Anthropic CCAR-P (Claude Certified Architect - Professional) Practice Test 7

Question 60

An executive requests 100% factual correctness for an open-ended knowledge assistant. Which solution best resolves this operational gap?

  1. Benchmark risk-adjusted total cost including retries and review effort before switching tiers.
  2. Measure end-to-end cost and quality rather than model inference duration alone.
  3. Negotiate measurable coverage, citation quality, abstention and review controls instead of impossible guarantees.
  4. Model token, retrieval, tooling and peak concurrency costs with conservative headroom.

Explanation

Correct Answer

C:

Operational commitments can describe tested error rates and escalation, not universal truth for every possible query.

Incorrect Answers

A:

That choice would be responsive to quality-adjusted model economics, but the reported defect is different. It leaves realistic quality commitments unproved because there is no commitment to negotiate measurable coverage, citation quality, abstention and review controls instead of impossible guarantees.

B:

There is a legitimate use case for full-workflow return on investment, but it does not supply the test or enforcement needed for realistic quality commitments. The design needs to negotiate measurable coverage, citation quality, abstention and review controls instead of impossible guarantees.

D:

The alternative could reduce risk associated with sustainable cost at peak usage, yet the described behavior requires realistic quality commitments. That outcome depends on the capability to negotiate measurable coverage, citation quality, abstention and review controls instead of impossible guarantees.

Question 61

A legal team values defensibility more than response speed for regulatory answers. What design decision would best fit these facts?

  1. Segment the workflow by complexity and send low-confidence exceptions to human handling.
  2. Measure end-to-end cost and quality rather than model inference duration alone.
  3. Benchmark risk-adjusted total cost including retries and review effort before switching tiers.
  4. Prioritize source traceability and specialist review over a low-latency autonomous response.

Explanation

Correct Answer

D:

The value function is risk-adjusted; faster unsupported answers have negative utility in this setting. Evaluate the whole operating process for risk-weighted value prioritization, including human review and failures, rather than only the API-level benchmark.

Incorrect Answers

A:

This would make sense as an additional project activity for selective automation by risk. The priority required here is risk-weighted value prioritization, so the workflow still needs a control that will prioritize source traceability and specialist review over a low-latency autonomous response.

B:

By itself, this improves full-workflow return on investment, but the case is not asking the architect to optimize that dimension. The requirement is risk-weighted value prioritization, supported by an approach to prioritize source traceability and specialist review over a low-latency autonomous response.

C:

Its main effect concerns quality-adjusted model economics. The case explicitly hinges on risk-weighted value prioritization; to resolve it reliably, the implementation has to prioritize source traceability and specialist review over a low-latency autonomous response, which this choice does not establish.

Question 62

A procurement pilot shows strong savings for simple invoices but expensive failures for unusual formats. Which intervention most directly addresses the failure?

  1. Prioritize source traceability and specialist review over a low-latency autonomous response.
  2. Weight evaluation and safeguards by consequence, not by frequency alone.
  3. Measure end-to-end cost and quality rather than model inference duration alone.
  4. Segment the workflow by complexity and send low-confidence exceptions to human handling.

Explanation

Correct Answer

D:

Value is maximized by automating reliable cases and controlling high-impact exceptions. Evaluate the whole operating process for selective automation by risk, including human review and failures, rather than only the API-level benchmark.

Incorrect Answers

A:

Although risk-weighted value prioritization can be valuable, it is not the acceptance condition in the question. The missing provision is selective automation by risk, addressed by the control to segment the workflow by complexity and send low-confidence exceptions to human handling.

B:

It is aimed at a different outcome, risk-adjusted performance priorities. No evidence is given that it provides selective automation by risk; the needed mechanism is to segment the workflow by complexity and send low-confidence exceptions to human handling at the appropriate boundary.

C:

It emphasizes full-workflow return on investment without closing the stated gap in selective automation by risk. By contrast, the remedy to segment the workflow by complexity and send low-confidence exceptions to human handling acts on the condition that makes this scenario fail.

