Microsoft AB-100 (Agentic AI Business Solutions Architect) Practice Test 79

Question 780

The architect wants evidence before altering the deployment. In a workflow where a request crosses Copilot Studio, a flow and the CRM backend, logs cannot reconstruct which API mutation followed an answer. What should the team examine first?

  1. Trace source exception through target state and error queue
  2. Inspect solution layers for unmanaged production changes
  3. Compare request IDs to backend transaction counts
  4. Trace one business request through each dependent component

Explanation

Correct Answer

D:

The investigation directly tests the causal connection: correlation identifiers link the distributed operational path; it collects evidence for a workflow where a request crosses Copilot Studio, a flow and the CRM backend but the outcome is instead that logs cannot reconstruct which API mutation followed an answer.

Incorrect Answers

A:

Tracing source exception through target state and error queue explores a separate connection, where failure-injection tests verify retries and compensation between systems; the diagnosed requirement is a request crosses Copilot Studio, a flow and the CRM backend, and the actual defect is logs cannot reconstruct which API mutation followed an answer.

B:

The evidence this method produces applies to a workflow where production components should not be edited casually; it cannot by itself establish why logs cannot reconstruct which API mutation followed an answer in the current scenario.

C:

Agents modify records through external APIs; its matching diagnostic check is reasonable for successful chat sessions conceal repeated backend writes; the present investigation needs evidence for the different failure where logs cannot reconstruct which API mutation followed an answer.

Question 781

After deployment, stakeholders need an interpretable measure for this use case: Product owners favor easy fixes over high-consequence failures. The pilot showed that a rare data leak risk remains below cosmetic response issues. Which metric captures the needed result?

  1. High-severity agent defects remediated within policy target
  2. Percentage of fixes that pass production effectiveness checks
  3. Unapproved external destinations reached through agent tools
  4. Quality-adjusted cost per approved business result

Explanation

Correct Answer

A:

High-severity agent defects remediated within policy target; this outcome measure belongs to the workflow where product owners favor easy fixes over high-consequence failures; a change in it can help detect whether the defect that a rare data leak risk remains below cosmetic response issues is resolved.

Incorrect Answers

B:

The alternative tracks percentage of fixes that pass production effectiveness checks, arising from the mechanism where outcome verification checks whether the change fixed the user problem; its movement alone says little about the failure that a rare data leak risk remains below cosmetic response issues.

C:

For the workflow where a prompt injection requests data export to an outside address, unapproved external destinations reached through agent tools is an actionable measure; it does not establish whether a rare data leak risk remains below cosmetic response issues in the situation under review.

D:

Quality-adjusted cost per approved business result; that indicator describes a distinct workflow where average accuracy hides failures on complex finance questions; it does not quantify high-severity agent defects remediated within policy target, which is needed for the stated case.

Question 782

A proposed deployment supports this business case: Management attributes a sales lift to Copilot without a comparison. Reviewers also warn that seasonal demand could be misrepresented as attributable AI benefit. Which restriction mitigates that concrete risk?

  1. Honor cross-app role permissions and data minimization
  2. Keep critical numeric calculations in tested services
  3. Separate recommendation privileges from posting permissions
  4. Avoid publishing benefits without attribution and uncertainty bounds

Explanation

Correct Answer

D:

Avoiding the consequence where seasonal demand could be misrepresented as attributable AI benefit requires a control for management attributes a sales lift to Copilot without a comparison; in this architecture, matched comparisons separate solution impact from unrelated shifts, which is the point at which the boundary can be applied.

Incorrect Answers

A:

Where sales and service agents need the same customer and case context, the guard can address cross-application aggregation might expose records outside the user's Dynamics permissions; the exposure that seasonal demand could be misrepresented as attributable AI benefit remains relevant to the present assessment.

B:

For the use case where a business process mixes deterministic calculations and open-ended judgments, the control addresses the threat that probabilistic generation might replace authoritative arithmetic used in financial controls; the specified agent must be protected against seasonal demand could be misrepresented as attributable AI benefit.

