Question 10
The fault affects an AI-powered workflow in which agents suggest payment adjustments and inventory holds. Specifically, a new connector allows a service identity to post directly. Which diagnostic investigation best distinguishes the relevant failure path?
- Inspect signatures, hashes and approved source records
- Compare action privileges before and after package import
- Trace source exception through target state and error queue
- Review endpoint access logs and allowed network paths
Explanation
Correct Answer
B:A new connector allows a service identity to post directly; to attribute the defect rather than guess at a fix, comparing action privileges before and after package import examines the dependency where authorization regression gates verify posting authority remains constrained.
Incorrect Answers
A:In the process where deployment automation pulls model artifacts from unverified origins, the diagnostic procedure could explain a replaced model package introduces unknown behavior; this particular incident instead turns on whether authorization regression gates verify posting authority remains constrained and why a new connector allows a service identity to post directly.
C:For the failure that a network fault leaves the two apps inconsistent, tracing source exception through target state and error queue is relevant; this question instead reports that a new connector allows a service identity to post directly, so different evidence is needed.
D:In the separate scenario where a sensitive model endpoint is reachable using a broadly shared key, the same investigation could explain untrusted clients submit requests and exfiltrate outputs; it does not isolate the reported cause in a system where agents suggest payment adjustments and inventory holds.
Question 11
A test plan focuses on a previously missed failure: New prompt automation bypasses supervisor approval. It occurred when service agents draft customer messages but cannot send without review. Which check directly validates that behavior?
- Rebuild a staged agent from recorded versions and run smoke tests
- Try to send high-impact notices without approval in staging
- Promote to staging and confirm correct connection targets
- Import a test solution and verify expected agent behavior
Explanation
Correct Answer
B:The check is targeted rather than generic: business approval checks remain independent of agent version; its pass/fail evidence applies to the workflow where service agents draft customer messages but cannot send without review and to the prior defect that new prompt automation bypasses supervisor approval.
Incorrect Answers
A:For model and tool changes are promoted independently, the alternate test is about whether a complete agent release manifest pins related runtime artifacts; it cannot prove the outcome needed after the reported regression where new prompt automation bypasses supervisor approval.
C:A pass on that verification would speak to a flow moved to production still points to a test endpoint; it would not show that business approval checks remain independent of agent version, which is needed to avoid the observed defect new prompt automation bypasses supervisor approval.
D:The verification concerning the same agent must move from development to test and production checks solution-based ALM transports supported components under controlled releases against manual recreation causes mismatched topics and actions; its result says little about the observed issue where new prompt automation bypasses supervisor approval.
Question 12
An investment review asks what measurable result would justify continuing an AI solution where AI proposes eligibility decisions affecting customers. The original issue is: Supervisors routinely click approve without seeing evidence. Which indicator should be tracked?
- High-impact decisions with substantive human review
- Generated test expectations independently validated
- Critical cases assigned to the correct escalation group
- Users correctly recognizing AI-generated recommendations
Explanation
Correct Answer
A:This scorecard needs an outcome related to AI proposes eligibility decisions affecting customers; high-impact decisions with substantive human review. That indicator can be checked against the earlier symptom that supervisors routinely click approve without seeing evidence.
Incorrect Answers
B:That scorecard concerns the activity in which Copilot writes a test expecting a plausible but wrong answer and its result of generated test expectations independently validated; the reported shortcoming here is supervisors routinely click approve without seeing evidence.
C:Although critical cases assigned to the correct escalation group is observable, the affected process involves AI proposes eligibility decisions affecting customers; its acceptance outcome is high-impact decisions with substantive human review, not the alternative measurement.
D:The premise of this indicator is transparency cues distinguish AI suggestions and human approvals; the architecture being evaluated instead serves AI proposes eligibility decisions affecting customers, and its success is measured by high-impact decisions with substantive human review.
Question 13
In a business situation where a support agent changes how it classifies priority intents, a previous build showed this defect: High-severity tickets begin routing to general queues. Which reproducible test belongs in the release gate?
- Submit urgent and ambiguous cases to the candidate agent
- Promote to staging and confirm correct connection targets
- Import a test solution and verify expected agent behavior
- Attempt an unapproved production modification in test governance
Explanation
Correct Answer
A:Submitting urgent and ambiguous cases to the candidate agent is aligned with the failure that high-severity tickets begin routing to general queues; its causal relevance is that controlled service evaluations validate routing before rollout for the situation where a support agent changes how it classifies priority intents.
Incorrect Answers
B:The verification concerning a flow moved to production still points to a test endpoint checks environment configuration separates deployable logic from endpoint bindings against agent writes real customer data into the sandbox; its result says little about the observed issue where high-severity tickets begin routing to general queues.
C:Although the exercise might catch manual recreation causes mismatched topics and actions, it would not directly validate the mechanism that controlled service evaluations validate routing before rollout in the target implementation.
