Voice AI Agents API

The Agents API enables you to programmatically create, manage, configure, and execute autonomous Voice AI agents. Agents handle inbound and outbound telephone calls, interact with customers via speech, access contextual knowledge, and trigger custom backend APIs and MCP tools in real time.

+-------------------------------------------------------------+
|                      Voice AI Agent                         |
|  +----------------+  +----------------+  +---------------+  |
|  |  ASR / Speech  |  |   LLM Brain    |  |  TTS / Voice  |  |
|  |  (Deepgram/...) |  | (GPT-4o/Claude)|  |  (Cartesia)   |  |
|  +----------------+  +----------------+  +---------------+  |
|          |                   |                   |          |
|  +-------------------------------------------------------+  |
|  |             Capabilities & Real-Time Tools            |  |
|  |  * Calendar Booking   * Custom REST APIs   * MCP SSE  |  |
|  +-------------------------------------------------------+  |
+-------------------------------------------------------------+

Key Capabilities

Ultra-Low Latency Voice

Streaming audio pipelines powered by Cartesia and WebRTC deliver sub-second natural voice conversations.

Tool Calling & Webhooks

Agents invoke custom REST APIs during live phone calls to check order status, trigger actions, or fetch CRM data.

MCP Server Integration

Native support for Model Context Protocol (MCP) servers via Server-Sent Events (SSE) or HTTP transports.

Calendar Scheduling

Automated meeting and appointment booking directly into Google Calendar and Calendly.

Agent Lifecycle & States

Every Voice AI agent transitions through well-defined operational states:

Status	Description	Permitted Actions
`active`	Agent is published and operational	Receive inbound calls, execute campaigns, run tool calls
`inactive`	Agent is paused or draft	View and update configurations; incoming calls receive fallback message
`archived`	Agent has been retired	Read-only access for historic call logs and analytics

Authentication

All Agents API endpoints require a valid organization API key or Bearer JWT token passed in the Authorization header:

Authorization: Bearer <your-api-key>

Always execute server-side agent creation and mutation requests from secure environments. Never expose secret keys in client-side bundles.

Available Endpoints

Explore the full suite of Agents API methods available in Sayvy AI:

Method	Endpoint	Description
GET	`/api/v1/agents/{agent_id}`	Retrieve detailed configuration, status, and tasks for an agent
POST	`/api/v1/agents`	Create a new Voice AI prompt agent or static bot
GET	`/api/v1/agents`	List all voice agents for your organization
PUT	`/api/v1/agents/{agent_id}`	Update complete configuration of an existing agent
DELETE	`/api/v1/agents/{agent_id}`	Permanently delete an agent

‹ Previous

Get agent

Retrieve detailed Voice AI agent informati…

Next ›

Need help?

Check out our quickstart guide or comprehensive API reference to get your first voice agent running in minutes.View guides →

Overview

Agents

Campaigns

Contacts

Templates

Knowledge Base

Webhooks

Call Logs

Quick Replies

Inbox

Agents Overview

Voice AI Agents API

Key Capabilities

Ultra-Low Latency Voice

Tool Calling & Webhooks

MCP Server Integration

Calendar Scheduling

Agent Lifecycle & States

Authentication

Available Endpoints

Need help?

​Voice AI Agents API

​Key Capabilities

Ultra-Low Latency Voice

Tool Calling & Webhooks

MCP Server Integration

Calendar Scheduling

​Agent Lifecycle & States

​Authentication

​Available Endpoints

Need help?

Voice AI Agents API

Key Capabilities

Agent Lifecycle & States

Authentication

Available Endpoints