Question 890
A solution architecture is being reviewed for this business situation: A regulated workload must keep data in permitted locations. The observed defect is that an external model call exports sensitive prompts to an unapproved region. Which architectural decision addresses the underlying requirement?
- Treat index schema and embeddings as versioned artifacts
- Map all agent data flows to approved geographic boundaries
- Apply data loss prevention policies to connector combinations
- Record immutable model deployment events with approver and version
Explanation
Correct Answer
B:A regulated workload must keep data in permitted locations; its failing behavior is that an external model call exports sensitive prompts to an unapproved region; applying the underlying design principle—data-flow architecture identifies processing, storage and telemetry locations—connects the requirement to an implementable control.
Incorrect Answers
A:The alternative addresses existing vectors are queried with incompatible assumptions by a design in which explicit index versions support compatible retrieval and rollback; it does not supply the missing relationship for a regulated workload must keep data in permitted locations, where an external model call exports sensitive prompts to an unapproved region.
C:In the case where a Copilot Studio agent can read corporate data and use public messaging, the alternate design depends on Power Platform data policies constrain connector use and movement pathways; it does not correct the target failure, an external model call exports sensitive prompts to an unapproved region.
D:In the case where a model is replaced without a traceable change ticket, the alternate design depends on deployment audit links actor, approved artifact and release time; it does not correct the target failure, an external model call exports sensitive prompts to an unapproved region.
Question 891
A pilot scorecard needs evidence of business improvement rather than only invocation counts. The use case is: A unique domain vocabulary produces persistent measured errors. Known defect: Teams propose custom training without any baseline. Which KPI should be selected?
- Incremental outcome lift relative to control
- Completed role-specific workflows within supported host surfaces
- Differentiated requirements satisfied without unsupported workarounds
- Format compliance improvement on unseen tasks
- Improvement in domain error rate versus approved baseline
Explanation
Correct Answer
E:Improvement in domain error rate versus approved baseline; this outcome measure belongs to the workflow where a unique domain vocabulary produces persistent measured errors; a change in it can help detect whether the defect that teams propose custom training without any baseline is resolved.
Incorrect Answers
A:A change in incremental outcome lift relative to control is informative for management attributes a sales lift to Copilot without a comparison; it does not establish whether the current process achieves improvement in domain error rate versus approved baseline.
B:Completed role-specific workflows within supported host surfaces; that indicator describes a distinct workflow where staff need a role-specific conversation within Copilot and Teams; it does not quantify improvement in domain error rate versus approved baseline, which is needed for the stated case.
C:Competitive requirements depend on unique multi-model control; its intended measurement is differentiated requirements satisfied without unsupported workarounds; substituting that result here would not capture whether teams propose custom training without any baseline has been corrected.
D:The proposed metric concerns a stable specialized classification style persists despite prompt examples, whereas the failure in this case is that teams propose custom training without any baseline; the indicator needed here is improvement in domain error rate versus approved baseline.
Question 892
Business leaders reject an activity-only metric for this process: An AI triage model under-serves one customer language. They need to detect the issue where overall performance masks unequal escalation rates. Which performance indicator fits?
- Worst-cohort error rate and unjustified outcome gap
- Users correctly recognizing AI-generated recommendations
- Critical cases assigned to the correct escalation group
- Resolved session rate alongside escalation and abandonment
Explanation
Correct Answer
A:The use case involves an AI triage model under-serves one customer language and the baseline failure that overall performance masks unequal escalation rates; the KPI tracks worst-cohort error rate and unjustified outcome gap, so improvement has an interpretable operational meaning.
Incorrect Answers
B:The alternative tracks users correctly recognizing AI-generated recommendations, arising from the mechanism where transparency cues distinguish AI suggestions and human approvals; its movement alone says little about the failure that overall performance masks unequal escalation rates.
C:The premise of this indicator is controlled service evaluations validate routing before rollout; the architecture being evaluated instead serves an AI triage model under-serves one customer language, and its success is measured by worst-cohort error rate and unjustified outcome gap.
D:That scorecard concerns the activity in which support leads need resolution and escalation insights over time and its result of resolved session rate alongside escalation and abandonment; the reported shortcoming here is overall performance masks unequal escalation rates.
Question 893
An investment review asks what measurable result would justify continuing an AI solution where stakeholders report hallucinations but cannot isolate search errors. The original issue is: Generator tuning fails because the required documents were never retrieved. Which indicator should be tracked?
- Precision at top results for identifier and natural-language queries
- Field-level extraction accuracy on representative documents
- Grounding recall for critical enterprise questions
- Correctly transcribed critical utterances
Explanation
Correct Answer
C:This scorecard needs an outcome related to stakeholders report hallucinations but cannot isolate search errors; grounding recall for critical enterprise questions. That indicator can be checked against the earlier symptom that generator tuning fails because the required documents were never retrieved.
Incorrect Answers
A:The premise of this indicator is hybrid ranking combines lexical precision and semantic relevance; the architecture being evaluated instead serves stakeholders report hallucinations but cannot isolate search errors, and its success is measured by grounding recall for critical enterprise questions.
