Microsoft AB-100 (Agentic AI Business Solutions Architect) Practice Test 85

Question 840

Troubleshooting starts from a business complaint: Parallel agent updates create conflicting authoritative values. The affected process is: Two applications try to overwrite the same customer credit field. Which investigation is most directly tied to the reported defect?

  1. Audit samples for missing owner, classification and source identifiers
  2. Inspect write conflicts and source-of-truth definitions
  3. Inspect baseline equivalence and external trend factors
  4. Compare metered usage with adoption and concurrency assumptions

Explanation

Correct Answer

B:

The reported fault is that parallel agent updates create conflicting authoritative values in a system where two applications try to overwrite the same customer credit field; evidence from inspecting write conflicts and source-of-truth definitions can show whether field ownership and integration contracts prevent competing writes.

Incorrect Answers

A:

Auditing samples for missing owner, classification and source identifiers explores a separate connection, where metadata enables filtering, review, retention and traceable answers; the diagnosed requirement is two applications try to overwrite the same customer credit field, and the actual defect is parallel agent updates create conflicting authoritative values.

C:

Management attributes a sales lift to Copilot without a comparison; its matching diagnostic check is reasonable for seasonality produces the same change in unassisted teams; the present investigation needs evidence for the different failure where parallel agent updates create conflicting authoritative values.

D:

The proposed inspection is associated with an agent scales from a trial to enterprise usage, where workload forecasts include model usage, connectors, storage and observability; it would not expose the failure mode that parallel agent updates create conflicting authoritative values with the required specificity.

Question 841

One enterprise rollout has 2 separately testable requirements. (1) Business use: An agent scales from a trial to enterprise usage. Known defect: Token and action costs exceed the pilot estimate. (2) Business use: A pilot reports gross labor savings but omits operating expenses. Known defect: The investment case overstates value by ignoring review and hosting. Which 2 diagnostic procedures directly satisfy these stated cases? Choose TWO.

  1. Audit rollout settings and incident timelines for uncontrolled changes
  2. Correlate throttling errors with fallback activation
  3. Reconcile model assumptions against invoices and actual time studies
  4. Compare metered usage with adoption and concurrency assumptions

Explanation

Correct Answers

C:

Reconciling model assumptions against invoices and actual time studies can reveal why the investment case overstates value by ignoring review and hosting when a pilot reports gross labor savings but omits operating expenses; that finding is needed alongside the independent diagnosis for an agent scales from a trial to enterprise usage.

D:

Comparing metered usage with adoption and concurrency assumptions can reveal why token and action costs exceed the pilot estimate when an agent scales from a trial to enterprise usage; that finding is needed alongside the independent diagnosis for a pilot reports gross labor savings but omits operating expenses.

Incorrect Answers

A:

This diagnosis is associated with the different failure where a prompt update changes decisions for all users at once in a workflow where new behavior must be evaluated without exposing the full workforce; it does not directly solve the listed conditions: token and action costs exceed the pilot estimate; the investment case overstates value by ignoring review and hosting.

B:

This diagnosis is associated with the different failure where requests fail despite another approved deployment being available in a workflow where peak load can saturate a primary model endpoint; it does not directly solve the listed conditions: token and action costs exceed the pilot estimate; the investment case overstates value by ignoring review and hosting.

Question 842

A release must prove that a generic instruction cannot express subtle category boundaries will not recur. The application is used in this situation: The classifier confuses complaints with refund requests. Which acceptance test most directly checks that failure mode?

  1. Test adversarial lookalike cases not present in prompt examples
  2. Ask counterfactual questions and verify claims map to retrieved evidence
  3. Replay boundary cases and verify the same branch and approver are selected
  4. Remove required sources and check that the agent declines

Explanation

Correct Answer

A:

Testing adversarial lookalike cases not present in prompt examples challenges the path where representative examples teach decision boundaries without changing underlying facts for the business setting the classifier confuses complaints with refund requests; the specific defect to exclude is that a generic instruction cannot express subtle category boundaries.

