Question 340
A sales presentation promises that a support assistant can keep and search unlimited uploaded documents within a single chat. Actual context and ingestion limits require durable indexing of large files. How should the architect reset the service expectation?
- Promise unlimited context anyway because the application can retry model calls as often as needed.
- Disable upload-size alerts but continue advertising that every attachment remains in model memory forever.
- Describe model output quality only and exclude indexing or document lifecycle behavior from the service agreement.
- Distinguish conversation context from durable document retrieval and publish tested file limits, indexing behavior and fallback commitments.
Explanation
Correct Answer
D:A chat context window is not an unlimited document repository. Communicating tested ingestion sizes, retrieval coverage and what happens before an index is ready creates a contract that the application can actually deliver.
Incorrect Answers
A:Retrying does not remove context-window or ingestion constraints. The product needs an explicit supported capacity and a defined large-file workflow.
B:Suppressing warnings makes an unsupported promise harder to detect. Durable storage and retrieval must be separately designed and disclosed.
C:Answer fluency cannot guarantee that an unindexed or expired document is retrievable. Input capacity and readiness are observable parts of the user-facing service commitment.
Question 341
A customer wants sub-second latency for reports that call three external research services. What design decision would best fit these facts?
- Offer a fast acknowledgment and asynchronous completion SLA with status updates.
- Commit to percentile latency and error budgets rather than the mean alone.
- Define supported scope and refusal behavior as part of the user-facing service description.
- Negotiate measurable groundedness, coverage, abstention and escalation targets.
Explanation
Correct Answer
A:Deep external work may be better governed by completion targets than interactive response deadlines. State realistic asynchronous delivery commitment as an operating commitment with a measurable boundary, dependencies and a named escalation owner.
Incorrect Answers
B:Consider its actual operating scope: tail-aware latency objectives. The requested assurance is realistic asynchronous delivery commitment, which calls for an implementation that can offer a fast acknowledgment and asynchronous completion SLA with status updates, not merely for a related improvement.
C:The alternative could reduce risk associated with explicit supported-answer scope, yet the described behavior requires realistic asynchronous delivery commitment. That outcome depends on the capability to offer a fast acknowledgment and asynchronous completion SLA with status updates.
D:The proposed measure is aimed at defensible answer-quality commitments, not realistic asynchronous delivery commitment. The architecture must instead offer a fast acknowledgment and asynchronous completion SLA with status updates; a separate change is still needed for realistic asynchronous delivery commitment.
Question 342
A legal department treats generated recommendations as final organizational decisions. Which intervention most directly addresses the failure?
- Define alert ownership, escalation time and incident response objectives.
- Define supported scope and refusal behavior as part of the user-facing service description.
- Define end-to-end SLIs that include retrieval, tools, queueing and external dependencies.
- Define system advisory status and the human authority needed for consequential action.
Explanation
Correct Answer
D:Clear responsibility boundaries prevent mistaken transfer of decision accountability to a model. State decision authority expectations as an operating commitment with a measurable boundary, dependencies and a named escalation owner.
Incorrect Answers
A:By itself, this improves actionable safety incident commitments, but the case is not asking the architect to optimize that dimension. The requirement is decision authority expectations, supported by an approach to define system advisory status and the human authority needed for consequential action.
B:Its main effect concerns explicit supported-answer scope. The case explicitly hinges on decision authority expectations; to resolve it reliably, the implementation has to define system advisory status and the human authority needed for consequential action, which this choice does not establish.
C:The option might improve whole-service reliability slas; that remains insufficient for decision authority expectations. The case is resolved by decision authority expectations, which depends on a design that will define system advisory status and the human authority needed for consequential action.
Question 343
A procurement contract quotes API uptime as though it guarantees end-user workflow availability. What should the solution architect prioritize?
- Define end-to-end SLIs that include retrieval, tools, queueing and external dependencies.
- Publish dependencies, decision gates and contingency dates in the delivery baseline.
- Define supported scope and refusal behavior as part of the user-facing service description.
- Commit to percentile latency and error budgets rather than the mean alone.
- Define alert ownership, escalation time and incident response objectives.
Explanation
Correct Answer
A:Model availability is only one component of the actual delivered service. State whole-service reliability SLAs as an operating commitment with a measurable boundary, dependencies and a named escalation owner.
Incorrect Answers
B:It emphasizes dependency-aware schedule promises without closing the stated gap in whole-service reliability slas. By contrast, the remedy to define end-to-end SLIs that include retrieval, tools, queueing and external dependencies acts on the condition that makes this scenario fail.
C:That is primarily a technique for explicit supported-answer scope. It lacks the explicit measure for whole-service reliability slas that is necessary here, namely a system that will define end-to-end SLIs that include retrieval, tools, queueing and external dependencies.
