Microsoft AB-100 (Agentic AI Business Solutions Architect) Practice Test 98

Question 970

The architect wants evidence before altering the deployment. In a workflow where candidate processes are attractive but permissions and data quality are weak, the first-wave pilot requests broad access to unreliable records. What should the team examine first?

  1. Benchmark accuracy, throughput and resource demands together
  2. Reconcile model assumptions against invoices and actual time studies
  3. Score use cases against data, identity, governance and integration readiness
  4. Audit decision state across sales and finance transactions

Explanation

Correct Answer

C:

The reported fault is that the first-wave pilot requests broad access to unreliable records in a system where candidate processes are attractive but permissions and data quality are weak; evidence from scoring use cases against data, identity, governance and integration readiness can show whether readiness assessment places safe high-value workloads ahead of risky ones.

Incorrect Answers

A:

In the process where high-volume narrow extraction needs predictable latency and cost, the diagnostic procedure could explain a general large model exceeds per-document unit economics; this particular incident instead turns on whether readiness assessment places safe high-value workloads ahead of risky ones and why the first-wave pilot requests broad access to unreliable records.

B:

The proposed probe concerns a pilot reports gross labor savings but omits operating expenses, where net benefit compares realized outcomes with implementation and recurring costs; this incident instead requires evidence about readiness assessment places safe high-value workloads ahead of risky ones and the failure that the first-wave pilot requests broad access to unreliable records.

D:

The evidence this method produces applies to a workflow where service, sales and finance jointly decide high-value concessions; it cannot by itself establish why the first-wave pilot requests broad access to unreliable records in the current scenario.

Question 971

Implementation alternatives must be compared for a workflow where two applications try to overwrite the same customer credit field. The pilot shows this gap: Parallel agent updates create conflicting authoritative values. Which alternative should lead the solution?

  1. Provide a unified approval handoff for AI-identified account risks
  2. Maintain system-of-record ownership across Dynamics integrations
  3. Use a controlled baseline to measure causal impact
  4. Tag documents with ownership, sensitivity, and provenance metadata

Explanation

Correct Answer

B:

Without the intended behavior, parallel agent updates create conflicting authoritative values; the proposed architecture establishes the specific design principle that field ownership and integration contracts prevent competing writes for the process where two applications try to overwrite the same customer credit field.

Incorrect Answers

A:

There is a valid place for the mechanism that cross-app workflow routes the recommendation to the accountable approver, namely service, sales and finance jointly decide high-value concessions; it does not remove the specific defect that parallel agent updates create conflicting authoritative values from this case.

C:

In the case where management attributes a sales lift to Copilot without a comparison, the alternate design depends on matched comparisons separate solution impact from unrelated shifts; it does not correct the target failure, parallel agent updates create conflicting authoritative values.

D:

In the case where search spans internal manuals with different owners and permissions, the alternate design depends on metadata enables filtering, review, retention and traceable answers; it does not correct the target failure, parallel agent updates create conflicting authoritative values.

Question 972

The architect wants evidence before altering the deployment. In a workflow where required knowledge stores are intermittently unreachable, the agent fabricates an answer during a search outage. What should the team examine first?

  1. Compare retrieval results for users with different permissions
  2. Compare API timestamps against retrieved document timestamps
  3. Inspect connector crawl scope, schema and ACL mappings
  4. Segment failures by route, token cost and difficulty
  5. Correlate retrieval timeouts with unsupported responses

Explanation

Correct Answer

E:

The reported fault is that the agent fabricates an answer during a search outage in a system where required knowledge stores are intermittently unreachable; evidence from correlating retrieval timeouts with unsupported responses can show whether health checks and explicit no-evidence responses handle missing context safely.

Incorrect Answers

A:

In the separate scenario where employee policies reside in governed Microsoft 365 documents, the same investigation could explain answers omit current employee handbook changes; it does not isolate the reported cause in a system where required knowledge stores are intermittently unreachable.

B:

That probe evaluates authorized actions query the system of record at request time in a workflow where customers ask for current delivery or payment status; the reported failure concerns the path for which health checks and explicit no-evidence responses handle missing context safely instead.