Question 63

A business unit demands sub-second responses while insisting on a deep multi-step investigative agent. What should the solution architect prioritize?

  1. Benchmark risk-adjusted total cost including retries and review effort before switching tiers.
  2. Segment the workflow by complexity and send low-confidence exceptions to human handling.
  3. Prioritize source traceability and specialist review over a low-latency autonomous response.
  4. Negotiate measurable coverage, citation quality, abstention and review controls instead of impossible guarantees.
  5. Negotiate a fast triage tier and slower asynchronous deep-analysis path.

Explanation

Correct Answer

E:

User experience and investigation depth impose incompatible latency budgets unless the workload is separated. Evaluate the whole operating process for latency-appropriate service tiers, including human review and failures, rather than only the API-level benchmark.

Incorrect Answers

A:

The alternative could reduce risk associated with quality-adjusted model economics, yet the described behavior requires latency-appropriate service tiers. That outcome depends on the capability to negotiate a fast triage tier and slower asynchronous deep-analysis path.

B:

The proposed measure is aimed at selective automation by risk, not latency-appropriate service tiers. The architecture must instead negotiate a fast triage tier and slower asynchronous deep-analysis path; a separate change is still needed for latency-appropriate service tiers.

C:

The relevant trade-off is not primarily risk-weighted value prioritization. A successful fix needs latency-appropriate service tiers; otherwise latency-appropriate service tiers remains unverified even after this change.

D:

The control would be justified if realistic quality commitments were the limiting factor. In this scenario the decisive target is latency-appropriate service tiers; success requires an architecture that will negotiate a fast triage tier and slower asynchronous deep-analysis path.

Question 64

A regional deployment must keep regulated data inside an approved jurisdiction at every processing stage. Which architecture adjustment is most appropriate?

  1. Choose hosting, logging, storage and connector placement after a full data-flow review.
  2. Choose the simplest maintainable architecture that meets objective thresholds.
  3. Negotiate measurable coverage, citation quality, abstention and review controls instead of impossible guarantees.
  4. Measure end-to-end cost and quality rather than model inference duration alone.

Explanation

Correct Answer

A:

Residency requirements cover derived data and operational telemetry as well as primary prompts. Evaluate the whole operating process for end-to-end residency compliance, including human review and failures, rather than only the API-level benchmark.

Incorrect Answers

B:

The proposed action prioritizes maintainability in architecture choice. That can coexist with the right solution, but it cannot replace end-to-end residency compliance: the application must still choose hosting, logging, storage and connector placement after a full data-flow review.

C:

The design concern is real but different: realistic quality commitments. It does not impose the specific requirement for end-to-end residency compliance; a suitable solution must choose hosting, logging, storage and connector placement after a full data-flow review and demonstrate the result.

D:

This addresses full-workflow return on investment; however, the problem asks for end-to-end residency compliance. Without the safeguard to choose hosting, logging, storage and connector placement after a full data-flow review, the original failure mode is still possible.

Question 65

The model provider offers a cheaper tier that lowers consistency on complex contractual analysis. Which next step is most defensible?

  1. Negotiate a fast triage tier and slower asynchronous deep-analysis path.
  2. Weight evaluation and safeguards by consequence, not by frequency alone.
  3. Model token, retrieval, tooling and peak concurrency costs with conservative headroom.
  4. Benchmark risk-adjusted total cost including retries and review effort before switching tiers.

Explanation

Correct Answer

D:

Lower per-token prices can be offset by worse quality or higher manual remediation. Evaluate the whole operating process for quality-adjusted model economics, including human review and failures, rather than only the API-level benchmark.

Incorrect Answers

A:

That is primarily a technique for latency-appropriate service tiers. It lacks the explicit measure for quality-adjusted model economics that is necessary here, namely a system that will benchmark risk-adjusted total cost including retries and review effort before switching tiers.

B:

Improving risk-adjusted performance priorities does not imply quality-adjusted model economics. The selected design must benchmark risk-adjusted total cost including retries and review effort before switching tiers, whereas this choice omits that direct protection or verification step.

