Correct Answer:

Explanation:
Extracted memory facts: Azure Cosmos DB; Cache: Azure Managed Redis with tenant-prefixed keys; Transcript exports: A shared blob container with lifecycle management policies.
The design separates durable memory, distributed cache, and retained transcripts. Azure Cosmos DB is appropriate for durable extracted facts that must survive deletion of a service conversation and remain available across stateless compute instances. Azure Managed Redis provides a shared low-latency cache; tenant-prefixed keys are a standard multitenant pattern that avoids local-process affinity. Transcript exports belong in Blob Storage, where lifecycle management rules can enforce retention and can be filtered by prefixes or tags so different tenant retention policies can coexist in one managed storage design. This combination keeps the compute tier stateless while avoiding unnecessary per-tenant infrastructure. The important security requirement is that every persistence layer also enforce tenant authorization, not just naming conventions. Within the options provided, the stated combination minimizes operational complexity while meeting cross-session persistence and retention needs. In production, add telemetry and regression tests around this behavior so changes to prompts, models, tools, or orchestration do not silently alter the intended contract. The selected approach is the one that best matches the platform ' s native execution semantics.
Official Microsoft reference: Azure Architecture Center - multitenant data and cache patterns

Correct Answer:

Explanation:
Quality of retrieved context: Retrieval evaluator; Directness of response to query: Relevance evaluator.
The dataset contains query, context, and response but does not contain document relevance labels. Microsoft ' s Retrieval evaluator is designed for exactly that situation: it uses an LLM judge to rate how relevant the retrieved context chunks are to the query and returns a 1-to-5 score with pass/fail behavior. The Relevance evaluator operates on the final response and measures whether the answer accurately, completely, and directly addresses the query. Document Retrieval is not appropriate because it requires retrieval ground truth such as known relevant documents or qrels. Groundedness answers a different question: whether response claims are supported by the provided context. Therefore the correct mapping is Retrieval for context quality and Relevance for response directness. For operational use, the measurement should be captured in a repeatable dataset, trace, or automated gate so that the same criterion can be compared across versions. That is more useful than a one-off manual observation and makes regressions visible before they become production incidents.
Official Microsoft reference: Microsoft Foundry - RAG evaluators

Correct Answer: D  🗳️

Explanation: Only visible for TestValid members. You can sign-up / login (it's free).

Correct Answer: B  🗳️

Explanation: Only visible for TestValid members. You can sign-up / login (it's free).

Correct Answer:

Explanation:
No / No / No
All three consultant-proposal statements should be rejected. Applying the highest sensitivity to every guardrail control is not automatically safer because it can increase false positives and unnecessarily block legitimate clinical interactions; controls should be tuned to the actual risk and tested. Telling the model in a system prompt to ignore hate speech is not equivalent to enabling the platform content-safety controls required for public-facing agents. Finally, repeatedly prompting patients to verify symptoms does not address long-context growth and can worsen token usage and latency. Microsoft guidance separates deterministic safety controls, evaluation, and context-management mechanisms from ordinary prompt instructions. The proposal also hardcodes medical behavior in prompts, contrary to the requirement to avoid embedding new clinical logic in the core prompt. A robust design would use appropriate content-safety/guardrail controls, validated intervention points, and a deliberate memory/compaction strategy rather than relying on increasingly restrictive or repetitive prompt text. The same configuration should be paired with auditable identity, trace, and evaluation data so reviewers can prove which principal acted, which policy was applied, and why a request was allowed or blocked. That is particularly important for production multi-agent systems with external tools.
Official Microsoft reference: AI-500 Study Guide - guardrails, context management, and evaluation