C:

The evidence this method produces applies to a workflow where teams want approved external records searchable in Copilot; it cannot by itself establish why the agent fabricates an answer during a search outage in the current scenario.

D:

Router saves tokens but sends difficult tasks to weak models; examining that other path does not determine whether health checks and explicit no-evidence responses handle missing context safely, which is what distinguishes the issue in the stated use case.

Question 973

A design adds autonomous behavior to a process where the classifier confuses complaints with refund requests. This creates a documented risk: A low-confidence label might be used for an irreversible eligibility decision. Which enforced control is the best fit?

  1. Require policy-owner approval for any rule change
  2. Disable autonomous regulated decisions under low evidence
  3. Require human review for high-impact low-confidence classifications
  4. Disallow factual claims unsupported by current policy data

Explanation

Correct Answer

C:

The business setting is the classifier confuses complaints with refund requests, with the exposure that a low-confidence label might be used for an irreversible eligibility decision; the enforceable protection must correspond to the path where representative examples teach decision boundaries without changing underlying facts.

Incorrect Answers

A:

Another process could need this guardrail because an unapproved change to the approval matrix could silently alter authorization decisions; the design being assessed supports the classifier confuses complaints with refund requests and has the explicit exposure that a low-confidence label might be used for an irreversible eligibility decision.

B:

This restriction is associated with the system cannot always find evidence for regulated guidance and addresses the exposure that a low-evidence agent might autonomously decide a regulated outcome; it does not specifically protect the target workflow against a low-confidence label might be used for an irreversible eligibility decision.

D:

The alternate boundary mitigates a model might state outdated rules as facts when authoritative policies have changed, which is relevant to answers depend on policies revised every week; it leaves the current workflow vulnerable to a low-confidence label might be used for an irreversible eligibility decision.

Question 974

Governance approves the business purpose but requires control of this exposure: Imported actions might retain staging endpoints after production deployment. The agent will be used when a flow moved to production still points to a test endpoint. Which boundary is most defensible?

  1. Block release when production variables or connections are unresolved
  2. Separate draft and send permissions in service integrations
  3. Restore prior version on any critical schema or safety failure
  4. Require a tested human escalation path for unmatched cases

Explanation

Correct Answer

A:

In the context where a flow moved to production still points to a test endpoint, the exposure is that imported actions might retain staging endpoints after production deployment; containing it requires a guardrail in the operation where environment configuration separates deployable logic from endpoint bindings.

Incorrect Answers

B:

Service agents draft customer messages but cannot send without review; its guardrail addresses the possibility that a service drafting assistant might gain permission to send unapproved notices; the present architecture must constrain the different threat that imported actions might retain staging endpoints after production deployment.

C:

This restriction is associated with business applications depend on the agent endpoint continuously and addresses the exposure that a canary model might cause critical schema or safety regressions without quick recovery; it does not specifically protect the target workflow against imported actions might retain staging endpoints after production deployment.

D:

Another process could need this guardrail because a new support agent might route urgent cases to a general queue; the design being assessed supports a flow moved to production still points to a test endpoint and has the explicit exposure that imported actions might retain staging endpoints after production deployment.

Question 975

Operations has not established the cause of this defect: A single-topic design cannot combine both operations naturally. The associated setting is: Users request a status check and an authorized follow-up action. Which inspection should precede remediation?

  1. Review plans for omitted or duplicate tool calls
  2. Inspect plugin capabilities and authentication configuration
  3. Inspect duplicate request IDs across retries
  4. Inspect step run histories and conditional branch outcomes

Explanation

Correct Answer

A:

In the setting where users request a status check and an authorized follow-up action, reviewing plans for omitted or duplicate tool calls can isolate the failure in which a single-topic design cannot combine both operations naturally, because the planner can select supported knowledge and tools for compound intents.

Incorrect Answers

B:

A Teams user should query and update a supported external service; its matching diagnostic check is reasonable for read-only knowledge alone cannot create a service request; the present investigation needs evidence for the different failure where a single-topic design cannot combine both operations naturally.

