Grok Voice Agent Builder

What it does

Voice Agent Builder is a no-code platform for creating phone and voice agents on Grok Voice, xAI's speech-to-speech model. You configure an agent's instructions and voice, add a knowledge source, tools and guardrails, and attach a phone connection. The result is an agent that talks with callers in real time. Telephony, knowledge retrieval, tools, guardrails, MCP support and observability are available in one place.

In practice, you can bring existing phone numbers over SIP, wire tools to your own APIs and MCP servers, or connect your own client over WebSocket. Instead of chaining separate speech-to-text, language-model and text-to-speech services, it uses one speech-to-speech path. xAI says Grok Voice was trained on difficult calls with noisy audio, accents and interruptions, and supports 25+ languages.

It is aimed at operators and developers who want high-volume production voice agents without building the surrounding stack. The Builder is in beta, so features and terms may change. Developers using the underlying Speech to Speech API need an xAI API key, and browser clients should use ephemeral tokens rather than raw keys. Request a quote or check the vendor's pricing page for Builder costs.

What it can do

• Configure voice agents without writing code
• Built-in telephony, knowledge retrieval, guardrails and observability
• Connect tools to your own APIs and MCP servers
• Bring existing phone numbers over SIP
• Connect a custom client over WebSocket
• Speech-to-speech model path for 25+ languages

Works with

• SIP
• MCP servers
• WebSocket
• LiveKit

Example workflows

Phone support line

Inputs: an existing phone number, agent instructions, knowledge content and tool connections to your APIs. Steps: configure the agent in the builder, attach the number over SIP, add guardrails and tools, then review calls through observability. Output: a voice agent that answers inbound calls.

Custom client voice app

Inputs: an xAI API key, session settings, and an MCP server or function tools. Steps: connect your own client over WebSocket, set the voice and instructions, and handle tool calls. Output: a real-time voice assistant inside your own application.

Before you choose

• The Builder is in beta, so capabilities and terms may change
• Request a quote or check the vendor's pricing page for Builder costs; the underlying voice API is priced separately
• Browser clients must use ephemeral tokens, not raw API keys

Listed under

• Support

Get started

Announcement offers a free try; terms unconfirmed. Page: x.ai/voice.

1. Open the Voice Agent Builder from x.ai/voice and sign in
2. Define the agent's instructions, knowledge and tools, or connect MCP servers
3. Attach a phone number via SIP or connect your own client over WebSocket

The voice agent builder offers beta access. Voice APIs have separate usage charges; confirm builder entitlements and the applicable voice rate table before deployment.

We couldn’t find published prices for Grok Voice Agent Builder. See their pricing page ↗

Sources for this overview

Product pages, documentation, and pricing checked 28 Oct 2026. Each source supports part of the overview; availability and prices can change.

• Introducing the Voice Agent Builder | SpaceXSI ↗x.ai
• Speech to Speech | SpaceXSI Docs ↗docs.x.ai

Company website

x.ai: product, pricing and policy pages

Web archive

When the site first appeared

Web search

Confirms the official site

Social accounts

Accounts the company links, and its posts on X

Human review

Checked by a person

knowledge_base

Last checked 28 Oct 2026. Every field keeps its source and date; unknown means we couldn’t confirm it.