another tenant with an identical project title.
C:The alternate boundary mitigates an inference outage might leave customers without any approved fallback, which is relevant to a specialized model must serve production apps reliably; it leaves the current workflow vulnerable to search might surface a record from another tenant with an identical project title.
D:For the use case where retraining uses revised labels and a different source snapshot, the control addresses the threat that a model release might combine training data with an unapproved candidate artifact; the specified agent must be protected against search might surface a record from another tenant with an identical project title.
Question 543
A planned AI solution must support this situation: The planner calls an order-cancel tool for status requests. Its prototype fails because final answer text conceals the erroneous tool choice. Which design best fits the constraint?
- Test long-running and partially completed plans
- Link negative feedback to grounded traces and source versions
- Inspect tool-selection spans when orchestration chooses wrong actions
- Compare traces before and after orchestration configuration changes
Explanation
Correct Answer
C:The planner calls an order-cancel tool for status requests; its failing behavior is that final answer text conceals the erroneous tool choice; applying the underlying design principle—step traces expose planning, selected tools and arguments—connects the requirement to an implementable control.
Incorrect Answers
A:In the case where an autonomous agent relies on slow approvals and delayed tool results, the alternate design depends on state-machine tests examine retries, expiry and compensation; it does not correct the target failure, final answer text conceals the erroneous tool choice.
B:For another type of architecture, trace and content-version linkage makes failures testable is useful; this design must instead support the planner calls an order-cancel tool for status requests despite the problem where final answer text conceals the erroneous tool choice.
D:For a new topic description changes which agent is invoked, this other approach relies on the fact that trace comparisons reveal shifts in plan selection and step order; the issue here is final answer text conceals the erroneous tool choice; that mechanism does not establish the required step traces expose planning, selected tools and arguments.
Question 544
An incident review has one confirmed symptom: Direct replacement removes the dependable baseline. For the use case where a tuned model claims higher accuracy but adds latency, which targeted diagnostic check would reveal the most pertinent evidence?
- Audit quarantined document reasons and promotion events
- Match release manifest to transformation and model checksums
- Compare errors by scan quality and source drift
- Analyze segment-specific score and throughput differences
Explanation
Correct Answer
D:Where direct replacement removes the dependable baseline, the useful evidence comes from analyzing segment-specific score and throughput differences; its focus matches the architectural mechanism that candidate comparison evaluates quality, latency and safety under matched load in the process where a tuned model claims higher accuracy but adds latency.
Incorrect Answers
A:The proposed probe concerns new documents enter a production source automatically, where ingestion gates check classification, freshness and ownership; this incident instead requires evidence about candidate comparison evaluates quality, latency and safety under matched load and the failure that direct replacement removes the dependable baseline.
B:A preprocessing release changes category labels incorrectly; its matching diagnostic check is reasonable for restoring code alone does not restore previous predictions; the present investigation needs evidence for the different failure where direct replacement removes the dependable baseline.
C:For the failure that scanned forms suddenly produce inaccurate predictions, comparing errors by scan quality and source drift is relevant; this question instead reports that direct replacement removes the dependable baseline, so different evidence is needed.
Question 545
Security and business owners want to keep this capability: Unused agents retain tokens and active integrations. They must also prevent the following: Retired agents might keep active tokens or leave no necessary audit history. Which guardrail should the design incorporate?
- Block release if privileged writes bypass approvals
- Limit production edit roles and monitor drift
- Preserve required audit records while disabling operations
- Mask sensitive input/output fields and control log access
Explanation
Correct Answer
C:Because retired agents might keep active tokens or leave no necessary audit history, a guardrail must constrain the actual system behavior: decommissioning revokes credentials, connectors and publishing rights; protecting a response alone cannot enforce that boundary.
Incorrect Answers
A:Where agents suggest payment adjustments and inventory holds, the guard can address an updated connector might gain a direct privileged posting route; the exposure that retired agents might keep active tokens or leave no necessary audit history remains relevant to the present assessment.
B:The alternate guard addresses direct production customization might diverge from the approved solution for production components should not be edited casually; the present threat, retired agents might keep active tokens or leave no necessary audit history, needs a different enforcement boundary.
D:For the use case where a newly deployed Foundry agent has no detailed diagnostics, the control addresses the threat that tracing might store secrets and personal content without adequate access controls; the specified agent must be protected against retired agents might keep active tokens or leave no necessary audit history.
Question 546
After deployment, stakeholders need an interpretable measure for this use case: Shared result caching makes answers fast across roles. The pilot showed that one user's cached confidential answer appears for another user. Which metric captures the needed result?
- Unauthorized cross-environment requests blocked
- Improvement in evidence-supported answer correctness
- Unauthorized grounding documents included in answers
- Cross-role cache disclosure incidents
Explanation
Correct Answer
D:This scorecard needs an outcome related to shared result caching makes answers fast across roles; cross-role cache disclosure incidents. That indicator can be checked against the earlier symptom that one user's cached confidential answer appears for another user.
Incorrect Answers
A:For the situation where test automation must not reach live customer records, unauthorized cross-environment requests blocked is a sensible indicator; the problem that one user's cached confidential answer appears for another user requires measurement of cross-role cache disclosure incidents.
B:Although improvement in evidence-supported answer correctness is observable, the affected process involves shared result caching makes answers fast across roles; its acceptance outcome is cross-role cache disclosure incidents, not the alternative measurement.
C:Users have different rights to the same knowledge corpus; its intended measurement is unauthorized grounding documents included in answers; substituting that result here would not capture whether one user's cached confidential answer appears for another user has been corrected.
