Community Q&A
AI-500 — frequently asked questions
Community-vetted answers to 13 common questions about this exam, each traced back to a specific practice question.
Questions from real practice questions
Each Q&A comes from a specific question — follow the link for the full analysisHow Should You Configure Foundry Multi-Agent Prompt Evaluations?
Which evaluator catches a Foundry connected agent that skips mandated evidence gathering?
A prompt-adherence evaluation, because it measures whether the subagent follows its prompt instructions, including the required evidence-gathering tool steps before it answers.
How do you detect when a subagent's JSON handoff payload cannot be processed downstream?
Add a schema-match evaluation that validates each handoff payload against the expected JSON structure, flagging any change that breaks deserialization by the next agent.
Can groundedness or task-completion evaluators detect out-of-scope agent work?
No. Those score factual support and goal attainment, so scope drift goes unnoticed; use a responsibility-boundary evaluation that tests in-scope responses and out-of-scope refusals.
Which Evaluator Isolates RAG Retrieval Failures per Agent
Why isn't groundedness the right evaluator for these failed traces?
Groundedness scores whether the final response is supported by retrieved context; it cannot show that an embedding or analyzer change altered which chunks were retrieved or their rank order.
Why do both agents get the same evaluator in this drag-and-drop?
Both the support agent and the policy agent query the same Foundry IQ knowledge base, so the embedding and analyzer changes degrade retrieval for both, requiring the document retrieval evaluator in each slot.
Yes/No: Multi-Agent Observability Monitoring in Microsoft Foundry
Why are token usage and TTFT counted as No in this Foundry observability hotspot?
The scenario says the dashboard includes token usage and TTFT, but the only alert rules configured are for latency anomalies and tool invocation failures, so no alerting coverage exists for those signals.
How do trace ID, span ID and conversation ID help in a multi-agent solution with concurrent agents?
The single conversation trace plus emitted trace and span IDs let you correlate Agent1 and Agent2 spans, MCP tool invocations, and external REST calls back to one user request for end-to-end tracing.
Physician Approval for ER Triage in Lead Orchestrator Workflow
Why can't a Deliver a message node satisfy the physician approval requirement?
A message node only outputs text to the conversation and never pauses execution, so the workflow would proceed without any physician decision being captured or enforced.
How does Ask a question differ from adding an ER recommendation to the system prompt?
Hardcoding ER rules in the prompt violates the safety requirement to keep logic out of core prompts; the Ask a question node enforces approval dynamically at the workflow level through human-in-the-loop review.
Which Knowledge Type for Foundry Multi-Agent Docs?
Why not use a File knowledge type for the shared documentation?
File grounding requires re-uploading content per agent, breaks the single maintained layer, and does not return structured query answers or citation metadata with each response.
How does Azure AI Search handle wording that doesn't match the documentation?
Vector/hybrid search plus the semantic ranker and synonym maps match meaning rather than exact tokens, so conversational follow-ups and unfamiliar terminology still retrieve relevant passages.
Which Identity Access Configs for Foundry Agent Tools?
Why does interactive ticket search need On-Behalf-Of delegated permissions?
OBO passes the signed-in employee's Entra token to the ticket tool, so the tool's existing per-employee authorization still applies to every query.
How do these configs avoid embedded secrets while staying least-privilege?
OBO exchanges tokens through Entra and app-only access uses a managed identity, so no key is stored, and each identity receives only the rights its context needs.