Microsoft AB-100 Practice Test 55

Question 540

A finance agent handles 10,000 purchase-order requests monthly. Seven percent require human review and three percent need retries. The proposed budget includes only successful model tokens. What must be added?

Treat any retry as a second business success to raise the productivity metric.
Expected reviewer time, retry and tool-call usage, exception overhead and the cost of correcting failed outcomes.
Exclude every escalated request from the cost model because it is handled by a human.
Multiply the number of attempted requests by the average token cost and assume every transaction completes.

Explanation

Correct Answer

Manual and retried cases consume capacity and affect cost per correctly completed purchase request.

Incorrect Answers

An additional attempt is not an independently completed purchase and may reflect a prior ambiguous failure.

Escalation remains part of the end-to-end agent-enabled workflow and its labor cost belongs in TCO.

A raw attempt estimate ignores the extra retries, manual processing and incomplete operations specified in the case.

Question 541

A Dynamics 365 service team calls tickets “overdue” when `dueDate` is before the user’s business day and `status` is not closed. Copilot searches for an `overdue` column that does not exist. What should the term configuration do?

Add a placeholder overdue field that agents fill by guessing whether each case is late.
Consider all cases with any populated due date overdue.
Define a governed business term that maps overdue to the documented due-date and active-status predicate.
Use only keyword search for the string overdue in case notes.

Explanation

Correct Answer

A domain synonym needs an explicit data interpretation rather than a nonexistent literal field.

Incorrect Answers

Model guesses cannot replace an auditable predicate over current structured records.

Having a due date is not sufficient; its value and the case status determine lateness.

A written mention can appear in cases that are already closed or whose actual due date has not passed.

Question 542

A finance user asks Copilot for “this quarter’s sales” in a Dynamics 365 app. The company’s fiscal quarter runs February through April, not calendar January through March. What must the business-term layer encode?

Use whichever calendar quarter is closest to today regardless of the organization’s reporting period.
Infer the fiscal quarter from the most frequent dates in the retrieved sales records.
Treat every open order as recognized sales for the quarter without checking posting status.
The approved fiscal calendar and timezone boundaries used by the business’s posted-sales measure.

Explanation

Correct Answer

A natural-language quarter reference must resolve to the correct accounting window and trusted transaction definition.

Incorrect Answers

Calendar months may include or exclude transactions that belong to the actual business fiscal quarter.

Record frequency cannot replace the company’s explicitly defined calendar start and end boundaries.

A pending order may not satisfy the approved revenue recognition and sales reporting definition.

Question 543

A Sales manager asks Copilot for “actionable deals.” The CRM includes open opportunities, suspended fraud-flagged opportunities and already won records. How should the query be grounded?

Use a documented predicate for open, permitted, non-suspended opportunities and retain the source field meaning.
Count all opportunities with a nonempty owner field as actionable.
Include won and closed-lost opportunities to maximize the number of suggested deals.
Ask the model to decide which records seem urgent based only on the opportunity title.

Explanation

Correct Answer

The business label must exclude blocked and completed records rather than match the word open alone.

Incorrect Answers

Ownership says who manages a record, not whether its status permits active sales work.

Completed outcomes cannot ordinarily be treated as current actionable pipeline opportunities.

Natural-language urgency cannot replace actual status and compliance flags in the CRM.

Question 544

An account manager asks a Copilot agent for the “customer balance.” Finance exposes posted receivables and unposted credit memos. Which semantic contract prevents a misleading answer?

Use the model’s explanation text as the authoritative recorded account balance.
Define the balance measure and cutoff status explicitly, identifying whether pending adjustments are included.
Subtract every draft credit memo automatically as though it were posted.
Return whichever measure has the highest amount without explaining its status.

Explanation

Correct Answer

Posted ledger values and prospective adjustments are different accounting states that need named business meanings.

Incorrect Answers

A generated narrative cannot establish an actual financial posting or reconciliation status.

Unapproved draft adjustments may never affect the ledger and would understate current receivables.

Magnitude does not define which accounting quantity the requester intended.

Question 545

A Dynamics 365 Sales user asks “show my accounts.” The account record has an assigned relationship manager while contacts have different individual owners. What should the agent’s business term resolve?

Show all accounts regardless of owner and hide unrelated names from the final prose.
Select the account whose display name most closely resembles the employee’s name.
Use the verified account-level owner field and the requester’s permissions, not contact-level ownership.
Include every account with at least one contact owned by the requester.

Explanation

Correct Answer

An account list should follow the company’s account stewardship definition rather than infer ownership from related contacts.

Incorrect Answers

Post-retrieval hiding cannot enforce actual record-level access boundaries.

Name similarity is unrelated to the CRM’s authenticated account ownership assignment.

Contact ownership does not necessarily grant access to or make the user owner of the parent account.

Question 546

A multinational support team calls canceled subscriptions “terminated” in one region and “closed” in another. Copilot reports different totals although both regions use the same approved lifecycle codes. What should be standardized?

