Microsoft AB-100 (Agentic AI Business Solutions Architect) Practice Test 6

Question 50

A pilot scorecard needs evidence of business improvement rather than only invocation counts. The use case is: A user asks the agent to read an unrelated executive record. Known defect: The model may try a tool outside the user's rights. Which KPI should be selected?

  1. Unauthorized write operations executed by read-only agents
  2. Safe failure responses under dependency loss
  3. Unauthorized retrievals completed in red-team cases
  4. Unapproved external destinations reached through agent tools

Explanation

Correct Answer

C:

For the business use where a user asks the agent to read an unrelated executive record, the result to quantify is unauthorized retrievals completed in red-team cases; it connects measurement to the mechanism where authorization evaluations verify server-side deny behavior, not merely to agent traffic.

Incorrect Answers

A:

For the situation where an agent should summarize account data but not alter it, unauthorized write operations executed by read-only agents is a sensible indicator; the problem that the model may try a tool outside the user's rights requires measurement of unauthorized retrievals completed in red-team cases.

B:

The premise of this indicator is failure injection checks controlled degradation and abstention; the architecture being evaluated instead serves a user asks the agent to read an unrelated executive record, and its success is measured by unauthorized retrievals completed in red-team cases.

D:

A change in unapproved external destinations reached through agent tools is informative for a prompt injection requests data export to an outside address; it does not establish whether the current process achieves unauthorized retrievals completed in red-team cases.

Question 51

A quality engineer is writing a regression case after the incident where answers omit current employee handbook changes. The underlying business scenario is: Employee policies reside in governed Microsoft 365 documents. What should the test exercise?

  1. Demonstrate a custom multi-step workflow and failure recovery
  2. Add a confidential draft and verify it is excluded for unauthorized users
  3. Install the agent for a pilot audience and test supported experiences
  4. Ask about a revised policy and verify an allowed citation

Explanation

Correct Answer

D:

The historical failure is that answers omit current employee handbook changes; for the setting where employee policies reside in governed Microsoft 365 documents, asking about a revised policy and verifying an allowed citation verifies the dependency where supported SharePoint grounding retrieves organizational content within access controls.

Incorrect Answers

A:

The tested behavior is that custom-engine architecture controls runtime, model calls and integrations; the necessary evidence here concerns supported SharePoint grounding retrieves organizational content within access controls and the previously observed answers omit current employee handbook changes.

B:

A pass on that verification would speak to search spans internal manuals with different owners and permissions; it would not show that supported SharePoint grounding retrieves organizational content within access controls, which is needed to avoid the observed defect answers omit current employee handbook changes.

C:

Installing the agent for a pilot audience and testing supported experiences verifies an adjacent objective where staff need a role-specific conversation within Copilot and Teams; it leaves untested the causal path in which supported SharePoint grounding retrieves organizational content within access controls for this case.

Question 52

An enterprise team is making a platform design decision. Business context: Agents suggest payment adjustments and inventory holds. Present failure: A new connector allows a service identity to post directly. Which approach resolves the mismatch?

  1. Test Finance-to-Supply-Chain exception handling after partial failure
  2. Test posting permissions and approvals after agent updates
  3. Validate finance and operations virtual-entity mappings after change
  4. Pin approved models and dependencies in a reviewed registry

Explanation

Correct Answer

B:

Without the intended behavior, a new connector allows a service identity to post directly; the proposed architecture establishes the specific design principle that authorization regression gates verify posting authority remains constrained for the process where agents suggest payment adjustments and inventory holds.

Incorrect Answers

A:

This design would make sense if a blocked invoice triggers a hold on related procurement actions, particularly where a network fault leaves the two apps inconsistent; the requested architecture instead needs the dependency that authorization regression gates verify posting authority remains constrained because a new connector allows a service identity to post directly.

C:

The alternative addresses an update changes a field mapping and grounded answers become stale by a design in which integration tests protect virtual-entity schema and source semantics; it does not supply the missing relationship for agents suggest payment adjustments and inventory holds, where a new connector allows a service identity to post directly.