Question 547
Security and business owners want to keep this capability: Several events refer to one business order at once. They must also prevent the following: Simultaneous events might produce duplicate financial side effects. Which guardrail should the design incorporate?
- Use idempotency keys and locking where the workflow requires it
- Use idempotency on any repeatable mutation action
- Require explicit approval when critical thresholds fail
- Define completion against system-of-record changes where relevant
Explanation
Correct Answer
A:Avoiding the consequence where simultaneous events might produce duplicate financial side effects requires a control for several events refer to one business order at once; in this architecture, concurrency evaluation tests idempotency and serialization, which is the point at which the boundary can be applied.
Incorrect Answers
B:The safeguard for agents modify records through external APIs deals with a different threat: retries might cause repeated transactions even though tool calls appear successful; even after it is enforced, the exposure that simultaneous events might produce duplicate financial side effects remains possible in the target workflow where concurrency evaluation tests idempotency and serialization.
C:For the use case where prompt revisions cause intermittent regressions, the control addresses the threat that a maker might publish a candidate despite a failed critical evaluation gate; the specified agent must be protected against simultaneous events might produce duplicate financial side effects.
D:The proposed control targets a friendly conversation might be called resolved even though no business action completed, while the given risk scenario is simultaneous events might produce duplicate financial side effects; these exposures arise in different operational contexts.
Question 548
An agent-assisted service does not behave as specified: Answers cite isolated sentences without section qualifiers. It is used in this situation: Retrieved policy fragments lose context when documents are split. Which observable diagnostic procedure is best?
- Trace page data bindings and invoked record APIs
- Compare training scores with independently curated validation results
- Inspect outbound payloads and role-based access decisions
- Inspect chunk boundaries and parent document links
Explanation
Correct Answer
D:In the setting where retrieved policy fragments lose context when documents are split, inspecting chunk boundaries and parent document links can isolate the failure in which answers cite isolated sentences without section qualifiers, because semantically coherent chunks with provenance preserve context during retrieval.
Incorrect Answers
A:That probe evaluates generated interface components can be grounded in governed app data and defined actions in a workflow where staff need a contextual workspace assembled around one customer case; the reported failure concerns the path for which semantically coherent chunks with provenance preserve context during retrieval instead.
B:A candidate appears accurate on its training examples; its matching diagnostic check is reasonable for overfitting hides failures on new business records; the present investigation needs evidence for the different failure where answers cite isolated sentences without section qualifiers.
C:A canvas app passes customer records to a model action; its matching diagnostic check is reasonable for hidden fields from the UI appear in the prompt payload; the present investigation needs evidence for the different failure where answers cite isolated sentences without section qualifiers.
Question 549
A staging test has to detect the specific defect that assistant retrieves another tenant's incident detail. In the workflow where two customers store similarly named support incidents, which exercise verifies the corrective behavior?
- Ask about identical case names from isolated tenants
- Try a forbidden refund despite a permissive model instruction
- Reindex a modified document and verify traceability
- Pilot a connector with synthetic data and review destinations
- Attempt a cross-environment write using staging credentials
- Modify a safe test purchase order and verify the mapped output
Explanation
Correct Answer
A:Assistant retrieves another tenant's incident detail; a release test must establish that this no longer occurs; asking about identical case names from isolated tenants measures the specific behavior for which partitioning and authorized filters constrain relevant candidates.
Incorrect Answers
B:The suggested test serves prompt wording persuades the agent to submit a privileged operation, in which API trusts the model's text assertion of permission; the present scenario instead concerns the reported problem where assistant retrieves another tenant's incident detail.
C:For agents begin citing an obsolete article after an ingestion job, the alternate test is about whether content lineage records update, transformation and index revision; it cannot prove the outcome needed after the reported regression where assistant retrieves another tenant's incident detail.
D:Piloting a connector with synthetic data and reviewing destinations verifies an adjacent objective where employees install external agents with unknown data pathways; it leaves untested the causal path in which partitioning and authorized filters constrain relevant candidates for this case.
E:The tested behavior is that environment-scoped identities isolate access and simplify incident containment; the necessary evidence here concerns partitioning and authorized filters constrain relevant candidates and the previously observed assistant retrieves another tenant's incident detail.
F:This check exercises a different system behavior: integration tests protect virtual-entity schema and source semantics; a pass there cannot establish that assistant retrieves another tenant's incident detail no longer occurs in the required workflow.
Question 550
In a business situation where the planner calls an order-cancel tool for status requests, a previous build showed this defect: Final answer text conceals the erroneous tool choice. Which reproducible test belongs in the release gate?
- Load-test burst traffic and compare latency percentiles
- Delay an approver past the defined execution deadline
- Create a synthetic trace and verify planned expiry behavior
- Reproduce the intent using a captured test trace
Explanation
Correct Answer
D:The historical failure is that final answer text conceals the erroneous tool choice; for the setting where the planner calls an order-cancel tool for status requests, reproducing the intent using a captured test trace verifies the dependency where step traces expose planning, selected tools and arguments.
Incorrect Answers
A:The suggested test serves a small portion of calls time out during peak demand, in which mean response time looks acceptable; the present scenario instead concerns the reported problem where final answer text conceals the erroneous tool choice.
B:Although the exercise might catch timeout creates an abandoned workflow with no owner, it would not directly validate the mechanism that step traces expose planning, selected tools and arguments in the target implementation.
C:The suggested test serves debugging logs include prompts containing customer data, in which historical traces remain available longer than approved retention; the present scenario instead concerns the reported problem where final answer text conceals the erroneous tool choice.