C:

For the use case where service, sales and finance jointly decide high-value concessions, the control addresses the threat that an AI recommendation could become an actual posting without an authorized reviewer; the specified agent must be protected against seasonal demand could be misrepresented as attributable AI benefit.

Question 783

One enterprise rollout has 2 separately testable requirements. (1) Business use: The system must detect customer sentiment and named entities at scale. Exposure: A sentiment score might become the only basis for an eligibility or disciplinary decision. (2) Business use: Support documents arrive in several customer languages. Exposure: A translated legal clause might lose meaning without access to the original text. Which 2 safeguards directly satisfy these stated cases? Choose TWO.

  1. Disable write tools and label generated material for review
  2. Require predefined acceptance thresholds before promotion
  3. Treat sentiment as advisory rather than a sole eligibility decision
  4. Store source text and flag low-confidence legal translations
  5. Prevent deployment of unapproved model/data combinations

Explanation

Correct Answers

C:

For the exposure that a sentiment score might become the only basis for an eligibility or disciplinary decision, purpose-built text analysis handles bounded language-analysis tasks explains the protection required when the system must detect customer sentiment and named entities at scale; this does not replace control of a translated legal clause might lose meaning without access to the original text.

D:

For the exposure that a translated legal clause might lose meaning without access to the original text, translation provides standardized text input while preserving original evidence explains the protection required when support documents arrive in several customer languages; this does not replace control of a sentiment score might become the only basis for an eligibility or disciplinary decision.

Incorrect Answers

A:

This security safeguard addresses the unrelated exposure that a text-only drafting assistant might be given unnecessary record-writing powers in a workflow where marketing specialists need tailored descriptions from existing briefings; the question instead names the specific risks a sentiment score might become the only basis for an eligibility or disciplinary decision; a translated legal clause might lose meaning without access to the original text; this choice does not directly mitigate those specific exposures.

B:

This security safeguard addresses the unrelated exposure that a team might promote a tuned model without meeting a critical holdout threshold in a workflow where a candidate appears accurate on its training examples; the question instead names the specific risks a sentiment score might become the only basis for an eligibility or disciplinary decision; a translated legal clause might lose meaning without access to the original text; this choice does not directly mitigate those specific exposures.

E:

This security safeguard addresses the unrelated exposure that a model release might combine training data with an unapproved candidate artifact in a workflow where retraining uses revised labels and a different source snapshot; the question instead names the specific risks a sentiment score might become the only basis for an eligibility or disciplinary decision; a translated legal clause might lose meaning without access to the original text; this choice does not directly mitigate those specific exposures.

Question 784

The rollout supports this workflow: A tool deletes a record following an ambiguous conversation. Leaders want a metric that responds to the reported shortcoming that logs show deletion but not the requesting principal or approval. Which measure best reflects success?

  1. Cross-app exceptions ending in consistent state
  2. Unauthorized retrievals completed in red-team cases
  3. High-risk tool actions with full decision-to-effect trace
  4. Model changes with complete actor and approval records

Explanation

Correct Answer

C:

For the use case where a tool deletes a record following an ambiguous conversation, the indicator high-risk tool actions with full decision-to-effect trace is worth baselining against the earlier issue where logs show deletion but not the requesting principal or approval.

Incorrect Answers

A:

For the workflow where a blocked invoice triggers a hold on related procurement actions, cross-app exceptions ending in consistent state is an actionable measure; it does not establish whether logs show deletion but not the requesting principal or approval in the situation under review.

B:

Although unauthorized retrievals completed in red-team cases is observable, the affected process involves a tool deletes a record following an ambiguous conversation; its acceptance outcome is high-risk tool actions with full decision-to-effect trace, not the alternative measurement.

D:

For the situation where a model is replaced without a traceable change ticket, model changes with complete actor and approval records is a sensible indicator; the problem that logs show deletion but not the requesting principal or approval requires measurement of high-risk tool actions with full decision-to-effect trace.