C:

This recommendation targets sustainable cost at peak usage at a different point in the workflow. It leaves quality-adjusted model economics outside the design contract and does not require the team to benchmark risk-adjusted total cost including retries and review effort before switching tiers.

Question 66

A CFO approves funding only if the assistant can prove avoided time against a matched control team. Which approach should be recommended?

  1. Weight evaluation and safeguards by consequence, not by frequency alone.
  2. Establish an evaluation design with baseline, comparable cohorts and attributable savings.
  3. Negotiate measurable coverage, citation quality, abstention and review controls instead of impossible guarantees.
  4. Model token, retrieval, tooling and peak concurrency costs with conservative headroom.

Explanation

Correct Answer

B:

Business value cannot be inferred from raw usage counts; a counterfactual measures actual improvement. Evaluate the whole operating process for credible attribution of business benefit, including human review and failures, rather than only the API-level benchmark.

Incorrect Answers

A:

The control would be justified if risk-adjusted performance priorities were the limiting factor. In this scenario the decisive target is credible attribution of business benefit; success requires an architecture that will establish an evaluation design with baseline, comparable cohorts and attributable savings.

C:

This shifts attention toward realistic quality commitments when the relevant decision is about credible attribution of business benefit. Unless the architecture can establish an evaluation design with baseline, comparable cohorts and attributable savings, the failure described in the stem remains unresolved.

D:

For a workflow whose primary weakness was sustainable cost at peak usage, that proposal could be reasonable. Here the unmet obligation is credible attribution of business benefit, and that requires the system to establish an evaluation design with baseline, comparable cohorts and attributable savings.

Question 67

A supply-chain assistant gets high adoption but fails at the few decisions causing the greatest financial exposure. What is the best production design response?

  1. Segment the workflow by complexity and send low-confidence exceptions to human handling.
  2. Model token, retrieval, tooling and peak concurrency costs with conservative headroom.
  3. Weight evaluation and safeguards by consequence, not by frequency alone.
  4. Benchmark risk-adjusted total cost including retries and review effort before switching tiers.

Explanation

Correct Answer

C:

Rare high-impact failures can dominate expected loss even when average accuracy looks high. Evaluate the whole operating process for risk-adjusted performance priorities, including human review and failures, rather than only the API-level benchmark.

Incorrect Answers

A:

This addresses selective automation by risk; however, the problem asks for risk-adjusted performance priorities. Without the safeguard to weight evaluation and safeguards by consequence, not by frequency alone, the original failure mode is still possible.

B:

That choice would be responsive to sustainable cost at peak usage, but the reported defect is different. It leaves risk-adjusted performance priorities unproved because there is no commitment to weight evaluation and safeguards by consequence, not by frequency alone.

D:

Consider its actual operating scope: quality-adjusted model economics. The requested assurance is risk-adjusted performance priorities, which calls for an implementation that can weight evaluation and safeguards by consequence, not by frequency alone, not merely for a related improvement.

Question 68

A platform team wants an expensive custom agent framework although standard API components already satisfy the requirements. Which change most directly satisfies this requirement?

  1. Choose the simplest maintainable architecture that meets objective thresholds.
  2. Model token, retrieval, tooling and peak concurrency costs with conservative headroom.
  3. Choose hosting, logging, storage and connector placement after a full data-flow review.
  4. Prioritize source traceability and specialist review over a low-latency autonomous response.

Explanation

Correct Answer

A:

Specialized infrastructure incurs long-term operations cost; complexity needs demonstrable capability benefits.

Incorrect Answers

B:

This would make sense as an additional project activity for sustainable cost at peak usage. The priority required here is maintainability in architecture choice, so the workflow still needs a control that will choose the simplest maintainable architecture that meets objective thresholds.

C:

By itself, this improves end-to-end residency compliance, but the case is not asking the architect to optimize that dimension. The requirement is maintainability in architecture choice, supported by an approach to choose the simplest maintainable architecture that meets objective thresholds.

