Microsoft AB-100 Practice Test 47

Question 460

A HR assistant currently drafts onboarding checklists. Management wants it to create real payroll records automatically, but the employee-ID system and approvals are not integrated. What is the appropriate expansion decision?

Treat a successful checklist generation as proof that the payroll database write path is safe.
Retain assistive drafting until typed employee identity, payroll authorization and verified commit workflows are implemented and tested.
Let the model infer payroll IDs from employee names and issue records in the background.
Add a prompt telling the assistant to be cautious, then immediately grant payroll administrator privileges.

Explanation

Correct Answer

Execution autonomy requires transaction-level controls that a checklist-only assistant never needed.

Incorrect Answers

Draft generation and durable payroll record creation are different capabilities with different failure and control risks.

Names may collide; inferred identifiers cannot establish the identity required for salary transactions.

Natural-language caution is not a credential, scope or approval enforcement layer.

Question 461

A support triage agent transfers a refund dispute to a billing agent. The triage agent keeps issuing decisions after the handoff, causing duplicate customer messages. What contract should be enforced?

Create a fresh case ID whenever an agent receives the same handoff.
Delete the ownership field to reduce the handoff payload size.
A typed handoff that transfers case ownership and records a terminal state for the originating agent.
Allow both agents to continue independently and merge their replies using whichever sounds more helpful.

Explanation

Correct Answer

Exclusive case ownership avoids two agents issuing contradictory or redundant actions for one customer request.

Incorrect Answers

Unnecessary new IDs break state reconciliation and make retries appear as separate incidents.

Removing routing responsibility eliminates the deterministic evidence needed to know which agent should act next.

Textual arbitration does not prevent duplicate downstream actions or conflicting commitments.

Question 462

A planner sends “ship 20” to a warehouse agent. One warehouse interprets it as pallets and another as cartons. What should the inter-agent protocol carry?

Only the free-text number 20 and a confidence score from the planning model.
A warehouse-wide default unit selected by whichever agent receives the request first.
A copy of the full chat history as the sole source of shipment quantity.
A validated quantity, explicit unit of measure, product identifier and correlation ID for the intended shipment.

Explanation

Correct Answer

Typed business fields remove ambiguous units and support safe deduplication during execution and retries.

Incorrect Answers

Confidence does not disambiguate physical units or identify which product should be shipped.

Different sites may use different handling units; implicit defaults can cause shipment-size errors.

Unstructured history may contain several obsolete numbers and lacks an executable typed unit contract.

Question 463

A travel planner agent reserves a hotel, then a transport agent fails to reserve a flight. The customer requested an all-or-nothing package. What recovery design is needed?

Track a shared workflow state and perform an authorized hotel cancellation or explicit exception review after the flight failure.
Declare the package complete because the hotel booking succeeded.
Retry the hotel reservation repeatedly while ignoring the flight error.
Delete the failed flight request trace so the hotel confirmation looks consistent.

Explanation

Correct Answer

Coordinated compensation prevents an orphan hotel charge when a multi-service transaction cannot complete.

Incorrect Answers

One successful component cannot satisfy the customer’s all-or-nothing itinerary requirement.

Uncoordinated retries can duplicate charges and do not resolve the missing transport component.

Removing the failure record undermines recovery and hides the reason for incomplete fulfilment.

Question 464

A finance assessment agent asks a shipping agent to release goods on credit hold. The shipping agent has a privileged tool but the initiating employee is not authorized. How should the delegation be handled?

Make the finance agent rewrite the user role in its tool arguments to match the shipping administrator.
Evaluate the requesting principal and the intended release action at the shipping execution boundary before proceeding.
Let the shipping agent act because its own connector has the required administrator permission.
Rely on the finance agent’s statement that the employee was already checked.

Explanation

Correct Answer

The receiving agent’s service identity cannot legitimately elevate the initiating employee’s authorization.

Incorrect Answers

Role impersonation evades the business security boundary and corrupts the audit record.

Connector capability does not prove the particular caller is allowed to release credit-held stock.

A handoff message is task data, not an independently verified authorization decision.

Question 465

An insurer runs a document-verification agent and a sanctions-screening agent in parallel before issuing a policy. Neither result may be skipped, and an outage can occur in either service. Which TWO controls are essential? (Select TWO.)

Retry failed checks using a new application ID so parallel outcomes are easier to count.
Allow either agent to overwrite the other’s verification result when its model confidence is higher.
Correlate both agent results to the same policy application and wait for the required validated statuses.
Keep policy issuance blocked or escalate when either mandatory screening result is unavailable.
Accept the first successful agent response and immediately treat the entire policy review as complete.

Explanation

Correct Answers

A shared transaction identity ensures evidence from different applications is not accidentally combined or omitted.

An incomplete mandatory check is not an implied pass merely because the other parallel task succeeded.

Incorrect Answers

Changing identity across retries prevents reliable correlation and can produce inconsistent issuance decisions.

Separate legal screening outcomes cannot be substituted using a cross-task confidence comparison.

The two screenings are independently required and cannot be replaced by whichever finishes first.

Question 466

A regional service team needs existing case summaries in four languages. Its Dynamics 365 agent supports the same records and languages, with approved user permissions. What should the architect do before commissioning a custom multilingual model?