Incorrect Answers

B:

In the alternate setting where analysts compare unstructured reports but must retain human interpretation, the test exercises retrieval supplies supporting evidence while analysts own analytical decisions; even a pass would not show that a generic instruction cannot express subtle category boundaries has been prevented.

C:

This check exercises a different system behavior: explicit conditions execute known business rules predictably and leave an auditable branch history; a pass there cannot establish that a generic instruction cannot express subtle category boundaries no longer occurs in the required workflow.

D:

Although the exercise might catch an agent guesses a policy instead of escalating, it would not directly validate the mechanism that representative examples teach decision boundaries without changing underlying facts in the target implementation.

Question 843

The operating team must monitor whether an architectural correction works. Business context: Employee policies reside in governed Microsoft 365 documents. Historical symptom: Answers omit current employee handbook changes. Which outcome measure is the best match?

  1. Authorized structured-record answer accuracy
  2. Correct citations to accessible policy material
  3. Custom workflow completion against the required execution contract
  4. Cross-domain tasks routed to the correct specialist

Explanation

Correct Answer

B:

For the use case where employee policies reside in governed Microsoft 365 documents, the indicator correct citations to accessible policy material is worth baselining against the earlier issue where answers omit current employee handbook changes.

Incorrect Answers

A:

Users ask about governed CRM records rather than prose manuals; its intended measurement is authorized structured-record answer accuracy; substituting that result here would not capture whether answers omit current employee handbook changes has been corrected.

C:

The premise of this indicator is custom-engine architecture controls runtime, model calls and integrations; the architecture being evaluated instead serves employee policies reside in governed Microsoft 365 documents, and its success is measured by correct citations to accessible policy material.

D:

For the workflow where a request spans procurement, finance and policy expertise, cross-domain tasks routed to the correct specialist is an actionable measure; it does not establish whether answers omit current employee handbook changes in the situation under review.

Question 844

A pilot scorecard needs evidence of business improvement rather than only invocation counts. The use case is: AI drafts requests quickly but specialists must correct some outputs. Known defect: Average handling-time savings disappear during rework. Which KPI should be selected?

  1. Fully loaded cost per correctly resolved case
  2. Monthly run-rate per active automated workflow
  3. Conflicting cross-application writes per thousand transactions
  4. Successful requests during primary endpoint failures

Explanation

Correct Answer

A:

Fully loaded cost per correctly resolved case; this outcome measure belongs to the workflow where AI drafts requests quickly but specialists must correct some outputs; a change in it can help detect whether the defect that average handling-time savings disappear during rework is resolved.

Incorrect Answers

B:

The alternative tracks monthly run-rate per active automated workflow, arising from the mechanism where workload forecasts include model usage, connectors, storage and observability; its movement alone says little about the failure that average handling-time savings disappear during rework.

C:

Two applications try to overwrite the same customer credit field; its intended measurement is conflicting cross-application writes per thousand transactions; substituting that result here would not capture whether average handling-time savings disappear during rework has been corrected.

D:

That scorecard concerns the activity in which peak load can saturate a primary model endpoint and its result of successful requests during primary endpoint failures; the reported shortcoming here is average handling-time savings disappear during rework.

Question 845

The architecture board has evidence that agent-generated pages render an unsafe instruction or link. The affected situation is: A tool response includes HTML or links supplied by third parties. What change most directly establishes the missing behavior?

  1. Link negative feedback to grounded traces and source versions
  2. Constrain high-risk tools with action-level policy checks
  3. Validate and sanitize tool output before rendering it
  4. Require independent backend authorization for generated tool calls

Explanation

Correct Answer

C:

The architecture must reconcile two facts: a tool response includes HTML or links supplied by third parties, yet agent-generated pages render an unsafe instruction or link; the selected pattern does so through the mechanism that trusted rendering and output validation constrain returned content.

