Microsoft AB-100: Deploy AI-powered business solutions Practice Questions

Provisioning the environment in an EU geography and disabling cross-region data movement is the platform-level control that keeps data processing within the EU boundary for Power Platform and Dataverse workloads. Reviewing the Microsoft Product Terms and DPA validates Microsoft's contractual commitments for the specific Copilot feature — essential for compliance assurance when data is processed by Microsoft-hosted AI services. Azure Policy on Azure regions governs IaaS/PaaS resource deployment and does not control where Power Platform or Dynamics 365 SaaS features process data. DDoS Protection Standard protects against volumetric network attacks and has no effect on data residency. Conditional Access IP-based policies restrict which users can authenticate — they do not control where Microsoft's backend services process AI workloads.

Storing connector definitions as managed solution components in source control ensures version control, auditability, and environment lock-down — tenants cannot accidentally modify a production-managed connector. Automated OpenAPI schema validation in the CI/CD pipeline catches breaking changes (e.g., renamed parameters, removed endpoints) before they affect any tenant. Manual recreation across 15 tenants is error-prone, inconsistent, and unscalable. Allowing tenant administrators to modify managed connector definitions defeats the purpose of managed solutions and breaks the central governance model. Embedding a shared API key in the connector definition is a severe security anti-pattern: a single compromised key affects all 15 tenants and the key is visible to anyone who exports the solution.

Precision and recall on a held-out test set measure whether classification quality is degrading over time — the most direct indicator of model drift for a classification task. Monitoring covariate shift in input features detects when the real-world data characteristics have diverged from training data, which is the root cause of predictive drift before label-level evidence accumulates. Compute node count is an infrastructure cost metric with no bearing on predictive quality. Monthly workspace cost is a financial operations concern, not a model validation criterion. Average token count is relevant for LLM prompt optimization, not for a structured classification model.

The cross-app handoff trigger (option E) is the core integration behavior that end-to-end testing must validate. Accuracy of agent-generated summaries (option C) tests the AI's grounding quality — critical for trust in agentic outputs. Graceful handling of API throttling (option B) validates operational resilience across the integrated services. Licensing SKU validation (option A) is an IT provisioning concern, not a test scenario for solution behavior. Data residency verification (option F) is a compliance validation activity, not a functional end-to-end test scenario. Flow run entitlement monitoring (option D) is a capacity management concern outside the scope of functional end-to-end testing.

Fairness evaluation against protected demographic groups (A) is a mandatory Responsible AI gate that ensures the model does not produce biased credit recommendations — a regulatory requirement in financial services. Baseline regression metric thresholds (D) such as RMSE and MAE are standard quantitative quality gates that confirm the retrained model has not regressed in predictive accuracy before promotion. Together these address both performance and fairness. — Container vulnerability scanning (B) is an important DevSecOps practice but is an infrastructure gate, not a model validation criterion; it does not validate model quality or fairness. — Human-in-the-loop review for every prediction (C) is operationally infeasible at scale and is not a validation criterion for the model itself; human oversight is better applied at high-risk decision thresholds. — Compute cost budget compliance (E) is a cost-governance concern, not a model quality or responsible AI validation gate. — Requiring full human re-labelling of all data monthly (F) is disproportionate, not a standard validation criterion, and would make the monthly retraining cycle impractical.