A Python team on Azure must implement complex multi-agent control flow with branches, loops, and human approval nodes for a claims workflow. They are selecting frameworks aligned to AI-500 skills. Which approach is appropriate?
Select an answer to reveal the explanation.
Short Explanation
You need to build this with a real orchestration framework like Microsoft Agent Framework, LangGraph, or LangChain where appropriate. Model the states, edges, loops, branches, and human approval nodes explicitly in code so the workflow is testable and observable.
Full Explanation
The correct approach is implementing orchestration with Microsoft Agent Framework, LangGraph, or LangChain where appropriate, modeling states, edges, and human-in-the-loop nodes in code. The exam expects framework-based control flow, and explicit states and edges support approvals, retries, and specialist routing. Encoding control flow only as unrestricted natural language in one prompt is wrong because it lacks deterministic structure for complex branches. Using spreadsheet macros as the sole orchestrator is wrong because spreadsheets are not a scalable agent runtime. Relying on random agent selection each turn without a graph or hub is wrong because random selection is not a designed pattern. Combine framework orchestration with middleware for logging, auth, and exception handling.