Create separate arbitrary lifecycle code values for each translation without coordinating the service data model.
Have the agent classify subscriptions by emotional tone of customer notes.
Combine every closed support case with every canceled subscription under one label.
Versioned multilingual synonyms mapped to the same underlying lifecycle codes and effective business definition.

Explanation

Correct Answer

Local wording may differ, but reporting must consistently interpret the authoritative subscription state.

Incorrect Answers

Unmapped code variants would fragment shared metrics and create incompatible regional definitions.

Conversation sentiment does not determine formal contract termination status.

Case closure and contract termination are different domain entities with different financial consequences.

Question 547

A customized Customer Service Copilot drafts resolution text. Some cases carry a legal-hold flag that forbids closure even when a customer reply is ready. Which customization is appropriate?

Draft the permitted response but make the closure action check legal-hold status and verified reviewer authorization.
Remove the hold field from the model input and let it close cases with polished summaries.
Treat a customer’s appreciative reply as proof that the legal hold has ended.
Close cases first and restore the hold if the compliance team objects.

Explanation

Correct Answer

Generating a note and closing a regulated case are different operations; the service must enforce the hold.

Incorrect Answers

Concealing the hold does not remove the legal restriction and can make unauthorized closure more likely.

Customer sentiment is not an authenticated order lifting retention or investigation requirements.

A post-hoc reversal may not undo a prohibited closure or missed retention action.

Question 548

A service Copilot prepares an email from a case containing public troubleshooting steps and private fraud-investigation commentary. How should its response grounding be built?

Copy private notes into the email footer with a small font to preserve completeness.
Supply only customer-authorized fields to the external reply prompt and retain private notes in a separate restricted context.
Give the model the full case and ask it to avoid quoting the private notes.
Remove the internal note label while keeping the same note text in the prompt.

Explanation

Correct Answer

Model scope must match the intended audience before generation so confidential case evidence is never exposed.

Incorrect Answers

Formatting does not create permission to disclose restricted internal investigation material.

A prose instruction cannot guarantee that hidden fraud details do not influence or leak through the draft.

Changing a field label does not restrict the sensitive content itself.

Question 549

A service agent can draft a customer compensation offer. The company limits goodwill credits to $150 without supervisor approval. What boundary is needed?

Allow the agent to promise any amount but block only the eventual payment action.
Reduce the agent’s response length and assume a shorter offer cannot exceed the limit.
A service-side amount check and authenticated supervisor signoff for any offer above the delegated threshold.
Treat any manager mentioned in the email chain as having approved the current offer.

Explanation

Correct Answer

A generated offer can create financial expectations and therefore must observe the same authority as a direct manual credit.

Incorrect Answers

An unauthorized binding promise can itself create business risk before financial posting occurs.

Text length does not measure the value or authorization of the proposed compensation.

A name in conversation text is not a verified signoff for this exact amount and case.

Question 550

A supervisor edits a Copilot-generated service summary because it omitted a promised callback date. What should be stored for accountable use?

Only the final summary, deleting the underlying case references and human correction history.
The generated text plus a generic “reviewed” label without an authenticated person.
A chat reaction emoji as the sole record of acceptance.
The source case version, original suggestion, corrected summary, reviewer identity and approved publication timestamp.

Explanation

Correct Answer

Separate machine suggestion and reviewer changes make customer communications reproducible and attributable.

Incorrect Answers

Removing the source and edit trail prevents auditors from reconstructing the omitted callback commitment.

A nonattributed label cannot establish who accepted the customer-facing correction.

An informal reaction does not preserve the published content or the exact evidence used during review.

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

Microsoft AB-100 Practice Test 53 - 10 Questions (Last Reviewed 22-Sep-2026)
Microsoft AB-100 Practice Test 54 - 10 Questions (Last Reviewed 21-Sep-2026)
Microsoft AB-100 Practice Test 55 - 10 Questions (Last Reviewed 20-Sep-2026)
Microsoft AB-100 Practice Test 56 - 10 Questions (Last Reviewed 19-Sep-2026)
Microsoft AB-100 Practice Test 57 - 10 Questions (Last Reviewed 14-Nov-2026)

Pass Your Microsoft AB-100 Exam Easy!

Question 540

Explanation

Correct Answer

Incorrect Answers

Question 541

Explanation

Correct Answer

Incorrect Answers

Question 542

Explanation

Correct Answer

Incorrect Answers

Question 543

Explanation

Correct Answer

Incorrect Answers

Question 544

Explanation

Correct Answer

Incorrect Answers

Question 545

Explanation

Correct Answer

Incorrect Answers

Question 546

Explanation

Correct Answer

Incorrect Answers

Question 547

Explanation

Correct Answer

Incorrect Answers

Question 548

Explanation

Correct Answer

Incorrect Answers

Question 549

Explanation

Correct Answer

Incorrect Answers

Question 550

Explanation

Correct Answer

Incorrect Answers

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