Skip to main content
Hero Light

​
Overview

Agentic AI is a CLI tool that enables AI-powered phone call agents with natural conversation capabilities. It combines OpenAI Realtime API for accurate voice transcription and generation, Twilio for phone connectivity, and ClawdBot integration via OpenClaw Gateway for executing commands during calls.

​
What Can Agentic AI Do?

Agentic AI enables you to:

Make AI Phone Calls

Create AI agents that can make and receive phone calls with natural voice conversation powered by OpenAI Realtime API.

Execute Commands

Let your AI agent execute real-world commands via ClawdBot integration - play music, send messages, check emails, and more.

Accurate Transcription

Built-in Whisper STT from OpenAI Realtime API ensures proper nouns and names are transcribed correctly.

Background Service

Run as a background daemon with auto-restart on crash, perfect for production deployments.

​
Key Features

Powered by OpenAI Realtime API with gpt-4o-realtime-preview-2024-12-17 model, enabling natural two-way conversations with ultra-low latency. Choose from 6 different voice options including alloy, echo, fable, onyx, nova, and shimmer.
Make outbound calls to any phone number using the CLI, or configure your Twilio number to have the AI answer incoming calls automatically. Perfect for appointment reminders, customer service, or interactive voice agents.
Built-in conversation brain analyzes user intent using Gemini AI to determine if the caller wants to execute an action. Actionable commands are automatically routed to ClawdBot for execution.
Execute real-world commands during calls via OpenClaw Gateway:
  • Open YouTube and play videos
  • Play songs on Spotify
  • Check and read emails
  • Send WhatsApp messages
  • Web search and more
Results are spoken back to the caller in real-time.
Optional Telegram bot integration sends live transcripts and command extractions to your Telegram chat, allowing you to monitor calls in real-time.
Built-in support for ngrok and Cloudflare tunnels to expose your local webhook server to Twilio. Includes CLI commands to start tunnels and manage webhook URLs.

​
Architecture

Agentic AI uses a modular architecture that bridges multiple services: How it works:
  1. Phone connectivity - Twilio handles the telephony layer, connecting phone calls to your server via WebSocket media streams
  2. Audio bridge - Agentic AI server routes audio between Twilio and OpenAI Realtime API with proper format conversion
  3. Voice processing - OpenAI Realtime API transcribes incoming speech with Whisper and generates natural voice responses
  4. Intent analysis - Gemini AI analyzes transcripts to identify actionable commands
  5. Command execution - ClawdBot executes commands via OpenClaw Gateway (YouTube, Spotify, email, etc.)
  6. Response delivery - ClawdBot results are spoken back to the caller through OpenAI’s voice synthesis

​
Use Cases

Appointment Reminders

Automatically call customers to confirm appointments, reschedule if needed, and execute calendar actions.

Customer Support

AI-powered support agent that can answer questions, look up information, and execute actions on behalf of customers.

Personal Voice Assistant

Call your personal AI assistant to play music, send messages, check emails, or control smart home devices.

Survey & Feedback Collection

Conduct automated phone surveys with natural conversation flow and intent understanding.

​
Get Started

Ready to build your AI voice agent? Follow our quickstart guide:

Quickstart Guide

Make your first AI phone call in under 10 minutes

​
Requirements

Before you begin, you’ll need:
The quickstart guide walks through obtaining all required API keys and credentials.

Build docs developers (and LLMs) love

Get started for freeTalk to us