100% Latest & Updated Microsoft AB-100 Practice Test Questions, Exam Dumps & Verified Answers!
30 Days Free Updates, Instant Download!
Question 130
A manager needs a weekly task that retrieves tickets, groups them, and posts an approved summary on schedule. Which specific remediation should the business adopt?
- Use a stable idempotency key on the transaction action.
- Combine scheduled orchestration with scoped retrieval and an authorized reporting action.
- Enable traceable agent execution spans and correlate model and tool latency.
- Define required task inputs and ask for missing values before invoking the action.
Explanation
Correct Answer
B:An orchestrated task has deterministic scheduling plus governed data access and output delivery.
Incorrect Answers
A:A retry-safe business transaction prevents duplicate side effects after network uncertainty, whereas an orchestrated task has deterministic scheduling plus governed data access and output delivery.
C:Execution traces expose where latency or errors occur in a multi-step agent run, but an orchestrated task has deterministic scheduling plus governed data access and output delivery.
D:A complete input contract prevents execution with invented or empty business fields. An orchestrated task has deterministic scheduling plus governed data access and output delivery.
Question 131
A purchasing agent must ask for a cost center before it can create a valid purchase request. Which solution approach best meets the stated requirement?
- Create a narrowly scoped prompt action with defined input and output.
- Design a bounded read-and-summarize task agent.
- Define required task inputs and ask for missing values before invoking the action.
- Use a stable idempotency key on the transaction action.
Explanation
Correct Answer
C:A complete input contract prevents execution with invented or empty business fields.
Incorrect Answers
A:A prompt action fits a bounded generative text transformation within a larger workflow. A complete input contract prevents execution with invented or empty business fields.
B:A limited task agent can collect evidence and generate a reviewable output without write privileges, but a complete input contract prevents execution with invented or empty business fields.
D:A retry-safe business transaction prevents duplicate side effects after network uncertainty; a complete input contract prevents execution with invented or empty business fields.
Question 132
An expense-processing agent must retry a timeout without creating two reimbursement requests. Which option most directly closes the documented gap?
- Design a bounded read-and-summarize task agent.
- Combine scheduled orchestration with scoped retrieval and an authorized reporting action.
- Add bounded retries and idempotent execution to the action flow.
- Use a stable idempotency key on the transaction action.
Explanation
Correct Answer
D:A retry-safe business transaction prevents duplicate side effects after network uncertainty.
Incorrect Answers
A:A limited task agent can collect evidence and generate a reviewable output without write privileges, but a retry-safe business transaction prevents duplicate side effects after network uncertainty.
B:An orchestrated task has deterministic scheduling plus governed data access and output delivery. A retry-safe business transaction prevents duplicate side effects after network uncertainty.
C:Retry limits and duplicate protection make deterministic automation safe under temporary failures; a retry-safe business transaction prevents duplicate side effects after network uncertainty.
Question 133
An inventory agent may place orders without approval only below a defined spend ceiling and for allowed vendors. Which concrete solution decision should the team make?
- Use a deterministic agent flow with explicit approval steps.
- Bound autonomy by enforceable spend and vendor policies.
- Separate autonomous case detection from consent-gated outbound communication.
- Keep final eligibility decisions with authorized humans and capture reasoning evidence.
Explanation
Correct Answer
B:Tool-level controls restrict autonomous actions to explicitly approved business transactions.
Incorrect Answers
A:Fixed workflow logic supports repeatable authorization and an auditable process sequence. Tool-level controls restrict autonomous actions to explicitly approved business transactions.
C:Proactive sensing does not grant automatic authority to contact external parties, whereas tool-level controls restrict autonomous actions to explicitly approved business transactions.
D:High-impact decisions need reviewable human accountability instead of silent autonomous denial. Tool-level controls restrict autonomous actions to explicitly approved business transactions.
Question 134
An agent evaluates customer eligibility where a false rejection can create legal exposure. Which design decision should the architect prioritize?
- Model durable task state, timeouts and a safe resume policy.
- Separate autonomous case detection from consent-gated outbound communication.
- Keep final eligibility decisions with authorized humans and capture reasoning evidence.
- Specify an approved at-risk definition with measurable input attributes.
Explanation
Correct Answer
C:High-impact decisions need reviewable human accountability instead of silent autonomous denial.
Incorrect Answers
A:Long-running autonomy requires inspectable lifecycle control and recovery from partial progress; high-impact decisions need reviewable human accountability instead of silent autonomous denial.
B:Proactive sensing does not grant automatic authority to contact external parties, whereas high-impact decisions need reviewable human accountability instead of silent autonomous denial.
D:An explicit business rule allows staff to reproduce and challenge customer risk classification, but high-impact decisions need reviewable human accountability instead of silent autonomous denial.
Question 135
A proactive service agent detects a case escalation trigger but must not contact external customers without consent. Which controlled change is most appropriate for this situation?
- Model durable task state, timeouts and a safe resume policy.
- Customize the supported service copilot instructions and escalation boundaries.
- Bound autonomy by enforceable spend and vendor policies.
- Separate autonomous case detection from consent-gated outbound communication.
