Microsoft AB-100 (Agentic AI Business Solutions Architect) Practice Test 62

Question 610

A test plan focuses on a previously missed failure: Agent-generated language promises the wrong financial remedy. It occurred when refund, credit, reimbursement and adjustment have different accounting impacts. Which check directly validates that behavior?

  1. Test duplicate legal names with different external IDs
  2. Change a permitted credit field and verify updated authorized retrieval
  3. Delay the external API and verify useful partial handling
  4. Generate stressful and adversarial user examples
  5. Test similar user requests that imply different posting types
  6. Query as multiple roles and confirm only authorized fields appear

Explanation

Correct Answer

E:

Agent-generated language promises the wrong financial remedy; a release test must establish that this no longer occurs; testing similar user requests that imply different posting types measures the specific behavior for which controlled definitions tie each term to approved policies and authorities.

Incorrect Answers

A:

The verification concerning two systems share company names but not primary keys checks a mapping service resolves stable identities before enrichment against Copilot attaches one customer's risk note to another account; its result says little about the observed issue where agent-generated language promises the wrong financial remedy.

B:

For sellers need current credit exposure held outside CRM, the alternate test is about whether a supported connector can retrieve external information with explicit identity mapping; it cannot prove the outcome needed after the reported regression where agent-generated language promises the wrong financial remedy.

C:

This check exercises a different system behavior: bounded timeouts and graceful fallback preserve essential user work; a pass there cannot establish that agent-generated language promises the wrong financial remedy no longer occurs in the required workflow.

D:

A pass on that verification would speak to compliance letters must preserve approved language and disclaimers; it would not show that controlled definitions tie each term to approved policies and authorities, which is needed to avoid the observed defect agent-generated language promises the wrong financial remedy.

F:

Querying as multiple roles and confirming only authorized fields appear verifies an adjacent objective where sales copilots should not expose sensitive margin assumptions; it leaves untested the causal path in which controlled definitions tie each term to approved policies and authorities for this case.

Question 611

An enterprise team is making a platform design decision. Business context: Competitive requirements depend on unique multi-model control. Present failure: A boxed product cannot express the execution and ownership rules. Which approach resolves the mismatch?

  1. Constrain agent tools by explicit allowlists and permissions
  2. Build a bespoke agent when differentiated orchestration is essential
  3. Require approval before irreversible agent actions
  4. Buy a supported managed capability for standard requirements

Explanation

Correct Answer

B:

Without the intended behavior, a boxed product cannot express the execution and ownership rules; the proposed architecture establishes the specific design principle that custom engineering trades more control for more operating responsibility for the process where competitive requirements depend on unique multi-model control.

Incorrect Answers

A:

There is a valid place for the mechanism that allowlisting and server-enforced identity restrict actual executable capability, namely agent instructions mention many APIs but only two are permitted; it does not remove the specific defect that a boxed product cannot express the execution and ownership rules from this case.

C:

For the agent may refund large amounts or deactivate accounts, this other approach relies on the fact that confirmation and human approval separate advice from execution authority; the issue here is a boxed product cannot express the execution and ownership rules; that mechanism does not establish the required custom engineering trades more control for more operating responsibility.

D:

For another type of architecture, managed functionality shifts routine platform operations to the service is useful; this design must instead support competitive requirements depend on unique multi-model control despite the problem where a boxed product cannot express the execution and ownership rules.

Question 612

After deployment, stakeholders need an interpretable measure for this use case: Unused agents retain tokens and active integrations. The pilot showed that former owners leave behind accessible service accounts. Which metric captures the needed result?

  1. AI-related incidents during financial-close windows
  2. Retired agents with no active unintended access
  3. Unauthorized finance postings in release tests
  4. Releases conforming to risk-tier controls
  5. Authorized tool calls succeeding after promotion

Explanation

Correct Answer

B:

The observed shortcoming is that former owners leave behind accessible service accounts; measuring retired agents with no active unintended access gives business owners a directly relevant indicator for the process where unused agents retain tokens and active integrations.