D:

Its main effect concerns risk-weighted value prioritization. The case explicitly hinges on maintainability in architecture choice; to resolve it reliably, the implementation has to choose the simplest maintainable architecture that meets objective thresholds, which this choice does not establish.

Question 69

A straightforward extraction product must choose a Claude model tier under a tight response-time budget. Its pilot used short documents, while production files have much longer inputs and outputs. Which TWO validation decisions make model selection defensible? Choose TWO.

  1. Verify approved model availability and the exact platform data path before choosing a provider endpoint.
  2. Test a stronger reasoning model on a curated difficult-case suite before choosing the production tier.
  3. Benchmark a lower-latency Claude model on extraction accuracy before assigning a higher-capability model by default.
  4. Benchmark language-specific intent, refusal and terminology accuracy before committing.
  5. Benchmark candidate tiers with realistic upper-percentile input and output lengths and expected concurrent throughput.

Explanation

Correct Answers

C:

Simple extraction should start with a lower-latency model candidate, not automatically consume the most capable tier. Selection remains conditional on meeting accuracy and response budgets under the separately specified production-length and concurrency tests.

E:

Longer real documents can alter latency, context fit, and cost even when short pilot cases appear easy. Load tests should include token distributions and concurrency rather than relying on single short-call measurements.

Incorrect Answers

A:

For a workflow whose primary weakness was residency-constrained model availability, that proposal could be reasonable. Here the unmet obligation is fit-for-purpose model choice and model-tier validation at production token lengths and load, and that requires the system to benchmark candidate tiers with realistic upper-percentile input and output lengths and expected concurrent throughput.

B:

Although quality for complex reasoning can be valuable, it is not the acceptance condition in the question. The missing provision is fit-for-purpose model choice and model-tier validation at production token lengths and load, addressed by the control to benchmark a lower-latency Claude model on extraction accuracy before assigning a higher-capability model by default.

D:

It emphasizes multilingual production capability without closing the stated gap in fit-for-purpose model choice and model-tier validation at production token lengths and load. By contrast, the remedy to benchmark a lower-latency Claude model on extraction accuracy before assigning a higher-capability model by default acts on the condition that makes this scenario fail.

Question 70

A litigation analysis pipeline handles long, contradictory evidence sets and errors require expensive attorney review. What design decision would best fit these facts?

  1. Run regression evaluations on contracts, tool calls and exception rates before cutover.
  2. Gate upgrades on measurable gains and compatibility risk, not release recency.
  3. Verify approved model availability and the exact platform data path before choosing a provider endpoint.
  4. Test a stronger reasoning model on a curated difficult-case suite before choosing the production tier.
  5. Benchmark a lower-latency Claude model on extraction accuracy before assigning a higher-capability model by default.

Explanation

Correct Answer

D:

Complex synthesis can justify increased inference cost when it substantially reduces legal-review rework. Confirm quality for complex reasoning with representative task cases and deployment constraints; rankings do not prove that quality for complex reasoning works under this deployment.

Incorrect Answers

A:

Consider its actual operating scope: safe migration between model options. The requested assurance is quality for complex reasoning, which calls for an implementation that can test a stronger reasoning model on a curated difficult-case suite before choosing the production tier, not merely for a related improvement.

B:

The alternative could reduce risk associated with evidence-based upgrade decisions, yet the described behavior requires quality for complex reasoning. That outcome depends on the capability to test a stronger reasoning model on a curated difficult-case suite before choosing the production tier.

C:

The proposed measure is aimed at residency-constrained model availability, not quality for complex reasoning. The architecture must instead test a stronger reasoning model on a curated difficult-case suite before choosing the production tier; a separate change is still needed for quality for complex reasoning.

E:

The control would be justified if fit-for-purpose model choice were the limiting factor. In this scenario the decisive target is quality for complex reasoning; success requires an architecture that will test a stronger reasoning model on a curated difficult-case suite before choosing the production tier.

Claude Certified Architect - Professional (CCAR-P) Practice Tests