Question 340
While using Claude for a staff training handout at a hospital education team, a training designer draws on approved training materials for new administrators. Two distinct defects require action: a project starts producing answers based on an obsolete procedural handbook; additionally, several independent workstreams need different instructions and reference sets. Choose TWO actions that directly remedy the two stated defects.
- Upload current approved files to project knowledge
- Separate access scopes and review sharing permissions before adding documents
- Create appropriately scoped projects with distinct knowledge and guidance
- Add the conventions as project instructions
- Replace or update the source and document the effective version
Explanation
Correct Answers
C:For a workflow at a hospital education team, separating workspaces limits context contamination and keeps each team's approved rules relevant. In a staff training handout, this directly addresses one of the stated issues: several independent workstreams need different instructions and reference sets. In checking a staff training handout at a hospital education team, the team can test the improvement addressing several independent workstreams need different instructions and reference sets against approved training materials before distributing a staff training handout to new administrators.
E:For a workflow at a hospital education team, knowledge maintenance matters because model fluency will not warn users that an uploaded file is superseded. In a staff training handout, this directly addresses one of the stated issues: a project starts producing answers based on an obsolete procedural handbook. In checking a staff training handout at a hospital education team, the team can test the improvement addressing a project starts producing answers based on an obsolete procedural handbook against approved training materials before distributing a staff training handout to new administrators.
Incorrect Answers
A:For a staff training handout, project knowledge makes curated source material available to relevant project discussions under the workspace's access rules. It would be reasonable in a case where a team has reference documents that should ground multiple project conversations, but the request for a staff training handout at a hospital education team asks for remedies to these particular issues: a project starts producing answers based on an obsolete procedural handbook; several independent workstreams need different instructions and reference sets. Neither the evidence nor the constraints in a staff training handout call for fixing the separate condition that a team has reference documents that should ground multiple project conversations as part of a training designer's next step.
B:For a staff training handout, workspace design should respect confidentiality boundaries, not assume every collaborator can see the same sources. It would be reasonable in a case where a workspace mixes confidential client material with a broadly shared internal project, but the request for a staff training handout at a hospital education team asks for remedies to these particular issues: a project starts producing answers based on an obsolete procedural handbook; several independent workstreams need different instructions and reference sets. Neither the evidence nor the constraints in a staff training handout call for fixing the separate condition that a workspace mixes confidential client material with a broadly shared internal project as part of a training designer's next step.
D:For a staff training handout, project instructions supply reusable behavioral guidance for chats within the project instead of relying on one chat's history. It would be reasonable in a case where a recurring project keeps losing its approved writing conventions between new chats, but the request for a staff training handout at a hospital education team asks for remedies to these particular issues: a project starts producing answers based on an obsolete procedural handbook; several independent workstreams need different instructions and reference sets. Neither the evidence nor the constraints in a staff training handout call for fixing the separate condition that a recurring project keeps losing its approved writing conventions between new chats as part of a training designer's next step.
Question 341
At a cooperative bank, audit team members will rely on a control walkthrough; an internal auditor is using Claude with approved procedures. Two distinct defects require action: the team cannot explain why a more expensive model is being used for every request; additionally, response cost dominates and an acceptable error tolerance has been measured. Choose TWO actions that directly remedy the two stated defects.
- Route work by complexity and measure quality, cost and latency per class
- Favor quality on a verified test set, then preserve human validation
- Pilot the lowest-cost qualifying model on representative tasks
- Benchmark candidate models on actual latency and outcome quality
- Build a representative benchmark including edge cases and failures
Explanation
Correct Answers
A:For a workflow at a cooperative bank, tiered routing can reserve stronger models for hard cases without degrading the simpler workload. In a control walkthrough, this directly addresses one of the stated issues: the team cannot explain why a more expensive model is being used for every request. In checking a control walkthrough at a cooperative bank, the team can test the improvement addressing the team cannot explain why a more expensive model is being used for every request against approved procedures before distributing a control walkthrough to audit team members.
C:For a workflow at a cooperative bank, cost optimization is justified only after checking that the less expensive option clears the required quality bar. In a control walkthrough, this directly addresses one of the stated issues: response cost dominates and an acceptable error tolerance has been measured. In checking a control walkthrough at a cooperative bank, the team can test the improvement addressing response cost dominates and an acceptable error tolerance has been measured against approved procedures before distributing a control walkthrough to audit team members.
