Microsoft AB-100 (Agentic AI Business Solutions Architect) Practice Test 48

Question 470

During integration testing, a trained model remembers obsolete guidance after source changes. The associated business context is: Answers depend on policies revised every week. Which architectural remedy is the strongest match?

  1. Extend an existing Dynamics or Copilot capability for a small gap
  2. Use a custom model only when general models fail a defined task
  3. Use Dataverse knowledge for structured business-record questions
  4. Use retrieval augmentation before tuning for changing facts

Explanation

Correct Answer

D:

Because retrieval separates volatile knowledge from model parameters, this approach corrects the reported breakdown where a trained model remembers obsolete guidance after source changes; the affected situation is that answers depend on policies revised every week.

Incorrect Answers

A:

For current users need one governed custom action inside their workflow, this other approach relies on the fact that extension retains familiar user experience and adds bounded custom behavior; the issue here is a trained model remembers obsolete guidance after source changes; that mechanism does not establish the required retrieval separates volatile knowledge from model parameters.

B:

The technical behavior of this alternative is that a measured capability gap, appropriate dataset and MLOps plan justify customization; the user case being assessed is answers depend on policies revised every week with a different failure: a trained model remembers obsolete guidance after source changes.

C:

Dataverse grounding exposes appropriate structured records to the agent; its main purpose is users ask about governed CRM records rather than prose manuals, not correction of the reported defect where a trained model remembers obsolete guidance after source changes in the target process.

Question 471

A staging test has to detect the specific defect that the automated suite approves the same defect it should detect. In the workflow where Copilot writes a test expecting a plausible but wrong answer, which exercise verifies the corrective behavior?

  1. Challenge generated cases using deliberately incorrect model output
  2. Seed paraphrased failures and verify useful grouping
  3. Promote to staging and confirm correct connection targets
  4. Replay complex and simple cases through revised thresholds

Explanation

Correct Answer

A:

The scenario is that Copilot writes a test expecting a plausible but wrong answer; the defect to rule out is that the automated suite approves the same defect it should detect; testing via challenging generated cases using deliberately incorrect model output challenges the design dependency where independent acceptance references prevent self-confirming evaluation.

Incorrect Answers

B:

Although the exercise might catch analysts overlook a widespread intent failure among unique phrasings, it would not directly validate the mechanism that independent acceptance references prevent self-confirming evaluation in the target implementation.

C:

For a flow moved to production still points to a test endpoint, the alternate test is about whether environment configuration separates deployable logic from endpoint bindings; it cannot prove the outcome needed after the reported regression where the automated suite approves the same defect it should detect.

D:

A pass on that verification would speak to average accuracy hides failures on complex finance questions; it would not show that independent acceptance references prevent self-confirming evaluation, which is needed to avoid the observed defect the automated suite approves the same defect it should detect.

Question 472

A product owner must measure whether an AI-enabled change actually improves this process: Agent supports refunds, order status and generic FAQs. The earlier problem was: The test suite spends most cases on low-risk FAQs. Which outcome indicator is most relevant?

  1. Security contract conditions represented in vetted tests
  2. Multi-turn tasks resolved without entity confusion
  3. Candidate releases satisfying every critical acceptance metric
  4. Critical journeys passing mandatory acceptance tests

Explanation

Correct Answer

D:

The observed shortcoming is that the test suite spends most cases on low-risk FAQs; measuring critical journeys passing mandatory acceptance tests gives business owners a directly relevant indicator for the process where agent supports refunds, order status and generic FAQs.

Incorrect Answers

A:

Security contract conditions represented in vetted tests; that indicator describes a distinct workflow where agents can call APIs with customer IDs and amount values; it does not quantify critical journeys passing mandatory acceptance tests, which is needed for the stated case.

B:

A change in multi-turn tasks resolved without entity confusion is informative for agent must remember order context while asking for missing details; it does not establish whether the current process achieves critical journeys passing mandatory acceptance tests.

C:

Candidate releases satisfying every critical acceptance metric; that indicator describes a distinct workflow where teams promote a custom model because it feels better; it does not quantify critical journeys passing mandatory acceptance tests, which is needed for the stated case.

Question 473

An enterprise team is making a platform design decision. Business context: A customer handoff depends on shared record and term mappings. Present failure: Independent releases cause contradictory customer statuses. Which approach resolves the mismatch?

  1. Preserve customer communication approvals across solution releases
  2. Test role-dependent visibility through Dynamics cross-app flows
  3. Coordinate Sales Copilot and Service Copilot customization changes
  4. Test Sales-to-Service customer context and case handoff

Explanation

Correct Answer

C:

The architecture must reconcile two facts: a customer handoff depends on shared record and term mappings, yet independent releases cause contradictory customer statuses; the selected pattern does so through the mechanism that cross-app release plans validate shared contracts and timing.

