Microsoft AB-100 Practice Test 6

Question 50

A sales opportunity agent draws data from Dynamics 365 Sales and customer payment status from Dynamics 365 Finance. What design prevents contradictory customer identities?

Have the model guess which Finance entity corresponds to each sales contact.
Use governed identity mapping and versioned integration contracts between the applications.
Link customers by normalized account display name without an authoritative CRM-to-Finance identifier and currency-aware contract.
Copy Finance balances into a free-text prompt and ignore reconciliation.

Explanation

Correct Answer

Cross-app workflows require stable keys, ownership and tested mappings before AI conclusions can be trusted.

Incorrect Answers

Identity inference cannot replace authoritative business system joins.

Display names can collide and change; financial data integration needs stable keys and explicit currency semantics.

Unreconciled prompt copies can misstate receivables.

Question 51

A customer refund agent identifies a possible fraud flag but the policy requires an investigator to approve large refunds. How should the execution be designed?

Issue the refund and send a fraud summary to investigators at the end of the day.
Let the agent use a shared investigator account when the queue is long.
Use a free-text “approved” label in chat to authorize the financial tool.
Provide a hold-and-review task and release the refund only after an authenticated approval.

Explanation

Correct Answer

High-impact actions need enforceable transaction authorization and a clear audit trail.

Incorrect Answers

Notification after payment is not a prior approval.

Shared investigator identity destroys accountability.

A chat label is not a verified authorization decision.

Question 52

An enterprise begins its AI program and lacks reliable data ownership or a deployment control plane. Select TWO sensible first actions before opening production agents.

Run a bounded, measurable proof of concept on a prioritized feasible use case.
Judge program success only by the number of prompts created.
Roll out full production agents before defining rollback and monitoring.
Grant every prototype unrestricted access to live customer data for quicker adoption.
Establish AI governance and data-owner responsibilities with explicit risk controls.

Explanation

Correct Answers

A bounded proof of concept provides evidence of value and feasibility before scaling.

Clear ownership and risk policies are required for safe organizational adoption.

Incorrect Answers

Prompt count is a poor proxy for real business outcomes.

Ungated production deployment makes failures hard to contain and diagnose.

Broad permissions increase exposure before controls have been evaluated.

Question 53

A public agency compares candidate agent initiatives: a high-volume document classification task has labeled historical data, while a rare discretionary decision lacks an accountable reviewer. Which should be piloted first?

The high-volume classification case with measurable error and throughput baselines.
The rare discretionary decision solely because it sounds more autonomous.
Choose the discretionary decision because its demo has the highest conversational satisfaction rating, without a task-outcome baseline.
Both initiatives at once without capacity or outcome measures.

Explanation

Correct Answer

Use-case ranking weighs validated data, value and risks; a measurable classification pipeline is an appropriate early pilot.

Incorrect Answers

Autonomy is not itself evidence of feasibility or impact.

A subjective demo rating is not a business impact, case volume, feasibility or safety measure for investment prioritization.

Concurrent unscoped pilots consume governance and engineering capacity.

Question 54

A project manager needs natural-language meeting summaries inside Microsoft 365 but not external writes. Another team requires approval-based work orders in Dynamics 365. Which platform strategy is sound?

Ban native Microsoft 365 assistance and rebuild both services from scratch.
Route every work order solely through an unreviewed meeting-summary prompt.
Configure Microsoft 365 Copilot for the read-focused summary and a separately scoped transactional agent for work orders.
Give the meeting assistant full Dynamics 365 administrative access to handle both needs.

Explanation

Correct Answer

A familiar productivity surface can meet read-only tasks; enterprise actions need dedicated authorization and tool contracts.

Incorrect Answers

Custom rebuilding wastes supported capabilities without a demonstrated gap.

A summarizer does not enforce the transactional order lifecycle.

Unnecessary privileges widen risk beyond the summary task.

Question 55

A finance-planning agent passes a proposed inventory purchase to an execution agent. Which handoff design best prevents duplicate orders after a retry?

Let two agents call the payment tool in parallel without correlation IDs.
Treat the execution agent’s acknowledgement message as order commitment without reconciling the purchase service receipt.
Send typed order fields with an idempotency key and check the transaction result before retrying.
Replay the same free-text request repeatedly until the model says it succeeded.

Explanation

Correct Answer

Idempotent contracts and verified acknowledgments prevent multiple writes from a single intended action.

Incorrect Answers

Uncoordinated parallel calls risk duplicated commitment.

A conversational or inter-agent acknowledgement cannot prove a durable transaction; check the authoritative purchase status before any retry.

Model-generated claims do not prove the purchase result.