Incorrect Answers
B:For a control walkthrough, when consequences of errors are high, performance evidence and review matter more than raw generation price. It would be reasonable in a case where a high-stakes report has difficult reasoning and rigorous review gates, but the request for a control walkthrough at a cooperative bank asks for remedies to these particular issues: the team cannot explain why a more expensive model is being used for every request; response cost dominates and an acceptable error tolerance has been measured. Neither the evidence nor the constraints in a control walkthrough call for fixing the separate condition that a high-stakes report has difficult reasoning and rigorous review gates as part of an internal auditor's next step.
D:For a control walkthrough, a model that is accurate but misses the operational deadline can be the wrong production choice. It would be reasonable in a case where a customer interaction imposes a tight response-time target, but the request for a control walkthrough at a cooperative bank asks for remedies to these particular issues: the team cannot explain why a more expensive model is being used for every request; response cost dominates and an acceptable error tolerance has been measured. Neither the evidence nor the constraints in a control walkthrough call for fixing the separate condition that a customer interaction imposes a tight response-time target as part of an internal auditor's next step.
E:For a control walkthrough, real workload diversity reveals weaknesses that a polished single example is unlikely to expose. It would be reasonable in a case where an evaluation was based only on one easy demonstration prompt, but the request for a control walkthrough at a cooperative bank asks for remedies to these particular issues: the team cannot explain why a more expensive model is being used for every request; response cost dominates and an acceptable error tolerance has been measured. Neither the evidence nor the constraints in a control walkthrough call for fixing the separate condition that an evaluation was based only on one easy demonstration prompt as part of an internal auditor's next step.
Question 342
A marketing analyst at a consumer brand has been assigned a campaign review for the marketing director; the source material is campaign tracking records. Two distinct defects require action: a colleague wants to send AI-generated client advice without the required review; additionally, a recurring workflow has produced errors but nobody has a reporting path. Choose TWO actions that directly remedy the two stated defects.
- Record the incident and trigger the defined escalation and remediation process
- Seek policy and data-owner approval for the proposed new purpose
- Apply the documented human-approval gate before release
- Register the use case and identify a responsible business owner
- Follow the organization's AI procurement and tool-approval process
Explanation
Correct Answers
A:For a workflow at a consumer brand, incident reporting supports accountability and prevents silent repetition of known failures. In a campaign review, this directly addresses one of the stated issues: a recurring workflow has produced errors but nobody has a reporting path. In checking a campaign review at a consumer brand, the team can test the improvement addressing a recurring workflow has produced errors but nobody has a reporting path against campaign tracking records before distributing a campaign review to the marketing director.
C:For a workflow at a consumer brand, governance requirements attach to the output's consequences, not to how confident the wording sounds. In a campaign review, this directly addresses one of the stated issues: a colleague wants to send AI-generated client advice without the required review. In checking a campaign review at a consumer brand, the team can test the improvement addressing a colleague wants to send AI-generated client advice without the required review against campaign tracking records before distributing a campaign review to the marketing director.
Incorrect Answers
B:For a campaign review, consent or permission for one workflow does not automatically extend to another. It would be reasonable in a case where a pilot uses data outside the purpose originally authorized by the organization, but the request for a campaign review at a consumer brand asks for remedies to these particular issues: a colleague wants to send AI-generated client advice without the required review; a recurring workflow has produced errors but nobody has a reporting path. Neither the evidence nor the constraints in a campaign review call for fixing the separate condition that a pilot uses data outside the purpose originally authorized by the organization as part of a marketing analyst's next step.
D:For a campaign review, inventory and ownership make oversight, auditing and incident response possible. It would be reasonable in a case where a department has no record of where AI is used in a high-impact workflow, but the request for a campaign review at a consumer brand asks for remedies to these particular issues: a colleague wants to send AI-generated client advice without the required review; a recurring workflow has produced errors but nobody has a reporting path. Neither the evidence nor the constraints in a campaign review call for fixing the separate condition that a department has no record of where AI is used in a high-impact workflow as part of a marketing analyst's next step.