C:

In the separate scenario where network retries can repeat a single purchase request, the same investigation could explain two purchase orders appear from one user intent; it does not isolate the reported cause in a system where users request a status check and an authorized follow-up action.

D:

That probe evaluates defined flow steps make sequencing and error paths explicit in a workflow where approval processing requires sequential verification and notification; the reported failure concerns the path for which the planner can select supported knowledge and tools for compound intents instead.

Question 976

An agent-assisted service does not behave as specified: AI recommends an action but downstream case data is incomplete. It is used in this situation: A won opportunity should create a properly owned onboarding case. Which observable diagnostic procedure is best?

  1. Compare outcomes by locale, channel and agent version
  2. Inspect conversation state and parameter mapping across turns
  3. Audit token use and secret inventory for agent services
  4. Audit send events and approver identities after rollout
  5. Compare action privileges before and after package import
  6. Compare opportunity and case identifiers and required fields

Explanation

Correct Answer

F:

In the setting where a won opportunity should create a properly owned onboarding case, comparing opportunity and case identifiers and required fields can isolate the failure in which AI recommends an action but downstream case data is incomplete, because end-to-end tests verify identity, mapping and assignment across apps.

Incorrect Answers

A:

The proposed probe concerns a conversational change affects only one support geography, where cohort telemetry supports safe staged validation; this incident instead requires evidence about end-to-end tests verify identity, mapping and assignment across apps and the failure that AI recommends an action but downstream case data is incomplete.

B:

For the failure that it substitutes a prior customer's order in a later turn, inspecting conversation state and parameter mapping across turns is relevant; this question instead reports that AI recommends an action but downstream case data is incomplete, so different evidence is needed.

C:

Auditing token use and secret inventory for agent services explores a separate connection, where managed identity provides credential-less workload authentication where supported; the diagnosed requirement is a won opportunity should create a properly owned onboarding case, and the actual defect is AI recommends an action but downstream case data is incomplete.

D:

New prompt automation bypasses supervisor approval; examining that other path does not determine whether end-to-end tests verify identity, mapping and assignment across apps, which is what distinguishes the issue in the stated use case.

E:

A new connector allows a service identity to post directly; examining that other path does not determine whether end-to-end tests verify identity, mapping and assignment across apps, which is what distinguishes the issue in the stated use case.

Question 977

Before production promotion, the team needs evidence for this setting: Executives request AI before defining an operational improvement target. A prior regression caused the following: Prototype activity increases but no business value is measured. Which validation is most persuasive?

  1. Prototype the distinguishing behavior under production constraints
  2. Run offline evaluations of the trained candidate against holdouts
  3. Propose a concession above limit and confirm approval enforcement
  4. Map each proposed use case to a baseline metric and named owner

Explanation

Correct Answer

D:

The scenario is that executives request AI before defining an operational improvement target; the defect to rule out is that prototype activity increases but no business value is measured; testing via mapping each proposed use case to a baseline metric and named owner challenges the design dependency where cloud adoption planning begins from motivations, outcomes and readiness constraints.

Incorrect Answers

A:

This check exercises a different system behavior: custom engineering trades more control for more operating responsibility; a pass there cannot establish that prototype activity increases but no business value is measured no longer occurs in the required workflow.

B:

The suggested test serves a unique domain vocabulary produces persistent measured errors, in which teams propose custom training without any baseline; the present scenario instead concerns the reported problem where prototype activity increases but no business value is measured.

C:

The verification concerning service, sales and finance jointly decide high-value concessions checks cross-app workflow routes the recommendation to the accountable approver against a suggested discount bypasses delegated approval authority; its result says little about the observed issue where prototype activity increases but no business value is measured.

Question 978

Security and business owners want to keep this capability: Several agents need the same inventory query operation. They must also prevent the following: A shared MCP tool might trust the agent prompt rather than checking actual authorization. Which guardrail should the design incorporate?