B:The proposed metric concerns vendors send variable-layout invoices and shipping documents, whereas the failure in this case is that generator tuning fails because the required documents were never retrieved; the indicator needed here is grounding recall for critical enterprise questions.
D:For the situation where quality teams need searchable insights from recorded calls, correctly transcribed critical utterances is a sensible indicator; the problem that generator tuning fails because the required documents were never retrieved requires measurement of grounding recall for critical enterprise questions.
Question 894
A pilot scorecard needs evidence of business improvement rather than only invocation counts. The use case is: Sales and service agents need the same customer and case context. Known defect: An opportunity owner sees a stale service escalation status. Which KPI should be selected?
- Wrong-entity retrieval rate in curated evaluation cases
- Share of grounding records with complete governance metadata
- Time from source exception to downstream acknowledgement
- Cross-app case-state agreement rate
Explanation
Correct Answer
D:For the use case where sales and service agents need the same customer and case context, the indicator cross-app case-state agreement rate is worth baselining against the earlier issue where an opportunity owner sees a stale service escalation status.
Incorrect Answers
A:The proposed metric concerns multiple systems use inconsistent customer identifiers and duplicate cases, whereas the failure in this case is that an opportunity owner sees a stale service escalation status; the indicator needed here is cross-app case-state agreement rate.
B:That scorecard concerns the activity in which search spans internal manuals with different owners and permissions and its result of share of grounding records with complete governance metadata; the reported shortcoming here is an opportunity owner sees a stale service escalation status.
C:The premise of this indicator is business events and durable orchestration propagate changes asynchronously; the architecture being evaluated instead serves sales and service agents need the same customer and case context, and its success is measured by cross-app case-state agreement rate.
Question 895
Release reviewers will not accept a generic success message. The business scenario is that staff need a contextual workspace assembled around one customer case, and the historical failure is: Generic chat loses visibility into related records and actions. Which test is appropriately targeted?
- Ask about a shared project name as users from different tenants
- Change a test case status and inspect audit events
- Try the action with a user lacking record access
- Open the page for different roles and verify context limits
Explanation
Correct Answer
D:Generic chat loses visibility into related records and actions; a release test must establish that this no longer occurs; opening the page for different roles and verifying context limits measures the specific behavior for which generated interface components can be grounded in governed app data and defined actions.
Incorrect Answers
A:A pass on that verification would speak to many tenants have identically named projects; it would not show that generated interface components can be grounded in governed app data and defined actions, which is needed to avoid the observed defect generic chat loses visibility into related records and actions.
B:The suggested test serves an agent needs to update a governed ticket status, in which its knowledge source can explain but cannot commit a database change; the present scenario instead concerns the reported problem where generic chat loses visibility into related records and actions.
C:This check exercises a different system behavior: server-side authorization and minimized inputs protect data flow; a pass there cannot establish that generic chat loses visibility into related records and actions no longer occurs in the required workflow.
Question 896
Governance approves the business purpose but requires control of this exposure: An uncertain generated field might silently become authoritative data. The agent will be used when staff reject suggestions because approval controls are unclear. Which boundary is most defensible?
- Let authorized incident commanders override recommendations
- Exclude sensitive attributes not needed by the receiving agent
- Do not auto-save an uncertain prediction as authoritative data
- Limit deployment to approved users and surfaces
Explanation
Correct Answer
C:Staff reject suggestions because approval controls are unclear; its risk scenario is that an uncertain generated field might silently become authoritative data; the safeguard fits because clear review interfaces surface provenance and editable proposals and the restriction addresses that specific exposure.
Incorrect Answers
A:This restriction is associated with a specialized support team uses a nonstandard severity matrix and addresses the exposure that automated severity suggestions might override an incident commander's decision; it does not specifically protect the target workflow against an uncertain generated field might silently become authoritative data.
B:The alternate boundary mitigates a case handoff could include personal attributes irrelevant to the receiving team's work, which is relevant to agents need structured case summaries at handoff; it leaves the current workflow vulnerable to an uncertain generated field might silently become authoritative data.
D:The alternate guard addresses a Teams-targeted agent package might be made available on unapproved hosts for agent experience must appear in selected Microsoft 365 contexts; the present threat, an uncertain generated field might silently become authoritative data, needs a different enforcement boundary.
Question 897
Before introducing a fix, the team must explain why separate assistants report contradictory totals for the same period. The affected setting is: Sales and finance agents must agree on revenue and margin calculations. What should be inspected?
- Identify workflow steps with fixed rules versus variable language
- Compare calculations to the approved finance definition catalogue
- Audit division practices against the common policy catalogue
- Compare event delivery logs with receiving workflow state
Explanation
Correct Answer
B:The reported fault is that separate assistants report contradictory totals for the same period in a system where sales and finance agents must agree on revenue and margin calculations; evidence from comparing calculations to the approved finance definition catalogue can show whether shared metric definitions and lineage prevent model-dependent arithmetic.
Incorrect Answers
A: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 separate assistants report contradictory totals for the same period, so different evidence is needed.
