# Worker Orchestration

Your Toolhouse workers can work directly with other workers. Rather than building one monolithic worker, you can create specialized workers and have a **manager** coordinate them — deciding who does what, in what order, and whether to run them in parallel or in sequence. The manager also delivers the final output to you.

You don't need to define a rigid workflow. The manager already knows how to interact with the workers you connect to it — but you can give it explicit instructions if you want more control.

### How the manager works

The manager is a worker like any other, but its job is coordination. It:

* Decides which specialized workers to involve for a given task
* Determines whether workers should run in sequence or in parallel
* Passes output from one worker as input to the next when needed
* Returns a final, consolidated result to you

### Building an orchestrated system

Build your specialized workers first, then create the manager **in a new chat** in Agent Builder. Creating the manager in a separate session ensures Agent Builder scopes it correctly and can discover the workers you've already built.

**Example:** Say you want a marketing worker that manages your online presence. You could build:

1. A **content generator** worker
2. An **email worker** connected to your email system
3. A **LinkedIn poster** worker for your page or profile

Once those exist, open a new chat and prompt Agent Builder with something like:

> Create a manager agent that's in charge of my marketing. It should call the content generator agent to generate the content, then feed the generated content to both the email and LinkedIn worker to adapt the content and post to the right channel.

The manager will call the content generator first, then pass its output to the email and LinkedIn workers — running them in parallel since they're independent.

### Parallel and sequential execution

The manager decides autonomously whether to run workers at the same time or one after another, based on data dependencies. You can also specify this explicitly in your prompt.

| Mode           | When it applies                                           |
| -------------- | --------------------------------------------------------- |
| **Sequential** | A worker's output is needed as input by the next          |
| **Parallel**   | Workers are independent and can run simultaneously        |
| **Mixed**      | A sequential step followed by parallel downstream workers |

### Hierarchical chains

Orchestration isn't limited to a single manager layer. Any worker can itself act as a manager for other workers — there's no limit to chain depth. A manager can call a worker that calls a worker that calls a worker, and so on.

### Adding orchestration manually

Agent Builder adds the **Call Agent** integration automatically based on what you're building. To add it yourself:

* Go to **Agents** in your Toolhouse dashboard
* Click on your worker to edit it
* Select **Integrations**, then click **Add Integration**
* Choose **Call Agent**
* Click **Save changes**


---

# Agent Instructions
This documentation is published with GitBook. GitBook is the documentation platform designed so that both humans and AI agents can read, navigate, and reason over technical content effectively. Learn more at gitbook.com.

## Querying This Documentation
If you need additional information that is not directly available in this page, you can query the documentation dynamically by asking a question.

Perform an HTTP GET request on the current page URL with the `ask` query parameter:

```
GET https://docs.toolhouse.ai/toolhouse/capabilites/worker-orchestration.md?ask=<question>
```

The question should be specific, self-contained, and written in natural language.
The response will contain a direct answer to the question and relevant excerpts and sources from the documentation.

Use this mechanism when the answer is not explicitly present in the current page, you need clarification or additional context, or you want to retrieve related documentation sections.