E:For a campaign review, an effective tool can still be impermissible if security, legal or purchasing controls are bypassed. It would be reasonable in a case where a team selects a convenient external tool without checking its approved-vendor status, but the request for a campaign review at a consumer brand asks for remedies to these particular issues: a colleague wants to send AI-generated client advice without the required review; a recurring workflow has produced errors but nobody has a reporting path. Neither the evidence nor the constraints in a campaign review call for fixing the separate condition that a team selects a convenient external tool without checking its approved-vendor status as part of a marketing analyst's next step.
Question 343
While using Claude for a partnership proposal at an environmental charity, a partnership specialist draws on approved program briefs for program partners. Two distinct defects require action: a solution brief mixes requirements, assumptions and design choices as if all were facts; additionally, several user groups have competing needs that cannot all be satisfied in the first release. Choose TWO actions that directly remedy the two stated defects.
- Separate must-have requirements from preferences and document trade-offs
- Elicit measurable outcomes, process owners and acceptance criteria
- Classify decision impact, escalation triggers and human-control points
- Produce a traceable requirements register separating confirmed needs from hypotheses
- Assess data readiness and failure cases before designing the solution
Explanation
Correct Answers
A:For a workflow at an environmental charity, prioritization makes a limited initial solution feasible and surfaces the implications of excluding a request. In a partnership proposal, this directly addresses one of the stated issues: several user groups have competing needs that cannot all be satisfied in the first release. In checking a partnership proposal at an environmental charity, the team can test the improvement addressing several user groups have competing needs that cannot all be satisfied in the first release against approved program briefs before distributing a partnership proposal to program partners.
D:For a workflow at an environmental charity, distinguishing evidence from inference makes stakeholder validation and later design changes more reliable. In a partnership proposal, this directly addresses one of the stated issues: a solution brief mixes requirements, assumptions and design choices as if all were facts. In checking a partnership proposal at an environmental charity, the team can test the improvement addressing a solution brief mixes requirements, assumptions and design choices as if all were facts against approved program briefs before distributing a partnership proposal to program partners.
Incorrect Answers
B:For a partnership proposal, requirements analysis turns a broad aspiration into testable goals and boundaries. It would be reasonable in a case where stakeholders describe a vague desire to automate reporting without defining success, but the request for a partnership proposal at an environmental charity asks for remedies to these particular issues: a solution brief mixes requirements, assumptions and design choices as if all were facts; several user groups have competing needs that cannot all be satisfied in the first release. Neither the evidence nor the constraints in a partnership proposal call for fixing the separate condition that stakeholders describe a vague desire to automate reporting without defining success as part of a partnership specialist's next step.
C:For a partnership proposal, risk-aware scoping identifies which stages can be assisted and which require accountable oversight. It would be reasonable in a case where the team has not established where errors would cause unacceptable harm, but the request for a partnership proposal at an environmental charity asks for remedies to these particular issues: a solution brief mixes requirements, assumptions and design choices as if all were facts; several user groups have competing needs that cannot all be satisfied in the first release. Neither the evidence nor the constraints in a partnership proposal call for fixing the separate condition that the team has not established where errors would cause unacceptable harm as part of a partnership specialist's next step.
E:For a partnership proposal, a model cannot compensate for missing authoritative inputs when outcomes depend on those inputs. It would be reasonable in a case where a proposed AI step appears useful but the available source data is incomplete, but the request for a partnership proposal at an environmental charity asks for remedies to these particular issues: a solution brief mixes requirements, assumptions and design choices as if all were facts; several user groups have competing needs that cannot all be satisfied in the first release. Neither the evidence nor the constraints in a partnership proposal call for fixing the separate condition that a proposed AI step appears useful but the available source data is incomplete as part of a partnership specialist's next step.
Question 344
A logistics planner at an import business has been assigned a shipment risk review for operations directors; the source material is carrier status files. Two distinct defects require action: a draft statement could create a binding contractual promise if sent to a client; additionally, a clinical education handout includes specific dosing and contraindication claims. Choose TWO actions that directly remedy the two stated defects.