  1. Ensure the backend rejects repeat writes for the same intent
  2. Confirm identity before sharing protected case details
  3. Keep authorization at the tool server, not just in agent instructions
  4. Provide accessible alternatives and honor channel policy

Explanation

Correct Answer

C:

The business setting is several agents need the same inventory query operation, with the exposure that a shared MCP tool might trust the agent prompt rather than checking actual authorization; the enforceable protection must correspond to the path where a supported MCP server exposes typed tool capabilities to clients.

Incorrect Answers

A:

Network retries can repeat a single purchase request; its guardrail addresses the possibility that an API retry might create a second purchase for the same user instruction; the present architecture must constrain the different threat that a shared MCP tool might trust the agent prompt rather than checking actual authorization.

B:

The alternate boundary mitigates a transferred voice conversation might disclose protected context before verifying identity, which is relevant to a voice bot cannot resolve a complex eligibility dispute; it leaves the current workflow vulnerable to a shared MCP tool might trust the agent prompt rather than checking actual authorization.

D:

Another process could need this guardrail because the voice channel might provide an inaccessible or unusable response form; the design being assessed supports several agents need the same inventory query operation and has the explicit exposure that a shared MCP tool might trust the agent prompt rather than checking actual authorization.

Question 979

Support engineers report this failure: Delayed reconciliation hides material fulfillment blockers. The business context is: Invoice holds must trigger exception handling without polling. What evidence should be collected to isolate the cause?

  1. Examine tool calls by transaction impact and approval requirement
  2. Audit rollout settings and incident timelines for uncontrolled changes
  3. Compare consumer errors to recent contract version changes
  4. Compare event delivery logs with receiving workflow state

Explanation

Correct Answer

D:

For the situation where invoice holds must trigger exception handling without polling, the defect is that delayed reconciliation hides material fulfillment blockers; comparing event delivery logs with receiving workflow state helps verify the cause-and-effect path: business events and durable orchestration propagate changes asynchronously.

Incorrect Answers

A:

For the failure that the same agent can read an invoice and commit a payment unchecked, examining tool calls by transaction impact and approval requirement is relevant; this question instead reports that delayed reconciliation hides material fulfillment blockers, so different evidence is needed.

B:

The proposed probe concerns new behavior must be evaluated without exposing the full workforce, where controlled release gates restrict blast radius and support rollback; this incident instead requires evidence about business events and durable orchestration propagate changes asynchronously and the failure that delayed reconciliation hides material fulfillment blockers.

C:

Comparing consumer errors to recent contract version changes explores a separate connection, where versioned schemas and compatibility policy make interfaces predictable; the diagnosed requirement is invoice holds must trigger exception handling without polling, and the actual defect is delayed reconciliation hides material fulfillment blockers.

Question 980

Security and business owners want to keep this capability: Standard capabilities draft customer communications but policy requires approval. They must also prevent the following: A drafting assistant could dispatch customer notices without a reviewer. Which guardrail should the design incorporate?

  1. Set enforceable monetary and scope thresholds outside prompts
  2. Quarantine unclassified or ownerless content pending review
  3. Honor cross-app role permissions and data minimization
  4. Separate draft permission from final dispatch permission

Explanation

Correct Answer

D:

When a drafting assistant could dispatch customer notices without a reviewer, the use case standard capabilities draft customer communications but policy requires approval requires a guardrail aligned with the behavior that a review step preserves accountability without discarding assisted drafting.

Incorrect Answers

A:

The proposed control targets a model could override a monetary limit through conversational instructions, while the given risk scenario is a drafting assistant could dispatch customer notices without a reviewer; these exposures arise in different operational contexts.

B:

For the use case where search spans internal manuals with different owners and permissions, the control addresses the threat that ownerless confidential drafts could become searchable before classification; the specified agent must be protected against a drafting assistant could dispatch customer notices without a reviewer.

C:

The alternate guard addresses cross-application aggregation might expose records outside the user's Dynamics permissions for sales and service agents need the same customer and case context; the present threat, a drafting assistant could dispatch customer notices without a reviewer, needs a different enforcement boundary.

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