D:

Artifact provenance and integrity checks guard model supply chain; its main purpose is deployment automation pulls model artifacts from unverified origins, not correction of the reported defect where a new connector allows a service identity to post directly in the target process.

Question 53

An investment review asks what measurable result would justify continuing an AI solution where invoice holds must trigger exception handling without polling. The original issue is: Delayed reconciliation hides material fulfillment blockers. Which indicator should be tracked?

  1. Error-rate delta between canary and control cohorts
  2. Time from source exception to downstream acknowledgement
  3. Cross-app case-state agreement rate
  4. Number of unannounced breaking data-contract changes

Explanation

Correct Answer

B:

The observed shortcoming is that delayed reconciliation hides material fulfillment blockers; measuring time from source exception to downstream acknowledgement gives business owners a directly relevant indicator for the process where invoice holds must trigger exception handling without polling.

Incorrect Answers

A:

New behavior must be evaluated without exposing the full workforce; its intended measurement is error-rate delta between canary and control cohorts; substituting that result here would not capture whether delayed reconciliation hides material fulfillment blockers has been corrected.

C:

That scorecard concerns the activity in which sales and service agents need the same customer and case context and its result of cross-app case-state agreement rate; the reported shortcoming here is delayed reconciliation hides material fulfillment blockers.

D:

The proposed metric concerns multiple AI teams consume a changing source API, whereas the failure in this case is that delayed reconciliation hides material fulfillment blockers; the indicator needed here is time from source exception to downstream acknowledgement.

Question 54

Before production promotion, the team needs evidence for this setting: An agent creates an order but reservation of inventory fails. A prior regression caused the following: Half-complete transactions leave operational state inconsistent. Which validation is most persuasive?

  1. Force a downstream failure after the initial write
  2. Disable inference and verify manual form submission remains possible
  3. Modify a test purchase order and validate authorized grounded retrieval
  4. Test ambiguous requests with incomplete required inputs
  5. Submit authorized and unauthorized return requests

Explanation

Correct Answer

A:

Forcing a downstream failure after the initial write is aligned with the failure that half-complete transactions leave operational state inconsistent; its causal relevance is that compensating operations reconcile earlier successful steps for the situation where an agent creates an order but reservation of inventory fails.

Incorrect Answers

B:

The tested behavior is that timeouts and graceful degradation protect core app operations; the necessary evidence here concerns compensating operations reconcile earlier successful steps and the previously observed half-complete transactions leave operational state inconsistent.

C:

Although the exercise might catch a static knowledge upload gives stale purchase-order states, it would not directly validate the mechanism that compensating operations reconcile earlier successful steps in the target implementation.

D:

The tested behavior is that specific tool metadata helps the planner choose and parameterize actions; the necessary evidence here concerns compensating operations reconcile earlier successful steps and the previously observed half-complete transactions leave operational state inconsistent.

E:

In the alternate setting where an employee requests a standard equipment return initiation, the test exercises single-purpose tools and typed parameters bound the task; even a pass would not show that half-complete transactions leave operational state inconsistent has been prevented.

Question 55

An agent-assisted service does not behave as specified: Shipping estimates reflect last week's warehouse data. It is used in this situation: Operational records update faster than the agent's indexed copy. Which observable diagnostic procedure is best?

  1. Compare source update timestamps with ingestion checkpoints
  2. Segment failures by route, token cost and difficulty
  3. Inspect table metadata and relationship mappings across applications
  4. Analyze route decisions by task class and observed error

Explanation

Correct Answer

A:

For the situation where operational records update faster than the agent's indexed copy, the defect is that shipping estimates reflect last week's warehouse data; comparing source update timestamps with ingestion checkpoints helps verify the cause-and-effect path: incremental refresh and source-to-index lag monitoring reduce staleness.

Incorrect Answers

B:

That probe evaluates feedback-driven thresholds rebalance accuracy, latency and spend in a workflow where real task mix differs from prelaunch performance tests; the reported failure concerns the path for which incremental refresh and source-to-index lag monitoring reduce staleness instead.