- Require qualified clinical validation against approved references
- Apply a proportionate lightweight review instead of a formal approval chain
- Require an authorized reviewer to make the decision after checking evidence
- Route it through the privacy and communications review before distribution
- Send it to the designated legal or commercial approver before release
Explanation
Correct Answers
A:For a workflow at an import business, health-related factual errors can cause harm; stylistic editing is not a substitute for professional verification. In a shipment risk review, this directly addresses one of the stated issues: a clinical education handout includes specific dosing and contraindication claims. In checking a shipment risk review at an import business, the team can test the improvement addressing a clinical education handout includes specific dosing and contraindication claims against carrier status files before distributing a shipment risk review to operations directors.
E:For a workflow at an import business, language models cannot authorize commitments on behalf of the organization, even when the draft sounds routine. In a shipment risk review, this directly addresses one of the stated issues: a draft statement could create a binding contractual promise if sent to a client. In checking a shipment risk review at an import business, the team can test the improvement addressing a draft statement could create a binding contractual promise if sent to a client against carrier status files before distributing a shipment risk review to operations directors.
Incorrect Answers
B:For a shipment risk review, review effort should track the consequences of error, not treat low-stakes ideation as a final regulated decision. It would be reasonable in a case where a noncritical internal brainstorming note contains no sensitive data and is clearly labeled as ideas, but the request for a shipment risk review at an import business asks for remedies to these particular issues: a draft statement could create a binding contractual promise if sent to a client; a clinical education handout includes specific dosing and contraindication claims. Neither the evidence nor the constraints in a shipment risk review call for fixing the separate condition that a noncritical internal brainstorming note contains no sensitive data and is clearly labeled as ideas as part of a logistics planner's next step.
C:For a shipment risk review, high-impact determinations need accountable human judgment and applicable procedural safeguards. It would be reasonable in a case where a generated recommendation would directly determine an individual's eligibility for a regulated benefit, but the request for a shipment risk review at an import business asks for remedies to these particular issues: a draft statement could create a binding contractual promise if sent to a client; a clinical education handout includes specific dosing and contraindication claims. Neither the evidence nor the constraints in a shipment risk review call for fixing the separate condition that a generated recommendation would directly determine an individual's eligibility for a regulated benefit as part of a logistics planner's next step.
D:For a shipment risk review, publication creates exposure that cannot be undone simply by revising the document later. It would be reasonable in a case where a draft public notice combines operational data with potentially identifying personal details, but the request for a shipment risk review at an import business asks for remedies to these particular issues: a draft statement could create a binding contractual promise if sent to a client; a clinical education handout includes specific dosing and contraindication claims. Neither the evidence nor the constraints in a shipment risk review call for fixing the separate condition that a draft public notice combines operational data with potentially identifying personal details as part of a logistics planner's next step.
Question 345
Using maintained help-center articles, a knowledge manager at a service organization is drafting an internal FAQ that front-line colleagues will use. Two distinct defects require action: the figures are correct but the reporting period in the conclusion is wrong; additionally, a report cites the right file but one numeric total differs from the source table. Choose TWO actions that directly remedy the two stated defects.
- Trace dates and scope through source, calculation and narrative
- Compare each recommendation with explicit constraints and resolve the conflict
- Recompute the total from underlying records and correct the discrepancy
- Evaluate both supporting evidence and consequential omissions before approval
- Check each requirement against a coverage checklist and fill the omission
Explanation
Correct Answers
A:For a workflow at a service organization, a correct statistic attached to the wrong period can mislead readers even when the arithmetic itself is sound. In an internal FAQ, this directly addresses one of the stated issues: the figures are correct but the reporting period in the conclusion is wrong. In checking an internal FAQ at a service organization, the team can test the improvement addressing the figures are correct but the reporting period in the conclusion is wrong against maintained help-center articles before distributing an internal FAQ to front-line colleagues.
C:For a workflow at a service organization, arithmetic accuracy must be checked against source values, not against the fluency of the explanation. In an internal FAQ, this directly addresses one of the stated issues: a report cites the right file but one numeric total differs from the source table. In checking an internal FAQ at a service organization, the team can test the improvement addressing a report cites the right file but one numeric total differs from the source table against maintained help-center articles before distributing an internal FAQ to front-line colleagues.
