> ## Documentation Index
> Fetch the complete documentation index at: https://docs.agentixhub.ai/llms.txt
> Use this file to discover all available pages before exploring further.

# Tasks

> The micro-units of work inside a job

## Definition

A **Task** is a single step inside a Job that needs to be performed by an AI agent, a human user or a system.

Each Task is created from a Workflow **Action** and includes the input, logic and expected outcome for that specific step.

## What Tasks Do

* Represent a **unit of work** inside the larger journey
* Represents a specific action to be taken, like uploading a document or speaking with an agent
* Can be handled by **Realtime** or **Task Agents**
* Carry all necessary **input/output**, instructions and agent configuration
* Are tracked for **status**, execution time and result
* Trigger assigned **tools** and use assigned **sources**

## Common Task Examples

* Upload and verify a document
* Join a video call with an AI agent
* Fill out a form with guided AI support
* Analyze an invoice and extract data
* Route a case based on form values

## 🖼️ Visual Overview

```mermaid theme={null}
graph TD
  Job --> Task1[Task: Upload Document]
  Task1 --> Task2[Task: Speak with AI Agent]
```

## Example Use Cases

<Card title="Identity Upload Task" horizontal>
  A Task where the customer uploads their ID, which is then processed by a Task Agent.
</Card>

<Card title="Live Help Session" horizontal>
  A Realtime Task where the user is assisted by an AI agent over voice or chat.
</Card>

<Card title="Form Validation" horizontal>
  A Task where a user fills a form and an MCP tool checks validity before continuing.
</Card>

## Task Metadata

Each Task contains:

* Assigned agent
* Status (pending, active, completed)
* Triggered tools
* User and session context
* Input/output data
* Logs and timestamps

***

> ✅ Tasks are how **Actions become real**, visible steps in a running workflow.