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

# Agent Node

> The main conversation node — where the agent talks, listens, and decides what to do next.

The **Agent Node** is the core of your agent's conversation. Each node handles one step — the agent speaks, listens, and then follows a pathway based on what the user said.

You can have as many Agent Nodes as needed. Each one owns a specific part of the conversation — qualification, objection handling, scheduling, confirmation, and so on.

## Settings

**Name**
A short label shown on the canvas and in call logs. Use a name that describes what this step does — for example, `Qualify Budget`, `Handle Objection`, or `Confirm Appointment`.

***

**Prompt**
The agent's instructions for this step. Write the goal of this step, what the agent should ask, and how to handle different user responses. Supports `{{template_variables}}` from variable extraction or Pre-Call Data Fetch.

***

**Allow Interruption**

* **On (default)** — User can speak while the agent is talking; the agent stops and listens immediately
* **Off** — Agent finishes speaking before listening; use this for disclosures or critical statements that must be heard in full

***

**Add Global Prompt**
When turned on, the [Global Node](/workspace/nodes/global) instructions are prepended to this node's prompt at runtime. Keep this on for most nodes to maintain consistent tone and behaviour throughout the call.

***

**Enable Variable Extraction**
When turned on, the agent extracts specific data points from the user's response at this step. You define what to capture — for example, the user's decision, their preferred date, or whether they agreed.

Each variable you add has:

* **Name** — what to call it (e.g. `payment_confirmed`, `preferred_date`)
* **Type** — text, number, or boolean
* **Description** — what the agent should listen for

Extracted variables are available as `{{gathered_context.variable_name}}` in any later node's prompt. They also appear in the **Leads** page and can be used as conditions in the **Follow-up** node.

***

**Tools**
Attach tools the agent can invoke during this step — for example, booking an appointment, looking up availability, or transferring the call. Tools run when the agent decides to call them based on the conversation.

See [Tools](/workspace/tools) for how to create and manage tools.

***

**Knowledge Base Documents**
Attach documents the agent can reference while handling this step. Useful when the agent needs to answer product questions, pricing queries, or policy-related questions mid-conversation.

See [Knowledge Base](/workspace/knowledge-base) for how to upload and manage documents.

***

## Conditions

Conditions are the arrows connecting this node to the next one on the canvas. Add as many as needed — one per possible outcome of this step.

See [Conditions](/workspace/nodes/conditions) for a full explanation of Condition Label, Condition, and Transition Speech.