Incorrect Answers
B:For an internal FAQ, internal consistency requires every proposed action to respect the same operating assumptions. It would be reasonable in a case where the output is logically plausible but contradicts a boundary stated earlier in the brief, but the request for an internal FAQ at a service organization asks for remedies to these particular issues: the figures are correct but the reporting period in the conclusion is wrong; a report cites the right file but one numeric total differs from the source table. Neither the evidence nor the constraints in an internal FAQ call for fixing the separate condition that the output is logically plausible but contradicts a boundary stated earlier in the brief as part of a knowledge manager's next step.
D:For an internal FAQ, a balanced completeness test checks whether decision-relevant disadvantages and limitations are represented. It would be reasonable in a case where a short summary accurately lists benefits but omits material implementation risks, but the request for an internal FAQ at a service organization asks for remedies to these particular issues: the figures are correct but the reporting period in the conclusion is wrong; a report cites the right file but one numeric total differs from the source table. Neither the evidence nor the constraints in an internal FAQ call for fixing the separate condition that a short summary accurately lists benefits but omits material implementation risks as part of a knowledge manager's next step.
E:For an internal FAQ, completeness is measured against the original brief, including exclusions and exceptions, not only the topics discussed. It would be reasonable in a case where the draft answers most questions yet silently drops a mandatory acceptance criterion, but the request for an internal FAQ at a service organization asks for remedies to these particular issues: the figures are correct but the reporting period in the conclusion is wrong; a report cites the right file but one numeric total differs from the source table. Neither the evidence nor the constraints in an internal FAQ call for fixing the separate condition that the draft answers most questions yet silently drops a mandatory acceptance criterion as part of a knowledge manager's next step.
Question 346
At a publishing house, section editors will rely on an editorial brief; a content editor is using Claude with author submissions. Two distinct defects require action: the drafted workflow would publish AI text without an assigned accountable editor; additionally, an assistant repeats work that an existing approved system already performs reliably. Choose TWO actions that directly remedy the two stated defects.
- Limit Claude to the judgment-heavy or language-intensive step and keep deterministic rules in existing tools
- Map the actual process and place Claude at a defined input/output handoff
- Add a named review and approval step before distribution
- Document fallback procedures and clear ownership of manual continuation
- Define the interface schema, validations and exception handling
Explanation
Correct Answers
A:For a workflow at a publishing house, complementing reliable systems reduces unnecessary complexity and avoidable error. In an editorial brief, this directly addresses one of the stated issues: an assistant repeats work that an existing approved system already performs reliably. In checking an editorial brief at a publishing house, the team can test the improvement addressing an assistant repeats work that an existing approved system already performs reliably against author submissions before distributing an editorial brief to section editors.
C:For a workflow at a publishing house, operational integration must preserve responsibility for final business communications. In an editorial brief, this directly addresses one of the stated issues: the drafted workflow would publish AI text without an assigned accountable editor. In checking an editorial brief at a publishing house, the team can test the improvement addressing the drafted workflow would publish AI text without an assigned accountable editor against author submissions before distributing an editorial brief to section editors.
Incorrect Answers
B:For an editorial brief, workflow fit matters more than merely adding an assistant to an unrelated step. It would be reasonable in a case where AI assistance is proposed without understanding how staff currently receive and complete work, but the request for an editorial brief at a publishing house asks for remedies to these particular issues: the drafted workflow would publish AI text without an assigned accountable editor; an assistant repeats work that an existing approved system already performs reliably. Neither the evidence nor the constraints in an editorial brief call for fixing the separate condition that AI assistance is proposed without understanding how staff currently receive and complete work as part of a content editor's next step.
D:For an editorial brief, a usable process remains functional when an optional AI component fails or cannot be accessed. It would be reasonable in a case where a small pilot works but staff do not know what to do when Claude is unavailable, but the request for an editorial brief at a publishing house asks for remedies to these particular issues: the drafted workflow would publish AI text without an assigned accountable editor; an assistant repeats work that an existing approved system already performs reliably. Neither the evidence nor the constraints in an editorial brief call for fixing the separate condition that a small pilot works but staff do not know what to do when Claude is unavailable as part of a content editor's next step.