Incorrect Answers

A:

That scorecard concerns the activity in which invoice posting agents affect critical close processes and its result of aI-related incidents during financial-close windows; the reported shortcoming here is former owners leave behind accessible service accounts.

C:

For the situation where agents suggest payment adjustments and inventory holds, unauthorized finance postings in release tests is a sensible indicator; the problem that former owners leave behind accessible service accounts requires measurement of retired agents with no active unintended access.

D:

The alternative tracks releases conforming to risk-tier controls, arising from the mechanism where zoned governance aligns controls with impact and data sensitivity; its movement alone says little about the failure that former owners leave behind accessible service accounts.

E:

Authorized tool calls succeeding after promotion; that indicator describes a distinct workflow where an imported agent can authenticate in dev but not production; it does not quantify retired agents with no active unintended access, which is needed for the stated case.

Question 613

The architect wants evidence before altering the deployment. In a workflow where analysts compare unstructured reports but must retain human interpretation, narratives present speculation as if it were validated insight. What should the team examine first?

  1. Inspect citations and human review notes for unsupported conclusions
  2. Inspect traces for instruction-following from retrieved content
  3. Inspect entity mappings and security context at retrieval
  4. Analyze confusion matrix at neighboring categories

Explanation

Correct Answer

A:

Narratives present speculation as if it were validated insight; to attribute the defect rather than guess at a fix, inspecting citations and human review notes for unsupported conclusions examines the dependency where retrieval supplies supporting evidence while analysts own analytical decisions.

Incorrect Answers

B:

Third-party documents include sentences telling the model to ignore rules; its matching diagnostic check is reasonable for answers adopt an instruction embedded in a source document; the present investigation needs evidence for the different failure where narratives present speculation as if it were validated insight.

C:

The proposed inspection is associated with users ask about governed CRM records rather than prose manuals, where Dataverse grounding exposes appropriate structured records to the agent; it would not expose the failure mode that narratives present speculation as if it were validated insight with the required specificity.

D:

For the failure that a generic instruction cannot express subtle category boundaries, analyzing confusion matrix at neighboring categories is relevant; this question instead reports that narratives present speculation as if it were validated insight, so different evidence is needed.

Question 614

A solution architecture is being reviewed for this business situation: Benchmarks cover only cooperative user wording. The observed defect is that a malicious instruction triggers a privileged tool. Which architectural decision addresses the underlying requirement?

  1. Test negative and adversarial cases alongside happy paths
  2. Automate repeatable evaluation gates before publishing
  3. Use production feedback to propose targeted regression tests
  4. Test concurrent agent runs for duplicate side effects

Explanation

Correct Answer

A:

The defect is that a malicious instruction triggers a privileged tool; the expected business case is that benchmarks cover only cooperative user wording; the design fits because negative tests expose policy and validation boundaries.

Incorrect Answers

B:

Makers validate only a few manual chats; the alternate design can address that situation because stable automated suites compare response quality across versions, whereas this question needs a solution to a malicious instruction triggers a privileged tool.

C:

Feedback-derived examples convert incidents into lasting safeguards; its main purpose is support tickets reveal an unusual multistep conversation failure, not correction of the reported defect where a malicious instruction triggers a privileged tool in the target process.

D:

In the case where several events refer to one business order at once, the alternate design depends on concurrency evaluation tests idempotency and serialization; it does not correct the target failure, a malicious instruction triggers a privileged tool.

Question 615

Before introducing a fix, the team must explain why chat responses mix replacement policies between product lines. The affected setting is: Customer-facing brands have several names for the same product. What should be inspected?

  1. Compare source timestamps and principal mapping on connector calls
  2. Capture response schemas and returned sensitive columns
  3. Trace product resolution to the canonical product identifier
  4. Compare exact-match misses and semantic recall errors

Explanation

Correct Answer

C:

Chat responses mix replacement policies between product lines; to attribute the defect rather than guess at a fix, tracing product resolution to the canonical product identifier examines the dependency where curated alias-to-product mapping grounds interpretations in the product catalogue.