Incorrect Answers

A:

In the case where users report stale answers after a policy refresh, the alternate design depends on trace and content-version linkage makes failures testable; it does not correct the target failure, agent-generated pages render an unsafe instruction or link.

B:

For another type of architecture, execution-time destination and permission checks block exfiltration is useful; this design must instead support a tool response includes HTML or links supplied by third parties despite the problem where agent-generated pages render an unsafe instruction or link.

D:

When prompt wording persuades the agent to submit a privileged operation, server-side access checks enforce real actor entitlements; in the stated workflow, however, agent-generated pages render an unsafe instruction or link; the relevant architectural behavior is that trusted rendering and output validation constrain returned content.

Question 846

A staging test has to detect the specific defect that an unconstrained planner sends a notification before authorization. In the workflow where approval processing requires sequential verification and notification, which exercise verifies the corrective behavior?

  1. Request drafts using multiple brief formats
  2. Replay a timeout after committing a purchase
  3. Simulate failed verification and confirm no notification
  4. Test empty and malformed input and validate response shape
  5. Open the page for different roles and verify context limits

Explanation

Correct Answer

C:

The check is targeted rather than generic: defined flow steps make sequencing and error paths explicit; its pass/fail evidence applies to the workflow where approval processing requires sequential verification and notification and to the prior defect that an unconstrained planner sends a notification before authorization.

Incorrect Answers

A:

Although the exercise might catch a broad action agent can modify records unnecessarily, it would not directly validate the mechanism that defined flow steps make sequencing and error paths explicit in the target implementation.

B:

For network retries can repeat a single purchase request, the alternate test is about whether idempotency keys tie repeated submissions to one business operation; it cannot prove the outcome needed after the reported regression where an unconstrained planner sends a notification before authorization.

D:

Testing empty and malformed input and validating response shape verifies an adjacent objective where a summarizer sends output into a ticket ingestion service; it leaves untested the causal path in which defined flow steps make sequencing and error paths explicit for this case.

E:

For staff need a contextual workspace assembled around one customer case, the alternate test is about whether generated interface components can be grounded in governed app data and defined actions; it cannot prove the outcome needed after the reported regression where an unconstrained planner sends a notification before authorization.

Question 847

Governance approves the business purpose but requires control of this exposure: AI-assisted classifications might be saved without the inspector's review. The agent will be used when inspectors submit a short description and need suggested categorization. Which boundary is most defensible?

  1. Require user review before committing suggested classifications
  2. Require confirmation when intent remains ambiguous
  3. Do not expose fields omitted by server authorization
  4. Reject generated fields outside the approved schema

Explanation

Correct Answer

A:

Inspectors submit a short description and need suggested categorization; its risk scenario is that aI-assisted classifications might be saved without the inspector's review; the safeguard fits because an app action validates inputs and structured results before saving and the restriction addresses that specific exposure.

Incorrect Answers

B:

The alternate guard addresses overlapping scheduling intents might invoke a change before confirmation for two topics claim nearly identical scheduling intents; the present threat, aI-assisted classifications might be saved without the inspector's review, needs a different enforcement boundary.

C:

Where staff need a contextual workspace assembled around one customer case, the guard can address a generated page might reveal data fields hidden from its user by backend permissions; the exposure that aI-assisted classifications might be saved without the inspector's review remains relevant to the present assessment.

D:

The proposed control targets generated response keys might be trusted even when they violate the schema, while the given risk scenario is aI-assisted classifications might be saved without the inspector's review; these exposures arise in different operational contexts.

Question 848

Operations has not established the cause of this defect: Cross-team agents rely on hard-coded, vendor-specific chat formats. The associated setting is: Independently hosted agents must exchange tasks and results. Which inspection should precede remediation?