D:A pass on that verification would speak to production components should not be edited casually; it would not show that controlled service evaluations validate routing before rollout, which is needed to avoid the observed defect high-severity tickets begin routing to general queues.
Question 14
A pilot scorecard needs evidence of business improvement rather than only invocation counts. The use case is: A bot only needs to check warranty and prepare a draft. Known defect: It autonomously contacts a supplier without approval. Which KPI should be selected?
- Valid returns completed without unrelated writes
- Regulated replies with complete mandatory clauses
- Numeric calculations matching the governed reference formula
- Unauthorized ancillary actions per task
Explanation
Correct Answer
D:The use case involves a bot only needs to check warranty and prepare a draft and the baseline failure that it autonomously contacts a supplier without approval; the KPI tracks unauthorized ancillary actions per task, so improvement has an interpretable operational meaning.
Incorrect Answers
A:The alternative tracks valid returns completed without unrelated writes, arising from the mechanism where single-purpose tools and typed parameters bound the task; its movement alone says little about the failure that it autonomously contacts a supplier without approval.
B:Compliance letters must preserve approved language and disclaimers; its intended measurement is regulated replies with complete mandatory clauses; substituting that result here would not capture whether it autonomously contacts a supplier without approval has been corrected.
C:The alternative tracks numeric calculations matching the governed reference formula, arising from the mechanism where deterministic application code computes numeric outputs from authorized data; its movement alone says little about the failure that it autonomously contacts a supplier without approval.
Question 15
Risk reviewers need a control, not merely a monitoring dashboard. The use case is: Quality teams need searchable insights from recorded calls. Potential harm: Recorded customer speech or transcript fields might be exposed to unintended viewers. Which protection fits?
- Require identity checks and human fallback for ambiguous audio
- Require predefined acceptance thresholds before promotion
- Prevent deployment of unapproved model/data combinations
- Protect recordings and redact sensitive transcript elements
Explanation
Correct Answer
D:Quality teams need searchable insights from recorded calls; its risk scenario is that recorded customer speech or transcript fields might be exposed to unintended viewers; the safeguard fits because speech transcription transforms audio to text for subsequent analysis and the restriction addresses that specific exposure.
Incorrect Answers
A:For the use case where callers provide account numbers and amounts by speech, the control addresses the threat that a speech misunderstanding might trigger a high-impact action for the wrong account; the specified agent must be protected against recorded customer speech or transcript fields might be exposed to unintended viewers.
B:A candidate appears accurate on its training examples; its guardrail addresses the possibility that a team might promote a tuned model without meeting a critical holdout threshold; the present architecture must constrain the different threat that recorded customer speech or transcript fields might be exposed to unintended viewers.
C:Where retraining uses revised labels and a different source snapshot, the guard can address a model release might combine training data with an unapproved candidate artifact; the exposure that recorded customer speech or transcript fields might be exposed to unintended viewers remains relevant to the present assessment.
Question 16
Business leaders reject an activity-only metric for this process: Employees express several intents and need interactive clarification. They need to detect the issue where a rigid menu fails whenever a request contains two intents. Which performance indicator fits?
- Percentage of decisions matching the signed decision table
- Successful task completion across unseen paraphrases
- Cross-domain tasks routed to the correct specialist
- Correct classification rate for ambiguous boundary cases
Explanation
Correct Answer
B:The observed shortcoming is that a rigid menu fails whenever a request contains two intents; measuring successful task completion across unseen paraphrases gives business owners a directly relevant indicator for the process where employees express several intents and need interactive clarification.
Incorrect Answers
A:Percentage of decisions matching the signed decision table; that indicator describes a distinct workflow where the task follows fixed approval branches with no interpretation needed; it does not quantify successful task completion across unseen paraphrases, which is needed for the stated case.
C:A change in cross-domain tasks routed to the correct specialist is informative for a request spans procurement, finance and policy expertise; it does not establish whether the current process achieves successful task completion across unseen paraphrases.
D:The proposed metric concerns the classifier confuses complaints with refund requests, whereas the failure in this case is that a rigid menu fails whenever a request contains two intents; the indicator needed here is successful task completion across unseen paraphrases.
Question 17
An enterprise team is making a platform design decision. Business context: An external API changes a required field used by an agent. Present failure: Agent actions fail only after the connector reaches production. Which approach resolves the mismatch?
- Run connector contract tests before every promotion
- Apply data loss prevention policies to connector combinations
- Deploy actions with compatible rollback behavior
- Validate finance and operations virtual-entity mappings after change
Explanation
Correct Answer
A:Because versioned API contracts and consumer checks detect breaking changes, this approach corrects the reported breakdown where agent actions fail only after the connector reaches production; the affected situation is that an external API changes a required field used by an agent.