Incorrect Answers

A:

In the process where sellers need current credit exposure held outside CRM, the diagnostic procedure could explain Copilot relies on last quarter's manually imported account report; this particular incident instead turns on whether curated alias-to-product mapping grounds interpretations in the product catalogue and why chat responses mix replacement policies between product lines.

B:

Sales copilots should not expose sensitive margin assumptions; its matching diagnostic check is reasonable for a connector returns the full ERP payload including protected fields; the present investigation needs evidence for the different failure where chat responses mix replacement policies between product lines.

D:

For the failure that pure semantic search misses precise part identifiers, comparing exact-match misses and semantic recall errors is relevant; this question instead reports that chat responses mix replacement policies between product lines, so different evidence is needed.

Question 616

During integration testing, agent invokes the wrong booking action for a reschedule. The associated business context is: Two topics claim nearly identical scheduling intents. Which architectural remedy is the strongest match?

  1. Disambiguate overlapping topic descriptions and triggers
  2. Design a fallback topic that escalates genuine nonmatches
  3. Invoke an AI-backed action from a canvas app with typed inputs
  4. Use an agent flow for repeatable ordered business steps

Explanation

Correct Answer

A:

Two topics claim nearly identical scheduling intents; its failing behavior is that agent invokes the wrong booking action for a reschedule; applying the underlying design principle—distinct descriptions and inputs improve topic selection—connects the requirement to an implementable control.

Incorrect Answers

B:

There is a valid place for the mechanism that fallback handles unmatched intent and can ask for clarification or escalate, namely users use novel requests outside the supported topic inventory; it does not remove the specific defect that agent invokes the wrong booking action for a reschedule from this case.

C:

Free-form AI output is written to the wrong Dataverse field; the alternate design can address that situation because an app action validates inputs and structured results before saving, whereas this question needs a solution to agent invokes the wrong booking action for a reschedule.

D:

For another type of architecture, defined flow steps make sequencing and error paths explicit is useful; this design must instead support two topics claim nearly identical scheduling intents despite the problem where agent invokes the wrong booking action for a reschedule.

Question 617

A proposed deployment supports this business case: A Copilot Studio agent can read corporate data and use public messaging. Reviewers also warn that a combination of connectors might transfer protected data to public messaging. Which restriction mitigates that concrete risk?

  1. Block release when production variables or connections are unresolved
  2. Limit production edit roles and monitor drift
  3. Roll back or withhold publication if acceptance checks fail
  4. Deny disallowed combinations at environment policy level

Explanation

Correct Answer

D:

In the scenario where a Copilot Studio agent can read corporate data and use public messaging, the explicit risk is that a combination of connectors might transfer protected data to public messaging; enforcement must follow the mechanism by which Power Platform data policies constrain connector use and movement pathways.

Incorrect Answers

A:

Another process could need this guardrail because imported actions might retain staging endpoints after production deployment; the design being assessed supports a Copilot Studio agent can read corporate data and use public messaging and has the explicit exposure that a combination of connectors might transfer protected data to public messaging.

B:

The alternate boundary mitigates direct production customization might diverge from the approved solution, which is relevant to production components should not be edited casually; it leaves the current workflow vulnerable to a combination of connectors might transfer protected data to public messaging.

C:

For the use case where solution import succeeds but users still see the previous behavior, the control addresses the threat that an imported but unpublished agent might cause users to see stale behavior; the specified agent must be protected against a combination of connectors might transfer protected data to public messaging.

Question 618

During integration testing, a custom replacement duplicates the host experience unnecessarily. The associated business context is: Users need familiar Copilot behavior plus an internal status lookup. Which architectural remedy is the strongest match?