C:

The proposed inspection is associated with agents and apps need shared definitions of customers and work orders, where Dataverse tables and governed relationships provide a reusable business-data surface; it would not expose the failure mode that shipping estimates reflect last week's warehouse data with the required specificity.

D:

The proposed probe concerns many calls require only a small fixed-label output, where task-aware routing chooses economical models after quality gates; this incident instead requires evidence about incremental refresh and source-to-index lag monitoring reduce staleness and the failure that shipping estimates reflect last week's warehouse data.

Question 56

An enterprise team is making a platform design decision. Business context: Operational records update faster than the agent's indexed copy. Present failure: Shipping estimates reflect last week's warehouse data. Which approach resolves the mismatch?

  1. Use retrieval citations with explicit no-source handling
  2. Normalize identifiers and remove duplicated grounding records
  3. Give prompts explicit role, task, constraints and response schema
  4. Synchronize an indexed source on an explicit freshness schedule

Explanation

Correct Answer

D:

For the need that operational records update faster than the agent's indexed copy, the relevant mechanism is that incremental refresh and source-to-index lag monitoring reduce staleness; that architectural dependency addresses the observed failure where shipping estimates reflect last week's warehouse data.

Incorrect Answers

A:

Helpful-looking responses contain unverifiable claims; the alternate design can address that situation because cited retrieval separates factual support from free-form generation, whereas this question needs a solution to shipping estimates reflect last week's warehouse data.

B:

There is a valid place for the mechanism that canonical keys and deduplication prevent incorrect cross-record joins, namely multiple systems use inconsistent customer identifiers and duplicate cases; it does not remove the specific defect that shipping estimates reflect last week's warehouse data from this case.

C:

In the case where agents return inconsistent fields for downstream integrations, the alternate design depends on clear instructions and structured outputs reduce avoidable ambiguity; it does not correct the target failure, shipping estimates reflect last week's warehouse data.

Question 57

After deployment, stakeholders need an interpretable measure for this use case: Several events refer to one business order at once. The pilot showed that parallel agents each submit a costly replacement. Which metric captures the needed result?

  1. Verified task resolution rate
  2. Rate of requests rejected by capacity constraints
  3. Orphaned business workflows after simulated timeouts
  4. Duplicate mutations per concurrent event batch

Explanation

Correct Answer

D:

The observed shortcoming is that parallel agents each submit a costly replacement; measuring duplicate mutations per concurrent event batch gives business owners a directly relevant indicator for the process where several events refer to one business order at once.

Incorrect Answers

A:

For the situation where a service bot handles many chats but few cases are resolved, verified task resolution rate is a sensible indicator; the problem that parallel agents each submit a costly replacement requires measurement of duplicate mutations per concurrent event batch.

B:

A regional launch multiplies concurrent agent requests; its intended measurement is rate of requests rejected by capacity constraints; substituting that result here would not capture whether parallel agents each submit a costly replacement has been corrected.

C:

Orphaned business workflows after simulated timeouts; that indicator describes a distinct workflow where an autonomous agent relies on slow approvals and delayed tool results; it does not quantify duplicate mutations per concurrent event batch, which is needed for the stated case.

Question 58

One enterprise rollout has 2 separately testable requirements. (1) Business use: Independent teams choose inconsistent evaluation and security methods. Exposure: A business unit could exempt its deployment from shared controls without documented ownership. (2) Business use: Teams rebuild similar assistants without seeing existing assets. Exposure: Unmaintained reusable components could continue being used after deprecation. Which 2 safeguards directly satisfy these stated cases? Choose TWO.

  1. Narrow allowed tool capabilities and verify server-side authorization
  2. Require exception ownership and policy-version review
  3. Retire obsolete shared components and notify dependent teams
  4. Restrict write APIs to the authoritative owning service

Explanation

Correct Answers

B:

For the exposure that a business unit could exempt its deployment from shared controls without documented ownership, a CoE publishes patterns, reusable controls, standards and escalation paths explains the protection required when independent teams choose inconsistent evaluation and security methods; this does not replace control of unmaintained reusable components could continue being used after deprecation.