E:For an editorial brief, reliable handoffs require structured data and rejection of malformed records, not just a fluent reply. It would be reasonable in a case where outputs would feed a business system that expects a consistent file layout, but the request for an editorial brief at a publishing house asks for remedies to these particular issues: the drafted workflow would publish AI text without an assigned accountable editor; an assistant repeats work that an existing approved system already performs reliably. Neither the evidence nor the constraints in an editorial brief call for fixing the separate condition that outputs would feed a business system that expects a consistent file layout as part of a content editor's next step.
Question 347
At a county authority, a transport planner is preparing a route-change briefing for decision makers using published ridership summaries. Two distinct defects require action: an improvement raises average quality but introduces severe errors on rare cases; additionally, a pilot shows that the assistant is fast but frequently confuses two categories. Choose TWO actions that directly remedy the two stated defects.
- Add exception coverage to the rubric and test the updated prompt on missed examples
- Collect misclassified cases and revise category definitions with boundary examples
- Run regression checks on a fixed representative sample and compare with baseline
- Revise tone only while freezing factual content and traceability
- Add edge-case tests and escalation triggers before expanding use
Explanation
Correct Answers
B:For a workflow at a county authority, error analysis targets the confused distinction instead of simply choosing a larger model by default. In a route-change briefing, this directly addresses one of the stated issues: a pilot shows that the assistant is fast but frequently confuses two categories. In checking a route-change briefing at a county authority, the team can test the improvement addressing a pilot shows that the assistant is fast but frequently confuses two categories against published ridership summaries before distributing a route-change briefing to decision makers.
E:For a workflow at a county authority, average gains can conceal unacceptable tail risk in consequential situations. In a route-change briefing, this directly addresses one of the stated issues: an improvement raises average quality but introduces severe errors on rare cases. In checking a route-change briefing at a county authority, the team can test the improvement addressing an improvement raises average quality but introduces severe errors on rare cases against published ridership summaries before distributing a route-change briefing to decision makers.
Incorrect Answers
A:For a route-change briefing, feedback is useful when converted into measurable acceptance tests rather than vague preferences. It would be reasonable in a case where users report that summaries miss exceptions that affect their work, but the request for a route-change briefing at a county authority asks for remedies to these particular issues: an improvement raises average quality but introduces severe errors on rare cases; a pilot shows that the assistant is fast but frequently confuses two categories. Neither the evidence nor the constraints in a route-change briefing call for fixing the separate condition that users report that summaries miss exceptions that affect their work as part of a transport planner's next step.
C:For a route-change briefing, controlled comparison shows which outcomes genuinely improved or degraded after a change. It would be reasonable in a case where a model upgrade changes output behavior in an existing workflow, but the request for a route-change briefing at a county authority asks for remedies to these particular issues: an improvement raises average quality but introduces severe errors on rare cases; a pilot shows that the assistant is fast but frequently confuses two categories. Neither the evidence nor the constraints in a route-change briefing call for fixing the separate condition that a model upgrade changes output behavior in an existing workflow as part of a transport planner's next step.
D:For a route-change briefing, a localized edit preserves already accepted material and minimizes regression risk. It would be reasonable in a case where a reviewer rejects a draft for tone but verifies the facts are correct, but the request for a route-change briefing at a county authority asks for remedies to these particular issues: an improvement raises average quality but introduces severe errors on rare cases; a pilot shows that the assistant is fast but frequently confuses two categories. Neither the evidence nor the constraints in a route-change briefing call for fixing the separate condition that a reviewer rejects a draft for tone but verifies the facts are correct as part of a transport planner's next step.
Question 348
A workstream at a conference organizer asks an event coordinator to create a post-event analysis from attendance and survey summaries for sponsors. Two distinct defects require action: a task refers to a file that has not actually been attached or made available; additionally, a user needs a current document kept in an authorized organizational drive. Choose TWO actions that directly remedy the two stated defects.
- Identify the authoritative edition and remove or archive superseded material
- Use an approved connector to retrieve the allowed source, after checking permissions
- Attach or retrieve the file before asking Claude to make source-specific claims
- Remove unnecessary fields or use an approved sanitized dataset
- Restrict access to the least necessary sources and confirm granted permissions
Explanation
Correct Answers
B:For a workflow at a conference organizer, connected retrieval can reduce stale copy-pasting while remaining bounded by authorization and tool access. In a post-event analysis, this directly addresses one of the stated issues: a user needs a current document kept in an authorized organizational drive. In checking a post-event analysis at a conference organizer, the team can test the improvement addressing a user needs a current document kept in an authorized organizational drive against attendance and survey summaries before distributing a post-event analysis to sponsors.