D:Improving tail-aware latency objectives does not imply whole-service reliability slas. The selected design must define end-to-end SLIs that include retrieval, tools, queueing and external dependencies, whereas this choice omits that direct protection or verification step.
E:This recommendation targets actionable safety incident commitments at a different point in the workflow. It leaves whole-service reliability slas outside the design contract and does not require the team to define end-to-end SLIs that include retrieval, tools, queueing and external dependencies.
Question 344
A partner asks for a fixed operating budget despite unknown seasonal traffic peaks. Which architecture adjustment is most appropriate?
- Define end-to-end SLIs that include retrieval, tools, queueing and external dependencies.
- Negotiate measurable groundedness, coverage, abstention and escalation targets.
- State demand assumptions, token estimates, variance bands and cost controls.
- Define alert ownership, escalation time and incident response objectives.
Explanation
Correct Answer
C:A cost forecast is credible only with explicit volume and complexity uncertainty. State transparent operating cost assumptions as an operating commitment with a measurable boundary, dependencies and a named escalation owner.
Incorrect Answers
A:The proposed measure is aimed at whole-service reliability slas, not transparent operating cost assumptions. The architecture must instead state demand assumptions, token estimates, variance bands and cost controls; a separate change is still needed for transparent operating cost assumptions.
B:The relevant trade-off is not primarily defensible answer-quality commitments. A successful fix needs transparent operating cost assumptions; otherwise transparent operating cost assumptions remains unverified even after this change.
D:The option belongs to the broader work on actionable safety incident commitments, but it cannot by itself deliver transparent operating cost assumptions. A decision on this case needs the team to state demand assumptions, token estimates, variance bands and cost controls and verify that outcome.
Question 345
A support agent's average response time is good but one in twenty tickets exceeds 20 seconds. Which next step is most defensible?
- Define end-to-end SLIs that include retrieval, tools, queueing and external dependencies.
- Publish dependencies, decision gates and contingency dates in the delivery baseline.
- Commit to percentile latency and error budgets rather than the mean alone.
- Define alert ownership, escalation time and incident response objectives.
Explanation
Correct Answer
C:Tail performance captures the delays that cause real user frustration. State tail-aware latency objectives as an operating commitment with a measurable boundary, dependencies and a named escalation owner.
Incorrect Answers
A:The proposed action prioritizes whole-service reliability slas. That can coexist with the right solution, but it cannot replace tail-aware latency objectives: the application must still commit to percentile latency and error budgets rather than the mean alone.
B:The design concern is real but different: dependency-aware schedule promises. It does not impose the specific requirement for tail-aware latency objectives; a suitable solution must commit to percentile latency and error budgets rather than the mean alone and demonstrate the result.
D:That choice would be responsive to actionable safety incident commitments, but the reported defect is different. It leaves tail-aware latency objectives unproved because there is no commitment to commit to percentile latency and error budgets rather than the mean alone.
Question 346
A pilot sponsor expects automation to remove reviewers, while the approved design includes mandatory dual control. Which approach should be recommended?
- Quantify review effort and explain its risk-control purpose before contracting for savings.
- Define alert ownership, escalation time and incident response objectives.
- Publish dependencies, decision gates and contingency dates in the delivery baseline.
- Define system advisory status and the human authority needed for consequential action.
Explanation
Correct Answer
A:Risk acceptance and staffing economics must align with governance obligations. State realistic human-review resource planning as an operating commitment with a measurable boundary, dependencies and a named escalation owner.
Incorrect Answers
B:This addresses actionable safety incident commitments; however, the problem asks for realistic human-review resource planning. Without the safeguard to quantify review effort and explain its risk-control purpose before contracting for savings, the original failure mode is still possible.
C:This is a valid architectural concern when optimizing dependency-aware schedule promises. Here it leaves realistic human-review resource planning unaddressed, whereas the intended change must quantify review effort and explain its risk-control purpose before contracting for savings before the case can be considered resolved.
D:This would make sense as an additional project activity for decision authority expectations. The priority required here is realistic human-review resource planning, so the workflow still needs a control that will quantify review effort and explain its risk-control purpose before contracting for savings.
Question 347
A deployment timeline assumes data access is already approved, although legal review has not started. What is the best production design response?
- Define alert ownership, escalation time and incident response objectives.
- Negotiate measurable groundedness, coverage, abstention and escalation targets.
- Publish dependencies, decision gates and contingency dates in the delivery baseline.
- Define end-to-end SLIs that include retrieval, tools, queueing and external dependencies.
Explanation
Correct Answer
C:Service commitments should distinguish engineering effort from external approvals on the critical path. State dependency-aware schedule promises as an operating commitment with a measurable boundary, dependencies and a named escalation owner.
