Microsoft AB-100 Practice Test 12

Question 110

A policy-support agent gives outdated answers because it retrieves superseded documents. A vendor recommends training a custom language model on the same historic policies. What architectural comparison should the solution architect make before paying for custom training?

Refresh and permission-filter the approved retrieval corpus, benchmark the grounded base model against task cases, and justify custom training only if a measured residual gap remains.
Train a custom model on all archived and current policies without preserving effective dates, because model weights automatically select the legally binding revision.
Select custom training based on a lower training loss rather than comparing factual support and total lifecycle cost on held-out policy questions.
Increase training compute to compensate for stale or unauthorized documents entering the retrieval pipeline.

Explanation

Correct Answer

A retrieval and source-governance failure is not proof of insufficient model capability. Correct the evidence and compare representative accuracy, governance and ongoing costs before tuning.

Incorrect Answers

Embedding obsolete policy text into weights does not establish version precedence and may make source corrections harder to audit or revoke.

Training fit alone cannot show improved policy correctness on unseen queries; customization also adds labeling, deployment, monitoring and retraining costs.

More compute does not fix the grounding source selection, effectivity filters or access controls that determine which policy is authoritative.

Question 111

An agent receives routine FAQ requests and rare high-risk contract interpretations. Sending everything to an expensive model would exceed the operating budget. Which model-routing design best balances cost with decision correctness?

Route validated low-risk FAQ requests to an economical model, escalate ambiguous or high-impact contract tasks to a stronger path, and monitor accuracy and cost by workload.
Route every request to the cheapest model without measuring whether legal contract interpretations meet the required evaluation threshold.
Choose model destinations from fixed traffic percentages alone, regardless of the request’s risk and complexity.
Route by prompt length only and treat a short high-impact contract request as automatically low risk.

Explanation

Correct Answer

Risk-aware routing reserves higher-cost capability for tasks needing it and verifies that cheap-path errors remain within threshold; cost savings must not conceal legal harm.

Incorrect Answers

Cheap inference is not an acceptable trade if contract advice fails grounded accuracy, refusal or review criteria.

Volume ratios measure traffic distribution but do not identify contract complexity, source requirements or routing misclassifications.

A short request can still require consequential legal reasoning, so token count alone is not a safe model-selection policy.

Question 112

A nonprofit projects a Copilot solution for 250 users but expects only 80 monthly active users. Licenses are provisioned to all 250, while agent inference depends on use. Which budgeting distinction prevents underestimation?

Separate provisioned per-user licensing from activity-based model and connector usage.
Charge license costs only to the 80 active users even if all 250 seats are billed.
Omit implementation, support and monitoring because users already have accounts.
Assume usage-based inference cannot vary once licenses have been purchased.

Explanation

Correct Answer

Per-seat charges and metered consumption have different cost drivers and must be modeled independently.

Incorrect Answers

Licensed seats can cost money even when assigned users are inactive.

Existing identities do not eliminate deployment and maintenance expense.

Consumption still depends on workload volume and token/tool usage.

Question 113

Dynamics 365 service records store a product field as product_code but users ask Copilot about “subscription tier.” What design improves term interpretation without changing the system of record?

Replace product_code labels in the agent prompt with “subscription tier” without validating any code-to-tier mapping against approved CRM metadata.
Define and validate a business-term mapping from subscription tier to the appropriate approved product field.
Tell the model to invent a tier whenever product_code is absent.
Rewrite every existing product record to match informal subscription labels without preserving official product codes or dependent integration keys.

Explanation

Correct Answer

A documented semantic mapping can make natural-language queries consistent with established business schema.

Incorrect Answers

A wording change alone does not establish which persisted product code maps to an actual subscription entitlement.

Inference without data should not fabricate commercial entitlements.

Renaming source records to fit colloquial language can break Dynamics 365 keys and reporting relationships; use an approved mapping instead.

Question 114

A contact agent summarizes long customer cases in Dynamics 365 but must exclude internal investigation notes from customer responses. What customization is appropriate?

Write a single generic prompt with no field-level access checks.
Use a scoped summary action with approved fields and an explicit exclusion policy before generation.
Provide every note to the model and ask it to suppress sensitive details afterward.
Copy internal investigation notes into a public knowledge source.

Explanation

Correct Answer

Source and field permissions should be enforced before a summary is composed or delivered.

Incorrect Answers

A generic prompt cannot define data entitlement.

Prompt-level suppression does not protect prior context exposure.

Publishing internal notes defeats their restricted classification.

Question 115

An account-manager agent needs to read Dynamics 365 Sales opportunities through a connector without changing them. Which access design is appropriate?

Configure a read-scoped authenticated connector that applies the requesting user’s opportunity permissions.
Let the agent bypass OAuth when its user question contains a valid account name.
Implement the connector as a write-only API because it returns richer JSON.
Use one permanent tenant admin connection for every opportunity query.

Explanation

Correct Answer

A secure connector requires supported authentication and least-privilege record retrieval.

Incorrect Answers

Account names are not authentication credentials.

Write-only operations are unrelated to the read requirement.

Global access ignores per-user permissions and increases disclosure risk.