C:For a workflow at a conference organizer, naming a document in a prompt does not grant access to its contents. In a post-event analysis, this directly addresses one of the stated issues: a task refers to a file that has not actually been attached or made available. In checking a post-event analysis at a conference organizer, the team can test the improvement addressing a task refers to a file that has not actually been attached or made available against attendance and survey summaries before distributing a post-event analysis to sponsors.
Incorrect Answers
A:For a post-event analysis, provenance and lifecycle control prevent an old source from winning simply because it is easier to find. It would be reasonable in a case where an uploaded reference lacks a version or effective date and conflicts with another file, but the request for a post-event analysis at a conference organizer asks for remedies to these particular issues: a task refers to a file that has not actually been attached or made available; a user needs a current document kept in an authorized organizational drive. Neither the evidence nor the constraints in a post-event analysis call for fixing the separate condition that an uploaded reference lacks a version or effective date and conflicts with another file as part of an event coordinator's next step.
D:For a post-event analysis, data minimization reduces exposure while preserving the aggregate information needed for analysis. It would be reasonable in a case where a spreadsheet upload contains private identifiers unrelated to the intended summary, but the request for a post-event analysis at a conference organizer asks for remedies to these particular issues: a task refers to a file that has not actually been attached or made available; a user needs a current document kept in an authorized organizational drive. Neither the evidence nor the constraints in a post-event analysis call for fixing the separate condition that a spreadsheet upload contains private identifiers unrelated to the intended summary as part of an event coordinator's next step.
E:For a post-event analysis, connector convenience does not override confidentiality or least-privilege principles. It would be reasonable in a case where a connector can read more customer records than the assistant's task requires, but the request for a post-event analysis at a conference organizer asks for remedies to these particular issues: a task refers to a file that has not actually been attached or made available; a user needs a current document kept in an authorized organizational drive. Neither the evidence nor the constraints in a post-event analysis call for fixing the separate condition that a connector can read more customer records than the assistant's task requires as part of an event coordinator's next step.
Question 349
During a review of a board packet, an administrator at a health foundation is working with signed committee minutes for trustees. Two distinct defects require action: a one-off user request is being hard-coded into instructions shared by unrelated teams; additionally, a configuration contains conflicting commands about output length and required sections. Choose TWO actions that directly remedy the two stated defects.
- Keep task-specific directions in the current prompt rather than broad persistent settings
- Resolve the priority conflict and define an explicit exception rule
- Remove secrets and rely on sanctioned access mechanisms
- State stable terminology and style rules in its persistent instructions
- Define the evidence and uncertainty policy in persistent guidance
Explanation
Correct Answers
A:For a workflow at a health foundation, separating enduring rules from local instructions avoids unnecessary effects on other work. In a board packet, this directly addresses one of the stated issues: a one-off user request is being hard-coded into instructions shared by unrelated teams. In checking a board packet at a health foundation, the team can test the improvement addressing a one-off user request is being hard-coded into instructions shared by unrelated teams against signed committee minutes before distributing a board packet to trustees.
B:For a workflow at a health foundation, contradictory directions invite inconsistent behavior and make evaluation harder. In a board packet, this directly addresses one of the stated issues: a configuration contains conflicting commands about output length and required sections. In checking a board packet at a health foundation, the team can test the improvement addressing a configuration contains conflicting commands about output length and required sections against signed committee minutes before distributing a board packet to trustees.
Incorrect Answers
C:For a board packet, instructions can be exposed in workflow contexts and are not a secure substitute for credential management. It would be reasonable in a case where a team embeds confidential credentials directly in a reusable instruction block, but the request for a board packet at a health foundation asks for remedies to these particular issues: a one-off user request is being hard-coded into instructions shared by unrelated teams; a configuration contains conflicting commands about output length and required sections. Neither the evidence nor the constraints in a board packet call for fixing the separate condition that a team embeds confidential credentials directly in a reusable instruction block as part of an administrator's next step.