Incorrect Answers
A:The option belongs to the broader work on actionable safety incident commitments, but it cannot by itself deliver dependency-aware schedule promises. A decision on this case needs the team to publish dependencies, decision gates and contingency dates in the delivery baseline and verify that outcome.
B:This shifts attention toward defensible answer-quality commitments when the relevant decision is about dependency-aware schedule promises. Unless the architecture can publish dependencies, decision gates and contingency dates in the delivery baseline, the failure described in the stem remains unresolved.
D:Although whole-service reliability slas can be valuable, it is not the acceptance condition in the question. The missing provision is dependency-aware schedule promises, addressed by the control to publish dependencies, decision gates and contingency dates in the delivery baseline.
Question 348
A client expects no downtime during all model migrations and zero change in outputs. Which change most directly satisfies this requirement?
- Define alert ownership, escalation time and incident response objectives.
- Publish dependencies, decision gates and contingency dates in the delivery baseline.
- Commit to tested rollout and rollback procedures plus measured regression tolerance.
- Negotiate measurable groundedness, coverage, abstention and escalation targets.
Explanation
Correct Answer
C:Observable compatibility and recoverability are defensible; identical outputs across models are not guaranteed.
Incorrect Answers
A:The proposed measure is aimed at actionable safety incident commitments, not credible upgrade service commitments. The architecture must instead commit to tested rollout and rollback procedures plus measured regression tolerance; a separate change is still needed for credible upgrade service commitments.
B:There is a legitimate use case for dependency-aware schedule promises, but it does not supply the test or enforcement needed for credible upgrade service commitments. The design needs to commit to tested rollout and rollback procedures plus measured regression tolerance.
D:The alternative could reduce risk associated with defensible answer-quality commitments, yet the described behavior requires credible upgrade service commitments. That outcome depends on the capability to commit to tested rollout and rollback procedures plus measured regression tolerance.
Question 349
A sales team promises that every question will be answered even if source documents are missing. Which solution best resolves this operational gap?
- Define system advisory status and the human authority needed for consequential action.
- Define end-to-end SLIs that include retrieval, tools, queueing and external dependencies.
- Negotiate measurable groundedness, coverage, abstention and escalation targets.
- Define supported scope and refusal behavior as part of the user-facing service description.
Explanation
Correct Answer
D:Truthful capability boundaries prevent unsupported completion being misinterpreted as a breach. State explicit supported-answer scope as an operating commitment with a measurable boundary, dependencies and a named escalation owner.
Incorrect Answers
A:This addresses decision authority expectations; however, the problem asks for explicit supported-answer scope. Without the safeguard to define supported scope and refusal behavior as part of the user-facing service description, the original failure mode is still possible.
B:By itself, this improves whole-service reliability slas, but the case is not asking the architect to optimize that dimension. The requirement is explicit supported-answer scope, supported by an approach to define supported scope and refusal behavior as part of the user-facing service description.
C:Its main effect concerns defensible answer-quality commitments. The case explicitly hinges on explicit supported-answer scope; to resolve it reliably, the implementation has to define supported scope and refusal behavior as part of the user-facing service description, which this choice does not establish.
Question 350
A governance board asks who receives alerts when the assistant's harmful-output rate crosses a threshold. What design decision would best fit these facts?
- Negotiate measurable groundedness, coverage, abstention and escalation targets.
- Define end-to-end SLIs that include retrieval, tools, queueing and external dependencies.
- Publish dependencies, decision gates and contingency dates in the delivery baseline.
- Commit to percentile latency and error budgets rather than the mean alone.
- Define alert ownership, escalation time and incident response objectives.
Explanation
Correct Answer
E:Operational service levels need accountable humans, not only machine metric dashboards. State actionable safety incident commitments as an operating commitment with a measurable boundary, dependencies and a named escalation owner.
Incorrect Answers
A:Although defensible answer-quality commitments can be valuable, it is not the acceptance condition in the question. The missing provision is actionable safety incident commitments, addressed by the control to define alert ownership, escalation time and incident response objectives.
B:It is aimed at a different outcome, whole-service reliability slas. No evidence is given that it provides actionable safety incident commitments; the needed mechanism is to define alert ownership, escalation time and incident response objectives at the appropriate boundary.
C:It emphasizes dependency-aware schedule promises without closing the stated gap in actionable safety incident commitments. By contrast, the remedy to define alert ownership, escalation time and incident response objectives acts on the condition that makes this scenario fail.
D:That is primarily a technique for tail-aware latency objectives. It lacks the explicit measure for actionable safety incident commitments that is necessary here, namely a system that will define alert ownership, escalation time and incident response objectives.