Question 56

A customer service platform already offers a supported agent for common case summarization. A team wants custom code solely to reproduce the same feature. What decision best controls cost?

Train a new model on all cases without a privacy or evaluation plan.
Use the prebuilt feature first and build only for a documented missing capability.
Discard the tested prebuilt agent without comparing support and integration expenses.
Assume custom code is always cheaper because it avoids configuration.

Explanation

Correct Answer

Capability reuse reduces redundant development and maintenance when requirements are met.

Incorrect Answers

Uncontrolled training increases risk without proving business benefit.

A build decision needs explicit capability and total-cost evidence.

Custom systems incur lifecycle, security and support costs.

Question 57

A human resources Copilot Studio action can update employee leave balances. Which tool contract is necessary to prevent accidental cross-employee edits?

Use the employee’s first name as a permanent unique key.
Verify a stable employee identifier and server-side caller permission at every action invocation.
Permit the model to choose any employee record it finds semantically similar.
Grant all actions application-owner rights to avoid authorization errors.

Explanation

Correct Answer

Typed identifiers and request-level authorization protect sensitive data during generated tool calls.

Incorrect Answers

Names are not unique or immutable record keys.

Vector similarity is not an identity check.

Broad privileges bypass the user’s proper access scope.

Question 58

A policy agent has two connected sources, an approved SharePoint policy library and an unmanaged personal drive. Which grounding strategy belongs in production?

Use the controlled library with access trimming and explicit source ownership.
Retrieve restricted policy documents first and hide them only in the response.
Let employees paste personal drafts into system instructions without review.
Search both equally and trust the newest-looking file title.

Explanation

Correct Answer

Authority and permission should govern retrieval, not merely textual relevance.

Incorrect Answers

Late concealment does not protect the model context or logs.

Uncontrolled drafts create conflicting policy facts.

Filename recency is not publication approval.

Question 59

Finance personnel want a new custom portal to ask questions already supported by Microsoft 365 Copilot in their existing workflow. What should the architect evaluate?

Whether training on all tenant files would remove the need for source permissions.
Whether copying the whole Microsoft 365 document store into the new portal is easiest.
Whether a supported scoped Copilot extension meets the needs before funding a parallel app.
Whether every user can be made a tenant administrator for convenience.

Explanation

Correct Answer

Extending an existing supported experience often avoids duplicated UI, identity and maintenance costs.

Incorrect Answers

Training does not remove data entitlements.

Bulk copying documents creates needless duplication and exposure.

Administrator access is not a reasonable extension prerequisite.

Question 60

A service company has a well-performing grounded model but experiences errors on internal product codes. What is an appropriate signal that custom tuning might be justified?

Measured failure on a representative held-out product-code task despite strong prompting and grounding baselines.
A single inaccurate conversation without reproducing its cause.
Assume a larger GPU allocation justifies fine-tuning even though held-out product-code accuracy already meets the business threshold.
Treat the model vendor’s published average benchmark as proof custom tuning is necessary without testing company-specific product-code errors.

Explanation

Correct Answer

A targeted validated gap and quality baseline can justify extra training and governance cost.

Incorrect Answers

One example may be bad retrieval rather than model deficiency.

Hardware availability does not prove a model-capability gap; training is justified by representative measured failures after grounding and prompting have been tested.

General vendor scores do not show that retrieval and prompting baselines fail on the actual internal code distribution.

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

Microsoft AB-100 Practice Test 4 - 10 Questions (Last Reviewed 10-Nov-2026)
Microsoft AB-100 Practice Test 5 - 10 Questions (Last Reviewed 09-Nov-2026)
Microsoft AB-100 Practice Test 6 - 10 Questions (Last Reviewed 08-Nov-2026)
Microsoft AB-100 Practice Test 7 - 10 Questions (Last Reviewed 07-Nov-2026)
Microsoft AB-100 Practice Test 8 - 10 Questions (Last Reviewed 06-Nov-2026)

Pass Your Microsoft AB-100 Exam Easy!

Question 50

Explanation

Correct Answer

Incorrect Answers

Question 51

Explanation

Correct Answer

Incorrect Answers

Question 52

Explanation

Correct Answers

Incorrect Answers

Question 53

Explanation

Correct Answer

Incorrect Answers

Question 54

Explanation

Correct Answer

Incorrect Answers

Question 55

Explanation

Correct Answer

Incorrect Answers

Question 56

Explanation

Correct Answer

Incorrect Answers

Question 57

Explanation

Correct Answer

Incorrect Answers

Question 58

Explanation

Correct Answer

Incorrect Answers

Question 59

Explanation

Correct Answer

Incorrect Answers

Question 60

Explanation

Correct Answer

Incorrect Answers

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