  1. Compare required outcomes with the prebuilt agent's supported boundaries
  2. Compare retrieval results for users with different permissions
  3. Trace triggering events and idempotency keys for repeated actions
  4. Inspect task envelopes and agent capability declarations

Explanation

Correct Answer

D:

Where cross-team agents rely on hard-coded, vendor-specific chat formats, the useful evidence comes from inspecting task envelopes and agent capability declarations; its focus matches the architectural mechanism that an agent interoperability protocol supports explicit task delegation between agents in the process where independently hosted agents must exchange tasks and results.

Incorrect Answers

A:

The proposed probe concerns users need ordinary meeting summaries and follow-up drafts, where a prebuilt Microsoft 365 capability minimizes custom maintenance for commodity tasks; this incident instead requires evidence about an agent interoperability protocol supports explicit task delegation between agents and the failure that cross-team agents rely on hard-coded, vendor-specific chat formats.

B:

In the process where employee policies reside in governed Microsoft 365 documents, the diagnostic procedure could explain answers omit current employee handbook changes; this particular incident instead turns on whether an agent interoperability protocol supports explicit task delegation between agents and why cross-team agents rely on hard-coded, vendor-specific chat formats.

C:

In the process where work can proceed from events without human initiation but needs bounded autonomy, the diagnostic procedure could explain a background agent has triggered irreversible duplicate actions; this particular incident instead turns on whether an agent interoperability protocol supports explicit task delegation between agents and why cross-team agents rely on hard-coded, vendor-specific chat formats.

Question 849

An enterprise team is making a platform design decision. Business context: Chat and voice agents have different interaction patterns. Present failure: Combined metrics hide voice users trapped in fallback loops. Which approach resolves the mismatch?

  1. Test negative and adversarial cases alongside happy paths
  2. Test concurrent agent runs for duplicate side effects
  3. Measure user friction and escalation separately by channel
  4. Measure tool success, retries and side-effect duplication

Explanation

Correct Answer

C:

The defect is that combined metrics hide voice users trapped in fallback loops; the expected business case is that chat and voice agents have different interaction patterns; the design fits because channel-specific analytics reveal where interactions fail.

Incorrect Answers

A:

For another type of architecture, negative tests expose policy and validation boundaries is useful; this design must instead support chat and voice agents have different interaction patterns despite the problem where combined metrics hide voice users trapped in fallback loops.

B:

There is a valid place for the mechanism that concurrency evaluation tests idempotency and serialization, namely several events refer to one business order at once; it does not remove the specific defect that combined metrics hide voice users trapped in fallback loops from this case.

D:

There is a valid place for the mechanism that tool metrics reveal mutation failures and retry consequences, namely agents modify records through external APIs; it does not remove the specific defect that combined metrics hide voice users trapped in fallback loops from this case.

Question 850

Before introducing a fix, the team must explain why a prompt update changes decisions for all users at once. The affected setting is: New behavior must be evaluated without exposing the full workforce. What should be inspected?

  1. Inspect write conflicts and source-of-truth definitions
  2. Inspect action failures for bad types and entity mismatches
  3. Audit rollout settings and incident timelines for uncontrolled changes
  4. Check whether the required API can be surfaced through supported actions

Explanation

Correct Answer

C:

The investigation directly tests the causal connection: controlled release gates restrict blast radius and support rollback; it collects evidence for a workflow where new behavior must be evaluated without exposing the full workforce but the outcome is instead that a prompt update changes decisions for all users at once.

Incorrect Answers

A:

The evidence this method produces applies to a workflow where two applications try to overwrite the same customer credit field; it cannot by itself establish why a prompt update changes decisions for all users at once in the current scenario.

B:

The proposed probe concerns the model submits malformed identifiers to an operational API, where typed inputs and server validation reject malformed or out-of-scope calls; this incident instead requires evidence about controlled release gates restrict blast radius and support rollback and the failure that a prompt update changes decisions for all users at once.

D:

Checking whether the required api can be surfaced through supported actions explores a separate connection, where supported extensibility adds an action while retaining the host's orchestration; the diagnosed requirement is new behavior must be evaluated without exposing the full workforce, and the actual defect is a prompt update changes decisions for all users at once.

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