  1. Use Microsoft 365 Copilot connectors for external knowledge
  2. Extend a prebuilt agent with an approved business action
  3. Build a bespoke agent when differentiated orchestration is essential
  4. Replace an unsuitable prebuilt design with a custom-engine agent

Explanation

Correct Answer

B:

Users need familiar Copilot behavior plus an internal status lookup; its failing behavior is that a custom replacement duplicates the host experience unnecessarily; applying the underlying design principle—supported extensibility adds an action while retaining the host's orchestration—connects the requirement to an implementable control.

Incorrect Answers

A:

There is a valid place for the mechanism that connectors expose external enterprise content to supported Microsoft search/Copilot experiences, namely teams want approved external records searchable in Copilot; it does not remove the specific defect that a custom replacement duplicates the host experience unnecessarily from this case.

C:

For another type of architecture, custom engineering trades more control for more operating responsibility is useful; this design must instead support users need familiar Copilot behavior plus an internal status lookup despite the problem where a custom replacement duplicates the host experience unnecessarily.

D:

When a workflow needs proprietary scheduling, model choice and runtime control, owning the engine provides control over models, routing and execution; in the stated workflow, however, a custom replacement duplicates the host experience unnecessarily; the relevant architectural behavior is that supported extensibility adds an action while retaining the host's orchestration.

Question 619

A risk analysis highlights a threat in this workflow: A blocked invoice triggers a hold on related procurement actions. Exposure: A procurement approval might proceed before a finance hold is resolved. Which enforceable safeguard addresses it at the relevant boundary?

  1. Do not approve procurement while financial hold status is unknown
  2. Restrict production to trusted vetted model versions
  3. Avoid recording personal identifiers as correlation keys
  4. Rotate compromised keys and enforce least-privilege callers

Explanation

Correct Answer

A:

A blocked invoice triggers a hold on related procurement actions; its risk scenario is that a procurement approval might proceed before a finance hold is resolved; the safeguard fits because failure-injection tests verify retries and compensation between systems and the restriction addresses that specific exposure.

Incorrect Answers

B:

Another process could need this guardrail because an attacker might swap a model artifact in the deployment pipeline; the design being assessed supports a blocked invoice triggers a hold on related procurement actions and has the explicit exposure that a procurement approval might proceed before a finance hold is resolved.

C:

Where a request crosses Copilot Studio, a flow and the CRM backend, the guard can address correlation identifiers might embed patient or customer details as stable identifiers; the exposure that a procurement approval might proceed before a finance hold is resolved remains relevant to the present assessment.

D:

The alternate guard addresses a leaked endpoint credential might permit inference from an untrusted client for a sensitive model endpoint is reachable using a broadly shared key; the present threat, a procurement approval might proceed before a finance hold is resolved, needs a different enforcement boundary.

Question 620

A quality engineer is writing a regression case after the incident where recommendations disappear in conversation history. The underlying business scenario is: Operations teams review multiple proposed next actions daily. What should the test exercise?

  1. Create, accept and reject a suggestion and inspect its lifecycle
  2. Test edge cases with conflicting documents and incomplete facts
  3. Change page layout in a test environment and observe recovery
  4. Pilot each intended host rather than assuming parity

Explanation

Correct Answer

A:

The scenario is that operations teams review multiple proposed next actions daily; the defect to rule out is that recommendations disappear in conversation history; testing via create,ing accept and rejecting a suggestion and inspecting its lifecycle challenges the design dependency where a persistent agent activity feed supports review, assignment and disposition.

Incorrect Answers

B:

In a solution where procurement exceptions require comparing several policies and records, testing edge cases with conflicting documents and incomplete facts could catch the defect that a simple scripted topic cannot handle legitimate variants; the current release gate must challenge the failure where recommendations disappear in conversation history.

C:

In a solution where a legacy portal has no integration API for a clerical workflow, changing page layout in a test environment and observing recovery could catch the defect that teams propose browser clicks despite an available stable API; the current release gate must challenge the failure where recommendations disappear in conversation history.

D:

For agent experience must appear in selected Microsoft 365 contexts, the alternate test is about whether host-specific support, packaging and admin approval shape deployment; it cannot prove the outcome needed after the reported regression where recommendations disappear in conversation history.

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