C:Independent teams choose inconsistent evaluation and security methods; its matching diagnostic check is reasonable for one division releases agents without risk review; the present investigation needs evidence for the different failure where separate assistants report contradictory totals for the same period.
D:Comparing event delivery logs with receiving workflow state explores a separate connection, where business events and durable orchestration propagate changes asynchronously; the diagnosed requirement is sales and finance agents must agree on revenue and margin calculations, and the actual defect is separate assistants report contradictory totals for the same period.
Question 898
Governance approves the business purpose but requires control of this exposure: An agent-feed suggestion might trigger a business update before approval. The agent will be used when operations teams review multiple proposed next actions daily. Which boundary is most defensible?
- Sandbox high-risk interactions and require approvals
- Keep authorization at the tool server, not just in agent instructions
- Cap tool depth and require evidence for consequential conclusions
- Require user approval before posting accepted changes
Explanation
Correct Answer
D:In the context where operations teams review multiple proposed next actions daily, the exposure is that an agent-feed suggestion might trigger a business update before approval; containing it requires a guardrail in the operation where a persistent agent activity feed supports review, assignment and disposition.
Incorrect Answers
A:A legacy portal has no integration API for a clerical workflow; its guardrail addresses the possibility that browser-based computer use might click a destructive UI control unexpectedly; the present architecture must constrain the different threat that an agent-feed suggestion might trigger a business update before approval.
B:That protection is tied to the mechanism that a supported MCP server exposes typed tool capabilities to clients; it does not prevent the distinct risk where an agent-feed suggestion might trigger a business update before approval in the system under review.
C:Procurement exceptions require comparing several policies and records; its guardrail addresses the possibility that a reasoning agent might continue invoking tools without evidence or budget limits; the present architecture must constrain the different threat that an agent-feed suggestion might trigger a business update before approval.
Question 899
One enterprise rollout has 2 separately testable requirements. (1) Business use: Multiple AI teams consume a changing source API. Known defect: An upstream field rename silently breaks downstream grounding. (2) Business use: Search spans internal manuals with different owners and permissions. Known defect: A model retrieves a draft manual without any accountable owner. Which 2 architectural choices directly satisfy these stated cases? Choose TWO.
- Publish discoverable data contracts with schema versions
- Use event-driven integration for finance-to-supply-chain exceptions
- Tag documents with ownership, sensitivity, and provenance metadata
- Filter retrieval by effective dates and authoritative version
Explanation
Correct Answers
A:For the independent requirement that multiple AI teams consume a changing source API, versioned schemas and compatibility policy make interfaces predictable; it addresses the defect an upstream field rename silently breaks downstream grounding, while search spans internal manuals with different owners and permissions needs its own architecture.
C:For the independent requirement that search spans internal manuals with different owners and permissions, metadata enables filtering, review, retention and traceable answers; it addresses the defect a model retrieves a draft manual without any accountable owner, while multiple AI teams consume a changing source API needs its own architecture.
Incorrect Answers
B:This design is associated with the different failure where delayed reconciliation hides material fulfillment blockers in a workflow where invoice holds must trigger exception handling without polling; it does not directly solve the listed conditions: an upstream field rename silently breaks downstream grounding; a model retrieves a draft manual without any accountable owner.
D:This design is associated with the different failure where the agent cites a superseded rule despite a newer policy existing in a workflow where policy documents contain draft, superseded, and currently effective revisions; it does not directly solve the listed conditions: an upstream field rename silently breaks downstream grounding; a model retrieves a draft manual without any accountable owner.
Question 900
A quality engineer is writing a regression case after the incident where existing vectors are queried with incompatible assumptions. The underlying business scenario is: A search index changes embedding model and chunking approach. What should the test exercise?
- Ask about identical case names from isolated tenants
- Rebuild a test index and switch traffic under observation
- Invoke both versions with boundary category samples
- Promote a candidate in staging and inspect generated audit entries
- Exercise both old and new agent builds against staged endpoints
Explanation
Correct Answer
B:Existing vectors are queried with incompatible assumptions; a release test must establish that this no longer occurs; rebuilding a test index and switch traffic under observation measures the specific behavior for which explicit index versions support compatible retrieval and rollback.
Incorrect Answers
A:In a solution where two customers store similarly named support incidents, asking about identical case names from isolated tenants could catch the defect that assistant retrieves another tenant's incident detail; the current release gate must challenge the failure where existing vectors are queried with incompatible assumptions.
C:Although the exercise might catch downstream apps interpret correct predictions as the wrong category, it would not directly validate the mechanism that explicit index versions support compatible retrieval and rollback in the target implementation.
D:In the alternate setting where a model is replaced without a traceable change ticket, the test exercises deployment audit links actor, approved artifact and release time; even a pass would not show that existing vectors are queried with incompatible assumptions has been prevented.
E:In a solution where a new action version changes how cases are updated, exercising both old and new agent builds against staged endpoints could catch the defect that rolling back the agent leaves an incompatible backend action; the current release gate must challenge the failure where existing vectors are queried with incompatible assumptions.