Explanation
Correct Answer
D:Proactive sensing does not grant automatic authority to contact external parties.
Incorrect Answers
A:Long-running autonomy requires inspectable lifecycle control and recovery from partial progress, whereas proactive sensing does not grant automatic authority to contact external parties.
B:Constrained response and escalation behavior adapts the copilot to governed service policy. Proactive sensing does not grant automatic authority to contact external parties.
C:Tool-level controls restrict autonomous actions to explicitly approved business transactions, but proactive sensing does not grant automatic authority to contact external parties.
Question 136
A background agent runs for hours and may need to pause when a dependency or approver is unavailable. Which design choice targets the observed problem?
- Model durable task state, timeouts and a safe resume policy.
- Design a bounded read-and-summarize task agent.
- Bound autonomy by enforceable spend and vendor policies.
- Keep final eligibility decisions with authorized humans and capture reasoning evidence.
Explanation
Correct Answer
A:Long-running autonomy requires inspectable lifecycle control and recovery from partial progress.
Incorrect Answers
B:A limited task agent can collect evidence and generate a reviewable output without write privileges, but long-running autonomy requires inspectable lifecycle control and recovery from partial progress.
C:Tool-level controls restrict autonomous actions to explicitly approved business transactions. Long-running autonomy requires inspectable lifecycle control and recovery from partial progress.
D:High-impact decisions need reviewable human accountability instead of silent autonomous denial; long-running autonomy requires inspectable lifecycle control and recovery from partial progress.
Question 137
A claims assistant turns one policy paragraph into a proposed explanation and must not change records. Which architecture response most directly addresses the risk?
- Include uncertainty detection and human fallback for conflicting sources.
- Define structured response fields and validate the generated output.
- Use a read-only prompt-and-response pattern grounded in the relevant policy text.
- Consolidate the response path so only one component emits the final answer.
Explanation
Correct Answer
C:A focused generation task produces an answer without granting mutation tools.
Incorrect Answers
A:Escalation avoids unsupported claims when authoritative inputs disagree, but a focused generation task produces an answer without granting mutation tools.
B:A validated schema makes the response dependable for subsequent workflow steps. A focused generation task produces an answer without granting mutation tools.
D:Avoiding overlapping answer generation prevents duplicate messages under generative orchestration, whereas a focused generation task produces an answer without granting mutation tools.
Question 138
A customer reply must contain case number, action summary and next step in a predictable machine-readable form. Which recommendation addresses the identified failure most directly?
- Use concise task-specific response guidance with essential support facts.
- Include uncertainty detection and human fallback for conflicting sources.
- Combine a bounded language step with a validated case-creation action.
- Define structured response fields and validate the generated output.
Explanation
Correct Answer
D:A validated schema makes the response dependable for subsequent workflow steps.
Incorrect Answers
A:Output bounds improve relevance and consumption while preserving necessary accuracy. A validated schema makes the response dependable for subsequent workflow steps.
B:Escalation avoids unsupported claims when authoritative inputs disagree; a validated schema makes the response dependable for subsequent workflow steps.
C:A hybrid flow uses AI only for unstructured interpretation, then validates the transactional result, whereas a validated schema makes the response dependable for subsequent workflow steps.
Question 139
An agent should stop and seek an employee when source evidence conflicts rather than choose a confident-looking answer. What should the solution architect recommend for this specific need?
- Include uncertainty detection and human fallback for conflicting sources.
- Use concise task-specific response guidance with essential support facts.
- Consolidate the response path so only one component emits the final answer.
- Use a read-only prompt-and-response pattern grounded in the relevant policy text.
Explanation
Correct Answer
A:Escalation avoids unsupported claims when authoritative inputs disagree.
Incorrect Answers
B:Output bounds improve relevance and consumption while preserving necessary accuracy, whereas escalation avoids unsupported claims when authoritative inputs disagree.
C:Avoiding overlapping answer generation prevents duplicate messages under generative orchestration. Escalation avoids unsupported claims when authoritative inputs disagree.
D:A focused generation task produces an answer without granting mutation tools, but escalation avoids unsupported claims when authoritative inputs disagree.
Question 140
An assistant generates lengthy explanations for simple password-reset eligibility, increasing cost and user confusion. What is the most defensible intervention for this case?
- Use a read-only prompt-and-response pattern grounded in the relevant policy text.
- Use concise task-specific response guidance with essential support facts.
- Improve tool descriptions and required input contracts before adding more tools.
- Define structured response fields and validate the generated output.
Explanation
Correct Answer
B:Output bounds improve relevance and consumption while preserving necessary accuracy.
Incorrect Answers
A:A focused generation task produces an answer without granting mutation tools, whereas output bounds improve relevance and consumption while preserving necessary accuracy.
C:Clear tool semantics improve selection and reduce orchestration confusion, but output bounds improve relevance and consumption while preserving necessary accuracy.
D:A validated schema makes the response dependable for subsequent workflow steps. Output bounds improve relevance and consumption while preserving necessary accuracy.
Agentic AI Business Solutions Architect (AB-100) Practice Tests