Incorrect Answers

A:

There is a valid place for the mechanism that business approval checks remain independent of agent version, namely service agents draft customer messages but cannot send without review; it does not remove the specific defect that independent releases cause contradictory customer statuses from this case.

B:

This design would make sense if sales and Finance roles have different record entitlements, particularly where a Copilot suggestion includes restricted invoice details; the requested architecture instead needs the dependency that cross-app release plans validate shared contracts and timing because independent releases cause contradictory customer statuses.

D:

In the case where a won opportunity should create a properly owned onboarding case, the alternate design depends on end-to-end tests verify identity, mapping and assignment across apps; it does not correct the target failure, independent releases cause contradictory customer statuses.

Question 474

Risk reviewers need a control, not merely a monitoring dashboard. The use case is: A Copilot prompt feeds records into downstream automation. Potential harm: A malformed JSON response might reach a write endpoint and corrupt records. Which protection fits?

  1. Route low-confidence tax values for human validation
  2. Reject invalid outputs before they reach mutation endpoints
  3. Avoid privileged shared accounts for user-specific retrieval
  4. Block promotion when authorization or data-leak tests fail

Explanation

Correct Answer

B:

When a malformed JSON response might reach a write endpoint and corrupt records, the use case a Copilot prompt feeds records into downstream automation requires a guardrail aligned with the behavior that schema validation catches responses that violate integration contracts.

Incorrect Answers

A:

That protection is tied to the mechanism that stratified field metrics protect high-impact values; it does not prevent the distinct risk where a malformed JSON response might reach a write endpoint and corrupt records in the system under review.

C:

That protection is tied to the mechanism that delegated authorization evaluates each caller's effective permissions; it does not prevent the distinct risk where a malformed JSON response might reach a write endpoint and corrupt records in the system under review.

D:

The proposed control targets a new Foundry workflow might fail safety evaluation but still be published, while the given risk scenario is a malformed JSON response might reach a write endpoint and corrupt records; these exposures arise in different operational contexts.

Question 475

The architect wants evidence before altering the deployment. In a workflow where solution import succeeds but users still see the previous behavior, deployment checks stop at package import success. What should the team examine first?

  1. Audit tokens and connection bindings for each stage
  2. Compare outcomes by locale, channel and agent version
  3. Compare target environment version and active channel metadata
  4. Inspect solution dependencies and common entity schema

Explanation

Correct Answer

C:

The reported fault is that deployment checks stop at package import success in a system where solution import succeeds but users still see the previous behavior; evidence from comparing target environment version and active channel metadata can show whether post-import validation includes publishing and channel behavior.

Incorrect Answers

A:

That probe evaluates environment-scoped identities isolate access and simplify incident containment in a workflow where test automation must not reach live customer records; the reported failure concerns the path for which post-import validation includes publishing and channel behavior instead.

B:

Global averages hide a regional escalation spike; examining that other path does not determine whether post-import validation includes publishing and channel behavior, which is what distinguishes the issue in the stated use case.

D:

In the separate scenario where a customer handoff depends on shared record and term mappings, the same investigation could explain independent releases cause contradictory customer statuses; it does not isolate the reported cause in a system where solution import succeeds but users still see the previous behavior.

Question 476

A release must prove that rate limiting causes intermittent failures will not recur. The application is used in this situation: A regional launch multiplies concurrent agent requests. Which acceptance test most directly checks that failure mode?

  1. Replay complex and simple cases through revised thresholds
  2. Run matched intent tests in chat and voice
  3. Use an annotated query with a known authoritative paragraph
  4. Simulate expected launch concurrency and validate headroom

Explanation

Correct Answer

D:

Simulating expected launch concurrency and validating headroom is aligned with the failure that rate limiting causes intermittent failures; its causal relevance is that capacity metrics and throttling errors show saturation risk for the situation where a regional launch multiplies concurrent agent requests.

Incorrect Answers

A:

This check exercises a different system behavior: stratified evaluation guides thresholds for difficult requests; a pass there cannot establish that rate limiting causes intermittent failures no longer occurs in the required workflow.

B:

Running matched intent tests in chat and voice verifies an adjacent objective where chat and voice agents have different interaction patterns; it leaves untested the causal path in which capacity metrics and throttling errors show saturation risk for this case.

C:

In a solution where an assistant responds fluently but cites unrelated support articles, using an annotated query with a known authoritative paragraph could catch the defect that surface quality scores appear healthy despite wrong documents; the current release gate must challenge the failure where rate limiting causes intermittent failures.

Question 477

During integration testing, direct replacement removes the dependable baseline. The associated business context is: A tuned model claims higher accuracy but adds latency. Which architectural remedy is the strongest match?

  1. Preserve reproducible evaluation evidence for major model changes
  2. Promote grounding datasets through quality gates
  3. Stage candidate models behind champion-challenger evaluation
  4. Assess robustness under realistic drift and noisy input
  5. Preserve model-compatible preprocessing and postprocessing contracts

Explanation

Correct Answer

C:

The defect is that direct replacement removes the dependable baseline; the expected business case is that a tuned model claims higher accuracy but adds latency; the design fits because candidate comparison evaluates quality, latency and safety under matched load.

