Hierarchical coordination
Hierarchical coordination is a multi-agent pattern in which agents at higher levels delegate, direct, or supervise work performed by agents at lower levels.
AI Agents in Action, Second Edition
An agent hierarchy is a multi-agent coordination pattern in which agents are organized into managerial levels that delegate commands and tasks downward.
- How it is used
- Organizes multi-agent delegation through managers and workers, including delegation on a user’s behalf.
- How deep it goes
- Defines managerial delegation and compares execution patterns, including parallel and sequential workers, without providing an implementation.
- Examples
- Illustrates a user interface agent delegating to multiple agents and a hierarchy combining parallel and sequential execution.
- What you can do after
- Evaluate when hierarchical delegation is appropriate and when a simpler coordination pattern may suffice.
Read how AI Agents in Action, Second Edition explains hierarchical coordination →
Designing AI Agents
A collaboration pattern that delegates bounded portions of work to agents arranged hierarchically.
- How it is used
- Coordinates dependent, bounded deliverables under a manager responsible for preserving and integrating a multi-stage goal.
- How deep it goes
- Details delegation boundaries, decision ownership, failure recovery, and operational metrics, with an implementation sketch.
- Examples
- Uses Manager Devin's incomplete information flow and Claude Code subagents to illustrate coordination and delegation.
- What you can do after
- Evaluate and apply hierarchical delegation by assigning decision ownership, checking deliverables, and comparing results with a single-agent baseline.
Read how Designing AI Agents explains hierarchical coordination →
AI Agents in Action
Hierarchical processing is the organization and execution of agents or tasks through multiple levels of authority or coordination.
- How it is used
- Coordinates CrewAI agents and tasks through a manager instead of processing tasks sequentially.
- How deep it goes
- Configures a manager LLM and runs a coding crew, but does not detail the manager’s internal operation.
- Examples
- Demonstrates a software engineering and QA crew creating a snake game and checking its code.
- What you can do after
- Configure a hierarchical CrewAI run and inspect its cost and repeated thoughts with AgentOps.
Read how AI Agents in Action explains hierarchical coordination →
Also covered in
- Grokking AI Applications
- Context EngineeringIn early access (MEAP)
What to know first
- A2AA2A is an open interoperability protocol that enables software agents to discover, communicate, and coordinate with one another across systems.
- multi-agent systemA multi-agent system is a computational system composed of multiple autonomous or semi-autonomous agents that interact or coordinate to perform tasks or pursue goals.
- subagentA subagent is an agent assigned by another agent to perform a delegated part of a larger task under its coordination or authority.
- context windowA context window is a bounded span of sequence elements or contextual information that a model or processing operation can consider together.
Related concepts
- collaborationimplements · Collaboration is the coordinated work of multiple participants who contribute toward a shared goal or outcome.
- CrewAIused by · CrewAI is an open-source framework for building and orchestrating collaborative AI agents and multi-agent workflows.
- fan-out/gathercontrasts with · Fan-out/gather is a coordination pattern that distributes work across multiple agents or workers and then aggregates their results.
- subagenthas part · A subagent is an agent assigned by another agent to perform a delegated part of a larger task under its coordination or authority.
Questions about hierarchical coordination
What is hierarchical coordination?
Hierarchical coordination is a multi-agent pattern in which agents at higher levels delegate, direct, or supervise work performed by agents at lower levels.
Which Manning books explain hierarchical coordination?
AI Agents in Action, Second Edition, Designing AI Agents and AI Agents in Action.
What should I know before learning hierarchical coordination?
Manning books that teach hierarchical coordination usually introduce A2A, multi-agent system, subagent and context window first.
What is the difference between hierarchical coordination and fan-out/gather?
Hierarchical coordination is a multi-agent pattern in which agents at higher levels delegate, direct, or supervise work performed by agents at lower levels. Fan-out/gather is a coordination pattern that distributes work across multiple agents or workers and then aggregates their results.
What are the parts of hierarchical coordination?
Hierarchical coordination includes subagent.
These guides are generated from the text of Manning books by Manning's knowledge graph and link to the sections they draw on. The full books are available on liveBook and at manning.com.


