Concurrent

What this lesson demonstrates

Concurrent orchestration runs several agents on the same prompt in parallel. Each works independently, then a built-in aggregator fans their answers back in. It fits ensemble reasoning, brainstorming, and voting — where diverse perspectives on one input are the whole point. Latency is max(agent), not the sum.

ConcurrentBuilder(participants=[...]).build() wires the fan-out/fan-in graph for you; you just hand it a list of agents (or custom Executors). The lesson runs a researcher, a marketer, and a legal reviewer on one product-launch prompt.

One real excerpt

The default aggregator yields one AgentResponse holding one assistant message per participant:

from agent_framework.orchestrations import ConcurrentBuilder

return ConcurrentBuilder(participants=[researcher, marketer, legal]).build()

# ...
events = await workflow.run("We are launching a budget-friendly electric bike...")
final: AgentResponse = events.get_outputs()[0]
for msg in final.messages:                       # one message per expert
    print(f"[{msg.author_name or 'assistant'}]:\n{msg.text}")

The gotcha

Participants run with no ordering — all in parallel — and the default aggregator’s single AgentResponse does not include the original user prompt, only the experts’ replies (read via events.get_outputs()[0].messages; msg.author_name labels each). Pass intermediate_output_from=[...] to also surface each listed participant’s own output as "intermediate" events (handy in stream=True mode). To replace fan-in entirely — e.g. a summarizer agent that consolidates every expert into one string — use .with_aggregator(callback).

How it maps to Azure AI Foundry

All three experts are FoundryChatClient + AzureCliCredential agents; the builder is client-agnostic. Because they run concurrently, the three Foundry calls overlap — total wall-clock time is the slowest single expert, not the sum of all three.

Run it

uv run tutorial/03-workflows/orchestrations/02_concurrent.py

Needs Foundry credentials (az login). You should see a ===== Final Aggregated Results ===== block with one section per expert (researcher / marketer / legal).

Next: Handoff