Tools Overview

Tools let your assistant call external APIs during a conversation. When the LLM decides it needs data or needs to trigger an action, it invokes the appropriate tool — the platform makes the HTTP request and returns the result to the LLM.

How Tools Work

Caller: "What's my order status?"
  → Assistant: calls get_order_status tool
    → Platform: sends HTTP request to your API
      → Your API: returns {status: "shipped", eta: "2 days"}
    → Platform: passes result to LLM
  → Assistant: "Your order has been shipped and will arrive in 2 days."

Tool Types

Type	Purpose
Function	Standard function calling (OpenAI-compatible). Define name, description, and JSON Schema parameters.
DTMF	Collect DTMF (touch-tone) input from the caller during a tool execution.
End Call	End the call when this tool is invoked.

Server Configuration

Each tool can have a server configuration that defines where the request is sent:

Setting	Description
URL	The endpoint the platform calls
Method	HTTP method (POST, GET, etc.)
Headers	Custom headers sent with each request
Credential	Reference to a server credential for authentication

The server can also use messages — a sequence of messages to play to the caller while waiting for the API response (e.g., “One moment while I look that up…”).

Next Steps

Create a Tool

Define your first function tool

Configure

Server config, credentials, and messaging

Unlink a Phone Number Create a Tool

Getting Started

Assistants

WebRTC & LiveKit

Workflows

Phone Numbers

Tools

Developer

Billing

Monitoring

Tools Overview

Tools Overview

How Tools Work

Tool Types

Server Configuration

Next Steps

Create a Tool

Configure

​Tools Overview

​How Tools Work

​Tool Types

​Server Configuration

​Next Steps

Create a Tool

Configure

Tools Overview

How Tools Work

Tool Types

Server Configuration

Next Steps