Incorrect Answers

A:

There is a valid place for the mechanism that versioned evaluation sets, model IDs and thresholds document release claims, namely an updated agent performs differently months after release; it does not remove the specific defect that direct replacement removes the dependable baseline from this case.

B:

Ingestion gates check classification, freshness and ownership; its main purpose is new documents enter a production source automatically, not correction of the reported defect where direct replacement removes the dependable baseline in the target process.

D:

For production documents differ from clean training samples, this other approach relies on the fact that stress tests measure sensitivity to expected input variation; the issue here is direct replacement removes the dependable baseline; that mechanism does not establish the required candidate comparison evaluates quality, latency and safety under matched load.

E:

In the case where a retraining release changes the tokenizer and label order, the alternate design depends on bundled preprocessing, model and label maps maintain contract consistency; it does not correct the target failure, direct replacement removes the dependable baseline.

Question 478

A product owner must measure whether an AI-enabled change actually improves this process: Analysts compare unstructured reports but must retain human interpretation. The earlier problem was: Narratives present speculation as if it were validated insight. Which outcome indicator is most relevant?

  1. Authorized structured-record answer accuracy
  2. Autonomous tasks completed without unauthorized changes
  3. Successful prompt-injection override rate
  4. Percentage of material analytical claims with checked citations

Explanation

Correct Answer

D:

Percentage of material analytical claims with checked citations; this outcome measure belongs to the workflow where analysts compare unstructured reports but must retain human interpretation; a change in it can help detect whether the defect that narratives present speculation as if it were validated insight is resolved.

Incorrect Answers

A:

The alternative tracks authorized structured-record answer accuracy, arising from the mechanism where Dataverse grounding exposes appropriate structured records to the agent; its movement alone says little about the failure that narratives present speculation as if it were validated insight.

B:

For the workflow where work can proceed from events without human initiation but needs bounded autonomy, autonomous tasks completed without unauthorized changes is an actionable measure; it does not establish whether narratives present speculation as if it were validated insight in the situation under review.

C:

The premise of this indicator is trust boundaries treat source text as evidence rather than authority; the architecture being evaluated instead serves analysts compare unstructured reports but must retain human interpretation, and its success is measured by percentage of material analytical claims with checked citations.

Question 479

Business leaders reject an activity-only metric for this process: Operations must distinguish healthy traffic from quality incidents. They need to detect the issue where teams discover failures through complaints several days later. Which performance indicator fits?

  1. Unauthorized action rate in adversarial evaluations
  2. Complete distributed traces per sampled transaction
  3. Correct source passage retrieved within top K
  4. User-impacting incidents detected within response target

Explanation

Correct Answer

D:

User-impacting incidents detected within response target; this outcome measure belongs to the workflow where operations must distinguish healthy traffic from quality incidents; a change in it can help detect whether the defect that teams discover failures through complaints several days later is resolved.

Incorrect Answers

A:

The premise of this indicator is negative tests expose policy and validation boundaries; the architecture being evaluated instead serves operations must distinguish healthy traffic from quality incidents, and its success is measured by user-impacting incidents detected within response target.

B:

The proposed metric concerns a request crosses Copilot Studio, a flow and the CRM backend, whereas the failure in this case is that teams discover failures through complaints several days later; the indicator needed here is user-impacting incidents detected within response target.

C:

The alternative tracks correct source passage retrieved within top K, arising from the mechanism where retrieval telemetry distinguishes evidence search from answer generation; its movement alone says little about the failure that teams discover failures through complaints several days later.

Question 480

Support engineers report this failure: Former owners leave behind accessible service accounts. The business context is: Unused agents retain tokens and active integrations. What evidence should be collected to isolate the cause?

  1. Trace token acquisition and authorization results by environment
  2. Compare structured output and errors across candidate cohorts
  3. Inspect solution layers for unmanaged production changes
  4. Compare inactive agents with still-valid tool grants

Explanation

Correct Answer

D:

Former owners leave behind accessible service accounts; to attribute the defect rather than guess at a fix, comparing inactive agents with still-valid tool grants examines the dependency where decommissioning revokes credentials, connectors and publishing rights.

Incorrect Answers

A:

An imported agent can authenticate in dev but not production; its matching diagnostic check is reasonable for users receive permission errors after successful package import; the present investigation needs evidence for the different failure where former owners leave behind accessible service accounts.

B:

A model upgrade breaks downstream response formatting; examining that other path does not determine whether decommissioning revokes credentials, connectors and publishing rights, which is what distinguishes the issue in the stated use case.

C:

The proposed inspection is associated with production components should not be edited casually, where managed solution deployment supports controlled downstream customization boundaries; it would not expose the failure mode that former owners leave behind accessible service accounts with the required specificity.

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