Question 785

After deployment, stakeholders need an interpretable measure for this use case: A specialized support team uses a nonstandard severity matrix. The pilot showed that default summaries treat critical outages as low priority. Which metric captures the needed result?

  1. Handoffs with complete approved case context
  2. Correct external-to-Dynamics account match rate
  3. Severity classification agreement with human triage
  4. Sales task completion under external API impairment

Explanation

Correct Answer

C:

The observed shortcoming is that default summaries treat critical outages as low priority; measuring severity classification agreement with human triage gives business owners a directly relevant indicator for the process where a specialized support team uses a nonstandard severity matrix.

Incorrect Answers

A:

For the situation where agents need structured case summaries at handoff, handoffs with complete approved case context is a sensible indicator; the problem that default summaries treat critical outages as low priority requires measurement of severity classification agreement with human triage.

B:

The proposed metric concerns two systems share company names but not primary keys, whereas the failure in this case is that default summaries treat critical outages as low priority; the indicator needed here is severity classification agreement with human triage.

D:

External credit service responds slowly during regional outages; its intended measurement is sales task completion under external API impairment; substituting that result here would not capture whether default summaries treat critical outages as low priority has been corrected.

Question 786

Implementation alternatives must be compared for a workflow where an intake task requires several prerequisite fields. The pilot shows this gap: Agent sends incomplete requests when users answer out of order. Which alternative should lead the solution?

  1. Use an agent flow for repeatable ordered business steps
  2. Choose bounded task-specific instructions over open-ended autonomy
  3. Use explicit state tracking for multi-step task completion
  4. Provide source-linked explanations for analytical responses

Explanation

Correct Answer

C:

The architecture must reconcile two facts: an intake task requires several prerequisite fields, yet agent sends incomplete requests when users answer out of order; the selected pattern does so through the mechanism that state tracking captures required inputs before any action.

Incorrect Answers

A:

This design would make sense if approval processing requires sequential verification and notification, particularly where an unconstrained planner sends a notification before authorization; the requested architecture instead needs the dependency that state tracking captures required inputs before any action because agent sends incomplete requests when users answer out of order.

B:

It autonomously contacts a supplier without approval; the alternate design can address that situation because limited intent and tool scope prevent unnecessary actions, whereas this question needs a solution to agent sends incomplete requests when users answer out of order.

D:

For finance asks why a variance was flagged, this other approach relies on the fact that grounded explanatory responses distinguish evidence from hypothesis; the issue here is agent sends incomplete requests when users answer out of order; that mechanism does not establish the required state tracking captures required inputs before any action.

Question 787

Security and business owners want to keep this capability: Executives request AI before defining an operational improvement target. They must also prevent the following: Pilot expansion might occur without anyone accountable for value and risk. Which guardrail should the design incorporate?

  1. Deny unapproved tools regardless of model-generated arguments
  2. Do not scale pilots without decision and benefit ownership
  3. Apply privacy review and model-version rollback before release
  4. Separate recommendation privileges from posting permissions

Explanation

Correct Answer

B:

Avoiding the consequence where pilot expansion might occur without anyone accountable for value and risk requires a control for executives request AI before defining an operational improvement target; in this architecture, cloud adoption planning begins from motivations, outcomes and readiness constraints, which is the point at which the boundary can be applied.

Incorrect Answers

A:

Another process could need this guardrail because planner-generated arguments might invoke tools that the user was never permitted to call; the design being assessed supports executives request AI before defining an operational improvement target and has the explicit exposure that pilot expansion might occur without anyone accountable for value and risk.

C:

The proposed control targets customization might promote a model trained on data without a privacy review, while the given risk scenario is pilot expansion might occur without anyone accountable for value and risk; these exposures arise in different operational contexts.

D:

The safeguard for service, sales and finance jointly decide high-value concessions deals with a different threat: an AI recommendation could become an actual posting without an authorized reviewer; even after it is enforced, the exposure that pilot expansion might occur without anyone accountable for value and risk remains possible in the target workflow where cloud adoption planning begins from motivations, outcomes and readiness constraints.