C:

For the exposure that unmaintained reusable components could continue being used after deprecation, discovery and ownership metadata support reuse and controlled deprecation explains the protection required when teams rebuild similar assistants without seeing existing assets; this does not replace control of a business unit could exempt its deployment from shared controls without documented ownership.

Incorrect Answers

A:

This security safeguard addresses the unrelated exposure that a shared MCP server might honor an unauthorized tool invocation in a workflow where a Copilot Studio agent needs a reusable interface to governed operations; the question instead names the specific risks a business unit could exempt its deployment from shared controls without documented ownership; unmaintained reusable components could continue being used after deprecation; this choice does not directly mitigate those specific exposures.

D:

This security safeguard addresses the unrelated exposure that two apps might overwrite each other's supposedly authoritative customer values in a workflow where two applications try to overwrite the same customer credit field; the question instead names the specific risks a business unit could exempt its deployment from shared controls without documented ownership; unmaintained reusable components could continue being used after deprecation; this choice does not directly mitigate those specific exposures.

Question 59

Before introducing a fix, the team must explain why prototype activity increases but no business value is measured. The affected setting is: Executives request AI before defining an operational improvement target. What should be inspected?

  1. Inspect table metadata and relationship mappings across applications
  2. Interview sponsors to locate missing success criteria
  3. Compare baseline model errors by language and domain
  4. Review available tool inventory versus approved actions

Explanation

Correct Answer

B:

The investigation directly tests the causal connection: cloud adoption planning begins from motivations, outcomes and readiness constraints; it collects evidence for a workflow where executives request AI before defining an operational improvement target but the outcome is instead that prototype activity increases but no business value is measured.

Incorrect Answers

A:

That probe evaluates Dataverse tables and governed relationships provide a reusable business-data surface in a workflow where agents and apps need shared definitions of customers and work orders; the reported failure concerns the path for which cloud adoption planning begins from motivations, outcomes and readiness constraints instead.

C:

Comparing baseline model errors by language and domain explores a separate connection, where a measured capability gap, appropriate dataset and MLOps plan justify customization; the diagnosed requirement is executives request AI before defining an operational improvement target, and the actual defect is prototype activity increases but no business value is measured.

D:

That probe evaluates allowlisting and server-enforced identity restrict actual executable capability in a workflow where agent instructions mention many APIs but only two are permitted; the reported failure concerns the path for which cloud adoption planning begins from motivations, outcomes and readiness constraints instead.

Question 60

A planned AI solution must support this situation: A flow moved to production still points to a test endpoint. Its prototype fails because agent writes real customer data into the sandbox. Which design best fits the constraint?

  1. Use managed solutions for governed downstream environments
  2. Preserve customer communication approvals across solution releases
  3. Coordinate Sales Copilot and Service Copilot customization changes
  4. Parameterize environment-specific connection references and variables

Explanation

Correct Answer

D:

Without the intended behavior, agent writes real customer data into the sandbox; the proposed architecture establishes the specific design principle that environment configuration separates deployable logic from endpoint bindings for the process where a flow moved to production still points to a test endpoint.

Incorrect Answers

A:

For production components should not be edited casually, this other approach relies on the fact that managed solution deployment supports controlled downstream customization boundaries; the issue here is agent writes real customer data into the sandbox; that mechanism does not establish the required environment configuration separates deployable logic from endpoint bindings.

B:

For service agents draft customer messages but cannot send without review, this other approach relies on the fact that business approval checks remain independent of agent version; the issue here is agent writes real customer data into the sandbox; that mechanism does not establish the required environment configuration separates deployable logic from endpoint bindings.

C:

In the case where a customer handoff depends on shared record and term mappings, the alternate design depends on cross-app release plans validate shared contracts and timing; it does not correct the target failure, agent writes real customer data into the sandbox.

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