Question

You are building a customer support solution in Azure AI Foundry that must handle three distinct capabilities: answering billing questions from an internal knowledge store, processing refund requests that require calling an external payment API, and escalating complex complaints to a human. You want a specialized agent for each capability, with a coordinating agent that inspects each incoming request and routes it to the appropriate specialist, receiving results back before responding to the customer. Which orchestration approach best fits this requirement?

Accepted Answer

Implement a connected-agent (orchestrator with sub-agents) pattern where a main agent delegates tasks to specialized child agents and aggregates their responses.

Answer

Combine all three capabilities into a single agent with multiple tool definitions and rely on the model to decide which tool to call each turn.

Answer

Deploy three independent agents behind separate endpoints and have the client application call each one sequentially in a fixed order.

Answer

Use a hard-coded conditional workflow with fixed branches that route inputs to different LLM deployments based on keyword matching.