Question 116

A Dynamics 365 Contact Center agent serves web chat and voice calls, and a case must follow the customer across channels. What should the design carry?

Separate disconnected records per channel so no history can be reconciled.
Use a caller telephone number as the universal case key across voice and chat without identity verification or authenticated cross-channel linkage.
An assumption that voice callers and chat users always share one verified identity.
A stable authenticated customer/case correlation with channel-specific consent and escalation handling.

Explanation

Correct Answer

Cross-channel case continuity depends on identity, correlation and safe handoff semantics.

Incorrect Answers

Disconnected records fragment the service timeline.

Telephone numbers can be shared or spoofed and may not identify a chat session. Match the verified customer and durable case reference.

Channel authentication differs and cannot be assumed equivalent.

Question 117

A customer-service task agent checks order state, asks for a return reason and creates a proposed return task for staff. Which agent design fits?

A completely unscoped autonomous agent with unrestricted refund authority.
A bounded task agent that validates each step and yields a reviewable return record.
A free-text summary that claims the return was created without calling the workflow.
Use a read-only case summarizer and treat its narrative as an executed return without invoking the authenticated return action.

Explanation

Correct Answer

Task agents orchestrate defined operations and should report actual persisted outcomes.

Incorrect Answers

Broad autonomy exceeds the limited delegated task.

A narrative cannot substitute for a confirmed record creation.

A summary may describe the correct intention, but it cannot create a persisted return transaction or validate eligibility.

Question 118

A supply chain team wants an event-triggered agent to monitor stockouts and propose urgent substitutions overnight. Changes above a purchasing threshold must remain manual. What architecture works?

An event-driven autonomous proposal agent with enforced spend limits and human approval for exceptions.
Use a timer-driven purchasing agent to commit substitutions directly overnight without rechecking the stockout event or approval ceiling.
An agent instructed to ignore stockout events when managers are offline.
A bot that makes every purchase without an inventory check or threshold.

Explanation

Correct Answer

Autonomous detection can be bounded by deterministic authorization and human-signoff controls.

Incorrect Answers

A schedule is not a reliable event signal and cannot authorize spend; monitor authoritative inventory and enforce the threshold.

Ignoring off-hours events defeats the monitoring objective.

Unverified purchases can breach procurement policy.

Question 119

A legal team needs an agent that answers policy questions but never changes client records. Which agent type and boundary are most suitable?

Use general model recall for the policy answer and generate source IDs afterward without checking the approved policy repository.
A grounded prompt-and-response agent with read-only source access and citations.
A full autonomous data-editing agent to answer static questions.
Choose an autonomous record-maintenance agent with policy-write permissions even though users require only read-only advice.

Explanation

Correct Answer

Read-only grounded assistance fits knowledge requests without unnecessary mutation authority.

Incorrect Answers

Synthetic citations cannot substantiate an authoritative policy answer. Retrieve approved, accessible policy text and preserve actual provenance.

Transactional autonomy adds risk where no write is required.

Unnecessary write privileges introduce policy-integrity risk; a grounded read-only answerer meets the stated boundary.

Question 120

A business app must read invoice totals from scanned PDFs before a downstream agent checks them. Which Microsoft AI capability category should be considered?

A generic model instructed to invent totals when scans are unreadable.
A document extraction service designed to return fields and supporting document evidence.
Run generic OCR text extraction without preserving table and invoice-field structure needed to validate totals and taxes.
Classify the document’s overall intent from extracted text without extracting the authoritative amount, tax and supplier fields.

Explanation

Correct Answer

Document-specific extraction should expose fields and uncertainty for validation before business actions.

Incorrect Answers

Invented totals cannot support financial reconciliation.

OCR text alone may omit invoice field relationships, confidence and source spans needed for downstream reconciliation.

An invoice type label does not supply structured amounts or document evidence required for financial checks.

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

Microsoft AB-100 Practice Test 10 - 10 Questions (Last Reviewed 04-Nov-2026)
Microsoft AB-100 Practice Test 11 - 10 Questions (Last Reviewed 03-Nov-2026)
Microsoft AB-100 Practice Test 12 - 10 Questions (Last Reviewed 02-Nov-2026)
Microsoft AB-100 Practice Test 13 - 10 Questions (Last Reviewed 01-Nov-2026)
Microsoft AB-100 Practice Test 14 - 10 Questions (Last Reviewed 31-Oct-2026)

Pass Your Microsoft AB-100 Exam Easy!

Question 110

Explanation

Correct Answer

Incorrect Answers

Question 111

Explanation

Correct Answer

Incorrect Answers

Question 112

Explanation

Correct Answer

Incorrect Answers

Question 113

Explanation

Correct Answer

Incorrect Answers

Question 114

Explanation

Correct Answer

Incorrect Answers

Question 115

Explanation

Correct Answer

Incorrect Answers

Question 116

Explanation

Correct Answer

Incorrect Answers

Question 117

Explanation

Correct Answer

Incorrect Answers

Question 118

Explanation

Correct Answer

Incorrect Answers

Question 119

Explanation

Correct Answer

Incorrect Answers

Question 120

Explanation

Correct Answer

Incorrect Answers

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