> For the complete documentation index, see [llms.txt](https://docs.flowiseai.com/llms.txt). Markdown versions of documentation pages are available by appending `.md` to page URLs; this page is available as [Markdown](https://docs.flowiseai.com/tutorials/agent-as-tool.md).

# Agent as Tool

In this tutorial, we are going to take a look at how to leverage other flows as tools to a parent Agent. This approach allows you to create a parent agent that can delegate specific tasks to specialized child agents

## Overview

1. Receives user input through a parent agent
2. Agent decides to retrieve data from document store, or call the Agentflow Tool.

<figure><img src="/files/kqNBqPWuSuntk8VIjnOo" alt="" width="375"><figcaption></figcaption></figure>

### Step 1: Setting Up the Start Node

Begin by adding a **Start** node to your canvas. This serves as the entry point for your agent system.

### Step 2: Creating the Parent Agent

Add an **Agent** node and connect it to the Start node.

### Step 3: Configuring the Agent Tool

The key feature of this flow is configuring another agent as a tool. In the Parent Agent's **Tools** section:

<figure><img src="/files/BiJk1PW3nbaK7p8alGTF" alt="" width="354"><figcaption></figcaption></figure>

#### Tool Configuration:

* **Tool**: Select "**Agent As Tool**"

#### Agent Tool Settings:

* **Selected Agentflow**: Choose your child agentflow
* **Name**: Name for the agentflow
* **Description**: Describe when this agentflow is useful. Example:

```
Useful for searching user availability, scheduling meetings and email related query
```

{% hint style="warning" %}
Name and Description for the tool are extremely important! They must be clear and correctly describe the purpose of the tool. Refer to [best practices](https://platform.openai.com/docs/guides/function-calling?api-mode=chat#best-practices-for-defining-functions) guide.
{% endhint %}

### Step 4: Adding Knowledge Sources

Configure the **Knowledge (Document Stores)** section to give your parent agent access to relevant information. This is the same as [RAG](/tutorials/rag.md) tutorial.

<figure><img src="/files/z0kwfoMOXv8vXEwHaw1d" alt="" width="518"><figcaption></figcaption></figure>

#### Document Store Configuration:

* **Document Store**: Select your pre-configured document store (e.g., "AI-Paper")
* **Describe Knowledge**: Describe what the knowledge is about

***

## Example Interactions

#### Sample Queries and Expected Behavior:

**Scheduling Query:**

* User: "Can you check my availability for next Tuesday?"
* Flow: Parent agent → personal\_assistant tool → specialized scheduling response

<figure><img src="/files/P1xc3CTTy9b8sMGdXDl1" alt="" width="563"><figcaption></figcaption></figure>

**Technical Query:**

* User: "What is AIGC and how does it work?"
* Flow: Parent agent → AI-Paper knowledge base → technical explanation with sources

<figure><img src="/files/CtT4kySpcnVuQaAqOcF4" alt="" width="563"><figcaption></figcaption></figure>

**General Query:**

* User: "Hello how are you?"
* Flow: Parent agent → direct response (no tools needed)

**Complex Query:**

* User: "Schedule a meeting about AIGC implementation next Tuesday, extract key insights and the talking points"
* Flow: Parent agent → both personal\_assistant tool AND AI-Paper knowledge → coordinated response

<figure><img src="/files/oOSiZFhiK2q876xlKliz" alt="" width="563"><figcaption></figcaption></figure>

***

## Best Practices

#### Design Guidelines:

1. **Clear Tool Descriptions**: Make tool name and descriptions specific and actionable
2. **Appropriate Delegation**: Better system prompt for parent agent to delegate effectively

#### Common Use Cases:

* **Customer Service**: Parent agent with specialized tools for billing, technical support, and general inquiries
* **Research Assistant**: Parent with tools for different research domains (legal, technical, market research)
* **Project Management**: Parent with tools for scheduling, resource allocation, and progress tracking
* **Content Creation**: Parent with tools for writing, editing, research, and formatting