Welcome to the VoiceAI documentation. Here you'll find clear explanations of all the key terms, technologies, and workflows that power modern voice AI telephony. Whether you're new or experienced, this page will help you understand how everything fits together—and how this app makes it easy.
Twilio is a cloud communications platform that provides APIs for voice, SMS, and other messaging services. In VoiceAI, Twilio is used to connect phone calls to AI agents, manage phone numbers, and handle telephony infrastructure without needing physical hardware.
LiveKit is a real-time audio/video infrastructure platform. In VoiceAI, it is used to bridge SIP telephony with WebRTC and AI agents, enabling scalable, low-latency, real-time voice communication between callers and AI.
UDP (User Datagram Protocol) and TCP (Transmission Control Protocol) are two core internet protocols. UDP is fast and connectionless, ideal for real-time audio (like calls), but less reliable. TCP is reliable and ordered, used for data where accuracy matters. Telephony often uses UDP for speed.
WebRTC (Web Real-Time Communication) is a technology that enables browsers and apps to communicate with audio, video, and data in real time. VoiceAI uses WebRTC to connect AI agents and callers with low latency.
WebSocket is a protocol for two-way communication between client and server over a single, long-lived connection. In VoiceAI, it is used for real-time signaling, control, and streaming data between services.
SIP is a signaling protocol used to initiate, maintain, and terminate real-time sessions (like voice calls) over IP networks. VoiceAI uses SIP to connect telephony providers (like Twilio) with real-time platforms (like LiveKit).
A SIP URL (e.g., sip:agent@domain.com) is an address used to route calls over SIP networks. It identifies endpoints (like AI agents or trunks) in telephony workflows.
An inbound trunk is a virtual phone line that receives incoming calls from the public telephone network and routes them into your system (e.g., to an AI agent).
An outbound trunk is a virtual phone line used to place outgoing calls from your system to the public telephone network.
A dispatch rule defines how incoming calls are routed to specific AI agents or workflows based on criteria like caller, time, or intent. This enables intelligent call distribution.
A Voice AI Agent is an AI-powered conversational entity that can answer phone calls, understand speech, and respond naturally. In VoiceAI, agents can handle tasks like customer support, scheduling, or information retrieval.
A workflow is a sequence of automated steps or logic that defines how calls are processed. For example, a workflow might route a call to Agent A, then escalate to Agent B if needed. Workflows can involve multiple agents and complex logic.
Function calling allows an AI agent to trigger external functions or APIs during a conversation (e.g., booking an appointment, sending a text). This enables VoiceAI agents to take real actions, not just converse.
Voice AI telephony solutions combine telephony infrastructure with AI to automate and enhance phone-based interactions. They enable businesses to scale support, sales, and operations with intelligent, automated voice agents.
Advances in AI, cloud telephony, and real-time infrastructure have made it possible to build scalable, intelligent voice solutions. Costs have dropped, APIs are mature, and customer expectations for automation are higher than ever.
This app integrates Twilio (telephony), LiveKit (real-time), OpenAI (AI), and workflow automation into a single, unified platform. It bridges the gap between phone calls, AI, and business logic.
Previously, integrating telephony, AI, and real-time infrastructure required deep expertise, custom code, and managing multiple vendors. Each piece had its own APIs, protocols, and quirks. This app abstracts and automates the hard parts.
This app provides a unified dashboard, prebuilt integrations, and workflow tools. You can connect providers, configure agents, and design call flows visually—no need to stitch together APIs or manage infrastructure.
Common workflows include: routing calls to the right agent, escalating from AI to human, booking appointments, sending follow-ups, and integrating with CRMs. This app lets you build these visually and automate them end-to-end.
A session is a complete interaction between a caller and an agent (AI or human), including audio, transcript, and metadata.
A task is a single action an AI agent can perform (e.g., answer a question, book a meeting). A workflow is a series of tasks and logic, possibly involving multiple agents and steps.
Set up your VoiceAI platform and make your first call
Understanding the VoiceAI ecosystem and components
Complete the onboarding process and configure your account
Make your first AI-powered phone call using the platform
Learn how to navigate the dashboard and access features
Configure Twilio for voice calls and phone number management
Create and configure your Twilio account for voice calls
Purchase and manage phone numbers for your AI agents
Set up SIP trunks for scalable voice communication
Configure webhook endpoints for call handling
Implement intelligent call routing and handling
Set up LiveKit for real-time communication and SIP connectivity
Create and configure your LiveKit project
Configure SIP trunks to receive incoming calls
Set up rules for routing calls to AI agents
Generate and manage SIP authentication tokens
Understanding WebRTC and real-time audio processing
Configure AI models and voice synthesis for your agents
Connect OpenAI models for natural language processing
Configure voice models and speech synthesis
Create custom personalities and conversation flows
Write effective prompts for your AI agents
Handle multi-turn conversations and context
Monitor, analyze, and manage your voice calls
Monitor call metrics, duration, and success rates
Record calls and manage audio files
Analyze call transcripts and extract insights
Schedule automated calls and reminders
Monitor audio quality and connection stability
Create automated workflows and business processes
Create automated call workflows and sequences
Implement if/then logic in your call flows
Connect with external systems via webhooks
Collect and store data from conversations
Handle errors and edge cases in workflows
Key terms and concepts used throughout the VoiceAI platform
A virtual phone line that connects your system to the public telephone network via the internet. In VoiceAI, SIP trunks handle incoming and outgoing calls.
A SIP trunk configured to receive incoming calls. LiveKit inbound trunks route calls to your AI agents.
A LiveKit configuration that determines how incoming calls are routed to specific AI agents or workflows.
A webhook endpoint that Twilio calls when making outbound calls. It contains the logic for call handling.
An authentication token used to connect to LiveKit SIP services. Required for SIP trunk authentication.
An AI-powered conversational agent that can handle phone calls, understand speech, and respond naturally.
Instructions given to the AI model that define the agent's personality, capabilities, and behavior.
Real-time communication protocol that enables voice and video calls directly in web browsers.
A complete interaction between a caller and an AI agent, including audio, transcript, and metadata.
A series of automated steps that define how calls are processed, including decision points and actions.