ALAgent Lab

Fan-out / fan-in pattern

Parallel Multi-Agent System

An orchestrator splits the task, several agents work at once, and their results are merged.

Best for tasks that break into independent parts

How it works

The Parallel Multi-Agent System pattern

One orchestrator decomposes the request into independent sub-tasks and dispatches them simultaneously. Each agent works without waiting for the others. When all have finished, an aggregator merges their outputs into a single response.

  1. 1

    Orchestrator

    Splits the request into independent sub-tasks.

  2. 2

    Agents run in parallel

    Each sub-task is handled at the same time by a dedicated agent.

  3. 3

    Aggregator

    Waits for all agents, then merges their outputs into one answer.

Why it matters

Wall-clock time is set by the slowest agent, not the sum of them, so independent work finishes far faster. It only helps when the sub-tasks genuinely do not depend on each other.


Playground

Run the parallel multi-agent system

Simulated
Try:

Your Question

Input

A question that breaks into independent parts

Orchestrator

Agent

Splits the task and dispatches sub-agents at once

Search Agent

Agent

Gathers current specs and pricing

Comparison Agent

Agent

Builds a spec-by-spec comparison

Review Agent

Agent

Summarises aggregate owner feedback

Aggregator

Aggregator

Merges the three results into a recommendation