Assume a bespoke model is necessary solely because the process spans four languages.
Disable existing source permissions so all language versions have equal access to every case.
Build a separate portal before testing whether the current agent fits the existing service workflow.
Configure and test the supported prebuilt capability with region-specific acceptance cases and escalation requirements.

Explanation

Correct Answer

A supported baseline can meet the need with less lifecycle burden if actual language and policy tests pass.

Incorrect Answers

Language count by itself does not show a capability gap in the existing platform.

Consistent translation does not require unrestricted cross-customer record visibility.

A duplicate interface increases maintenance without demonstrating an unmet process or permission requirement.

Question 467

A prebuilt returns agent meets normal request handling, but the company must retain each approval signature under a legal hold. The prebuilt workflow cannot expose or export the required signature evidence. What is the appropriate decision?

Document the retention gap and evaluate a supported extension or custom controlled action before deployment.
Use the prebuilt flow as is and assume its conversation summary proves a signed retention record.
Remove the retention requirement from the use case because prebuilt agents are standard software.
Grant global administrator access to the agent to make the absent audit export feature appear.

Explanation

Correct Answer

A missing mandatory audit artifact is a genuine capability gap that must be addressed before approving the prebuilt flow.

Incorrect Answers

Conversational text is not the authenticated signature or legal-hold evidence the policy explicitly requires.

Product support status does not waive the organization’s legal recordkeeping obligations.

Broader permissions cannot create a missing evidence capability and would introduce additional security exposure.

Question 468

A contact-center product offers case classification and escalation, but the company needs one additional reviewed classification field. A bespoke replacement would rebuild all routing. What should be compared?

Discard the extra field from the business requirements before measuring its compliance impact.
A supported configuration or extension for the field against the total cost and risk of replacing the agent.
Replace the whole service because any extra field makes prebuilt software unsuitable.
Store the field only in the system prompt and treat that as persistent CRM state.

Explanation

Correct Answer

A narrow additive requirement often favors extension if it preserves existing routing, security and release controls.

Incorrect Answers

Requirements should be assessed and negotiated explicitly rather than silently omitted to protect a product choice.

A single field gap does not show that rebuilding authentication, orchestration and support is economical.

Prompt text does not enforce typed field persistence or downstream integration compatibility.

Question 469

A vendor pitches a custom customer-service agent with a more natural voice but it lacks the prebuilt agent’s validated transfer-to-human behavior. What is the minimum acceptance requirement?

Disable escalation tests in the custom benchmark because the new speech is more natural.
Grant the custom agent supervisor permissions to handle cases that its handoff cannot transfer.
Demonstrate at least equivalent routing, authorization, accessibility and escalation reliability on representative service cases.
Accept the custom option solely because its responses are longer and more conversational.

Explanation

Correct Answer

A conversational improvement is not a replacement for operational controls essential to actual customer support.

Incorrect Answers

Removing a failed control from evaluation would overstate the replacement’s production readiness.

Permission escalation does not repair the missing workflow and could enable unauthorized customer actions.

Response polish does not prove that unresolved customers reach the proper authenticated service queue.

Question 470

An insurer has an approved case-routing agent. A custom alternative may improve unusual claim triage but has not been measured. Which migration approach protects current service?

Switch all customer traffic to the custom agent immediately and retire the known-good rollback path.
Run both agents as independent writers for every claim without a shared case state.
Judge the custom agent solely on its development-team demo without independent case measurements.
Pilot the custom triage for the bounded exception class while preserving the prebuilt primary flow and comparing outcomes.

Explanation

Correct Answer

A narrow trial can establish incremental value without replacing a working platform before acceptance criteria are met.

Incorrect Answers

A full cutover before measured acceptance exposes routine service to unverified failures.

Concurrent uncoordinated action paths risk duplicate updates and contradictory customer communications.

A polished demo cannot establish error, escalation and permission behavior under actual claim conditions.

Agentic AI Business Solutions Architect (AB-100) Practice Tests

Microsoft AB-100 Practice Test 45 - 10 Questions (Last Reviewed 30-Sep-2026)
Microsoft AB-100 Practice Test 46 - 10 Questions (Last Reviewed 29-Sep-2026)
Microsoft AB-100 Practice Test 47 - 10 Questions (Last Reviewed 28-Sep-2026)
Microsoft AB-100 Practice Test 48 - 10 Questions (Last Reviewed 27-Sep-2026)
Microsoft AB-100 Practice Test 49 - 10 Questions (Last Reviewed 26-Sep-2026)

Pass Your Microsoft AB-100 Exam Easy!

Question 460

Explanation

Correct Answer

Incorrect Answers

Question 461

Explanation

Correct Answer

Incorrect Answers

Question 462

Explanation

Correct Answer

Incorrect Answers

Question 463

Explanation

Correct Answer

Incorrect Answers

Question 464

Explanation

Correct Answer

Incorrect Answers

Question 465

Explanation

Correct Answers

Incorrect Answers

Question 466

Explanation

Correct Answer

Incorrect Answers

Question 467

Explanation

Correct Answer

Incorrect Answers

Question 468

Explanation

Correct Answer

Incorrect Answers

Question 469

Explanation

Correct Answer

Incorrect Answers

Question 470

Explanation

Correct Answer

Incorrect Answers

Agentic AI Business Solutions Architect (AB-100) Practice Tests