Incorrect Answers
B:For a Copilot Studio agent can read corporate data and use public messaging, this other approach relies on the fact that Power Platform data policies constrain connector use and movement pathways; the issue here is agent actions fail only after the connector reaches production; that mechanism does not establish the required versioned API contracts and consumer checks detect breaking changes.
C:When a new action version changes how cases are updated, coordinated versioning preserves consumer/provider compatibility; in the stated workflow, however, agent actions fail only after the connector reaches production; the relevant architectural behavior is that versioned API contracts and consumer checks detect breaking changes.
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 external API changes a required field used by an agent with a different failure: agent actions fail only after the connector reaches production.
Question 18
An incident review has one confirmed symptom: Makers validate only a few manual chats. For the use case where prompt revisions cause intermittent regressions, which targeted diagnostic check would reveal the most pertinent evidence?
- Reconcile session outcomes to actual completed records
- Compare generated tests to authentication and schema contracts
- Review evaluator results by task segment and release
- Review the signed acceptance matrix and results
- Audit quarantined document reasons and promotion events
Explanation
Correct Answer
C:The investigation directly tests the causal connection: stable automated suites compare response quality across versions; it collects evidence for a workflow where prompt revisions cause intermittent regressions but the outcome is instead that makers validate only a few manual chats.
Incorrect Answers
A:In the process where a service bot handles many chats but few cases are resolved, the diagnostic procedure could explain a dashboard calls the bot successful because messages increased; this particular incident instead turns on whether stable automated suites compare response quality across versions and why makers validate only a few manual chats.
B:The proposed inspection is associated with agents can call APIs with customer IDs and amount values, where specification-guided generation covers missing fields and denied roles; it would not expose the failure mode that makers validate only a few manual chats with the required specificity.
D:Candidate improves one metric while violating another; examining that other path does not determine whether stable automated suites compare response quality across versions, which is what distinguishes the issue in the stated use case.
E:New documents enter a production source automatically; its matching diagnostic check is reasonable for unreviewed confidential drafts appear in agent answers; the present investigation needs evidence for the different failure where makers validate only a few manual chats.
Question 19
A design adds autonomous behavior to a process where a search index changes embedding model and chunking approach. This creates a documented risk: Mixed embedding or index versions might corrupt retrieval quality. Which enforced control is the best fit?
- Prevent mixed-version retrieval without compatibility checks
- Restrict modifications and protect audit storage from tampering
- Reject mismatched schema/model combinations
- Limit retroactive alteration of approved QA evidence
Explanation
Correct Answer
A:A search index changes embedding model and chunking approach; its risk scenario is that mixed embedding or index versions might corrupt retrieval quality; the safeguard fits because explicit index versions support compatible retrieval and rollback and the restriction addresses that specific exposure.
Incorrect Answers
B:For the use case where a model is replaced without a traceable change ticket, the control addresses the threat that audit logs might be altered to obscure who deployed a model; the specified agent must be protected against mixed embedding or index versions might corrupt retrieval quality.
C:The proposed control targets a new tokenizer might be deployed with an old label mapping, while the given risk scenario is mixed embedding or index versions might corrupt retrieval quality; these exposures arise in different operational contexts.
D:The alternate boundary mitigates old evaluation evidence might be overwritten to conceal a failed release gate, which is relevant to an updated agent performs differently months after release; it leaves the current workflow vulnerable to mixed embedding or index versions might corrupt retrieval quality.
Question 20
A policy review covers an agent for which contact Center agents sometimes cannot resolve complex billing intent. Its most relevant exposure is that a caller might get trapped in repeated fallback without a safe escalation route. What safeguard should be required before release?
- Require explicit consent for consequential create or update operations
- Preserve conversation context and avoid infinite re-prompting
- Limit creation privileges and prevent duplicate escalations
- Limit tool availability and confirmations for risky steps
Explanation
Correct Answer
B:In the scenario where contact Center agents sometimes cannot resolve complex billing intent, the explicit risk is that a caller might get trapped in repeated fallback without a safe escalation route; enforcement must follow the mechanism by which confidence, retry ceilings and escalation routing enable resolution.
Incorrect Answers
A:For the use case where a Teams user should query and update a supported external service, the control addresses the threat that a plugin might create business records without user consent or approval; the specified agent must be protected against a caller might get trapped in repeated fallback without a safe escalation route.
C:For the use case where a representative must initiate an exception workflow from the case, the control addresses the threat that repeated agent calls might create duplicate escalation records or excess privileges; the specified agent must be protected against a caller might get trapped in repeated fallback without a safe escalation route.
D:The alternate boundary mitigates a generative planner might chain tools that exceed the approved workflow scope, which is relevant to users request a status check and an authorized follow-up action; it leaves the current workflow vulnerable to a caller might get trapped in repeated fallback without a safe escalation route.