Question 788

In a business situation where agent handles threshold-based return decisions, a previous build showed this defect: Human testers overlook values just above approval limits. Which reproducible test belongs in the release gate?

  1. Compare controlled tests before and after tool curation
  2. Use an annotated query with a known authoritative paragraph
  3. Replay complex and simple cases through revised thresholds
  4. Generate cases immediately below, at and above the limit

Explanation

Correct Answer

D:

For the required workflow where agent handles threshold-based return decisions, the test needs to reproduce or prevent the failure where human testers overlook values just above approval limits; generating cases immediately below, at and above the limit exercises the path where assisted test generation explores stated boundary conditions.

Incorrect Answers

A:

Comparing controlled tests before and after tool curation verifies an adjacent objective where planner calls the wrong tool among overlapping capabilities; it leaves untested the causal path in which assisted test generation explores stated boundary conditions for this case.

B:

For an assistant responds fluently but cites unrelated support articles, the alternate test is about whether retrieval telemetry distinguishes evidence search from answer generation; it cannot prove the outcome needed after the reported regression where human testers overlook values just above approval limits.

C:

The tested behavior is that stratified evaluation guides thresholds for difficult requests; the necessary evidence here concerns assisted test generation explores stated boundary conditions and the previously observed human testers overlook values just above approval limits.

Question 789

An incident review has one confirmed symptom: Transcription confusion risks wrong account modifications. For the use case where callers provide account numbers and amounts by speech, which targeted diagnostic check would reveal the most pertinent evidence?

  1. Compare source timestamps and principal mapping on connector calls
  2. Inspect transcript confidence and correction requests
  3. Audit approval tokens attached to payment attempts
  4. Inspect match decisions and false-positive entity joins

Explanation

Correct Answer

B:

The reported fault is that transcription confusion risks wrong account modifications in a system where callers provide account numbers and amounts by speech; evidence from inspecting transcript confidence and correction requests can show whether spoken confirmation and verification reduce high-impact recognition errors.

Incorrect Answers

A:

Copilot relies on last quarter's manually imported account report; examining that other path does not determine whether spoken confirmation and verification reduce high-impact recognition errors, which is what distinguishes the issue in the stated use case.

C:

In the separate scenario where agent can research options but cannot approve a vendor payment, the same investigation could explain a background process posts expenses beyond its delegated scope; it does not isolate the reported cause in a system where callers provide account numbers and amounts by speech.

D:

The evidence this method produces applies to a workflow where two systems share company names but not primary keys; it cannot by itself establish why transcription confusion risks wrong account modifications in the current scenario.

Question 790

Release reviewers will not accept a generic success message. The business scenario is that planner calls the wrong tool among overlapping capabilities, and the historical failure is: The team proposes adding another similar tool. Which test is appropriately targeted?

  1. Submit a malicious mislabeled training record and inspect review
  2. Classify a mixed set of complaints and check routing consistency
  3. Compare controlled tests before and after tool curation
  4. Reproduce a sanitized failure and confirm the new regression catches it

Explanation

Correct Answer

C:

Comparing controlled tests before and after tool curation is aligned with the failure that the team proposes adding another similar tool; its causal relevance is that clear capabilities and restricted sets improve selection reliability for the situation where planner calls the wrong tool among overlapping capabilities.

Incorrect Answers

A:

The verification concerning business analysts can submit examples but not change production behavior checks separated data-curation and release permissions preserve integrity against unvetted labeled data poisons a retrained classifier; its result says little about the observed issue where the team proposes adding another similar tool.

B:

The tested behavior is that structured defect taxonomy assigns the right remediation owner; the necessary evidence here concerns clear capabilities and restricted sets improve selection reliability and the previously observed the team proposes adding another similar tool.

D:

A pass on that verification would speak to support tickets reveal an unusual multistep conversation failure; it would not show that clear capabilities and restricted sets improve selection reliability, which is needed to avoid the observed defect the team proposes adding another similar tool.

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