A. Incorrect.
Agentspace is not primarily an infrastructure management tool.

B. Correct.
Agentspace helps employees search, reason, and act across enterprise data and systems, reducing knowledge silos.

C. Incorrect.
Security matters, but the core business advantage in the scenario is knowledge access and collaboration.

D. Incorrect.
Performance review automation is not the stated use case.

A. Correct.
Free-form text has no fixed table schema and is considered unstructured data.

B. Incorrect.
Structured data is organized in predefined rows, columns, and types.

C. Incorrect.
Labeled data has target annotations or categories; the raw feedback is not labeled yet.

D. Incorrect.
Quantitative data is numeric and measurable, unlike open-ended text.

A. Correct.
The agent automates a multi-step operational business process.

B. Incorrect.
Productivity agents usually help individuals with tasks such as drafting or summarizing.

C. Incorrect.
Customer service agents focus on customer interactions, not end-to-end fulfillment operations.

D. Incorrect.
Conversational agents focus on dialogue; this scenario emphasizes workflow execution.

A. Incorrect.
Fine-tuning is expensive and unnecessary before trying better prompt context.

B. Correct.
Clear prompt details are the fastest way to guide style, audience, tone, and required brand themes.

C. Incorrect.
More tokens may make the output longer, but not necessarily more relevant.

D. Incorrect.
Lower temperature can make output more consistent, but it does not add missing brand context.

A. Incorrect.
That describes a broader Contact Center as a Service platform.

B. Incorrect.
That describes Conversational Insights.

C. Correct.
Agent Assist supports live agents with real-time suggestions and relevant information.

D. Incorrect.
Automated chatbot building is a different customer engagement capability.

A. Incorrect.
A full knowledge-base Q&A solution usually needs retrieval and enterprise search components.

B. Incorrect.
Numerical forecasting is better suited to analytics or predictive ML systems.

C. Incorrect.
Document extraction is better suited to specialized document AI workflows.

D. Correct.
Gemma is an open LLM family well suited for text generation, summarization, and drafting.

A. Incorrect.
Prompt chaining breaks tasks into steps; it does not directly establish persona and tone.

B. Correct.
Role prompting sets the professional persona, tone, and context for consistent customer service replies.

C. Incorrect.
A single example can help, but it is less direct than assigning the desired role and context.

D. Incorrect.
Few-shot examples can help, but the question asks for guiding tone and policy-aware persona.

A. Incorrect.
RAG is an architecture pattern for retrieval and grounding, not a front-end generator.

B. Incorrect.
RAG avoids retraining for every document update by retrieving relevant context at answer time.

C. Correct.
RAG grounds responses in relevant source documents, improving freshness and reducing hallucination.

D. Incorrect.
Summaries may be produced, but the main benefit is grounded retrieval before generation.

A. Correct.
Model Garden is the place to discover, test, customize, and deploy a wide range of models.

B. Incorrect.
Evaluation is part of the broader platform, but not the unique purpose of Model Garden.

C. Incorrect.
Model registry and metadata tools address version management.

D. Incorrect.
Custom training is separate from discovering and using existing foundation models.

A. Incorrect.
Security is important, but it does not explain why applicants are rejected or address fairness concerns.

B. Correct.
Explainability helps audit decisions, detect bias, support compliance, and build trust in high-stakes lending.

C. Incorrect.
Faster processing could worsen harm if decisions are biased or unexplained.

D. Incorrect.
More data can improve performance, but it does not by itself establish fairness or accountability.