D:For a board packet, reusable guidance is appropriate for consistent behavior across multiple tasks, not for one transient fact. It would be reasonable in a case where a shared assistant repeatedly changes approved terminology across otherwise similar outputs, but the request for a board packet at a health foundation asks for remedies to these particular issues: a one-off user request is being hard-coded into instructions shared by unrelated teams; a configuration contains conflicting commands about output length and required sections. Neither the evidence nor the constraints in a board packet call for fixing the separate condition that a shared assistant repeatedly changes approved terminology across otherwise similar outputs as part of an administrator's next step.
E:For a board packet, stable safeguards can encourage verifiable responses across tasks but still require human checking. It would be reasonable in a case where a recurring process requires Claude to disclose uncertainty and avoid invented sources, but the request for a board packet at a health foundation asks for remedies to these particular issues: a one-off user request is being hard-coded into instructions shared by unrelated teams; a configuration contains conflicting commands about output length and required sections. Neither the evidence nor the constraints in a board packet call for fixing the separate condition that a recurring process requires Claude to disclose uncertainty and avoid invented sources as part of an administrator's next step.
Question 350
For funding partners, a program officer at a nonprofit must deliver a grant update grounded in program delivery reports. Two distinct defects require action: a customer-facing draft needs a specific voice and action, not a technical evaluation; additionally, a research request needs a factual answer with support rather than confident speculation. Choose TWO actions that directly remedy the two stated defects.
- Ask for sourced findings, dates, uncertainty and unresolved questions
- Request multiple divergent options before assessing them against criteria
- Specify audience, channel, tone, call to action and prohibited claims
- Ask for a recommendation that distinguishes facts, assumptions, alternatives and risks
- Request a comparison against explicit evaluation criteria and trade-offs
Explanation
Correct Answers
A:For a workflow at a nonprofit, research prompting separates verified observations from inference and makes important claims independently checkable. In a grant update, this directly addresses one of the stated issues: a research request needs a factual answer with support rather than confident speculation. In checking a grant update at a nonprofit, the team can test the improvement addressing a research request needs a factual answer with support rather than confident speculation against program delivery reports before distributing a grant update to funding partners.
C:For a workflow at a nonprofit, drafting prompts constrain persuasive language and prevent the assistant from inventing commitments. In a grant update, this directly addresses one of the stated issues: a customer-facing draft needs a specific voice and action, not a technical evaluation. In checking a grant update at a nonprofit, the team can test the improvement addressing a customer-facing draft needs a specific voice and action, not a technical evaluation against program delivery reports before distributing a grant update to funding partners.
Incorrect Answers
B:For a grant update, separating idea generation from selection increases variety without treating every proposal as equally practical. It would be reasonable in a case where a brainstorming session is being constrained to one obvious recommendation too early, but the request for a grant update at a nonprofit asks for remedies to these particular issues: a customer-facing draft needs a specific voice and action, not a technical evaluation; a research request needs a factual answer with support rather than confident speculation. Neither the evidence nor the constraints in a grant update call for fixing the separate condition that a brainstorming session is being constrained to one obvious recommendation too early as part of a program officer's next step.
D:For a grant update, decision-support structure exposes the basis for a judgment instead of hiding uncertainty inside fluent prose. It would be reasonable in a case where the team must make a defensible decision from conflicting information, but the request for a grant update at a nonprofit asks for remedies to these particular issues: a customer-facing draft needs a specific voice and action, not a technical evaluation; a research request needs a factual answer with support rather than confident speculation. Neither the evidence nor the constraints in a grant update call for fixing the separate condition that the team must make a defensible decision from conflicting information as part of a program officer's next step.
E:For a grant update, comparative analysis weighs strengths and limitations on common dimensions before choosing an approach. It would be reasonable in a case where the team wants an analysis of competing proposals rather than another summary, but the request for a grant update at a nonprofit asks for remedies to these particular issues: a customer-facing draft needs a specific voice and action, not a technical evaluation; a research request needs a factual answer with support rather than confident speculation. Neither the evidence nor the constraints in a grant update call for fixing the separate condition that the team wants an analysis of competing proposals rather than another summary as part of a program officer's next step.