xAI has rolled out Voice Agent Builder in beta, giving businesses and developers a no-code way to create Grok-powered voice agents for customer support, sales, scheduling, and other call-based workflows.
TL;DR
- xAI launched Voice Agent Builder in beta on July 1, 2026.
- Users can build voice agents in under two minutes without code.
- It supports 25+ languages, 80+ built-in voices, custom voice cloning, and browser testing.
- Pricing starts at $0.05 per minute, with telephony on a free provisioned number costing $0.01 per minute.
xAI has launched Voice Agent Builder, a no-code platform that lets users create and deploy AI-powered voice agents on Grok Voice in under two minutes. The tool is currently available in beta and is aimed at operators and developers who want production-ready voice agents without building the underlying voice stack from scratch.
“Today, we’re announcing Voice Agent Builder in beta,” xAI said in its official announcement, describing it as “a no-code platform to configure production voice agents on Grok Voice.”
The platform is designed for customer support, sales, personal assistance, scheduling, and similar voice-led workflows. Users can describe how calls should flow in plain language, attach documents, connect tools, add guardrails, and test the agent directly in the browser before deployment.
xAI says the product brings telephony, knowledge retrieval, tools, guardrails, Model Context Protocol integrations, and observability into one interface. This matters because many voice AI systems require separate services for speech-to-text, a language model, and text-to-speech, which can add latency, cost, and failure points.
The company claims Grok Voice is built for real-world calls, including noisy audio, interruptions, strong accents, ambiguous requests, and callers who change their minds mid-sentence. Voice Agent Builder supports more than 25 languages and lets businesses upload documents in formats such as plain text, Markdown, Word, PowerPoint, Excel, HTML, JSON, and others, so agents can retrieve relevant information during live calls.
The tool also integrates with Gmail, Google Calendar, Outlook Calendar, Linear, Notion, Google Drive, OneDrive, web search, X search, and custom APIs or MCP servers. For support or sales lines, this means an agent could check order status, book appointments, create tickets, issue refunds, send confirmations, or transfer a call to a human when needed.
Voice options are another major part of the rollout. xAI says users can choose from more than 80 built-in voices or clone a brand voice from around two minutes of audio. Each account gets a free phone number, while businesses can also connect existing numbers through direct SIP from major telephony providers.
Topics For More Insights
For reviewing performance, every call is recorded and transcribed. Businesses can play back audio, read transcripts, and see what tools the agent used during the conversation. xAI also says guardrails can prevent agents from doing things such as reading card numbers aloud or going off script.
The pricing is usage-based. xAI says agents are billed at the Voice API rate, currently $0.05 per minute of audio, with voices included and no separate platform fee. Telephony on a free provisioned number costs an additional $0.01 per minute. xAI’s pricing documentation lists realtime Voice API usage at $0.05 per minute, text-to-speech at $15 per 1 million characters, and speech-to-text at $0.10 per hour for REST or $0.20 per hour for streaming.
The rollout builds on xAI’s earlier Grok Voice Think Fast 1.0 release, which the company described as its flagship voice model for complex, multi-step customer support, sales, and enterprise applications. xAI said that Grok Voice has already powered Starlink’s phone sales and customer support experience, claiming a 20% conversion rate for sales inquiries and a 70% autonomous resolution rate for support inquiries.
With this beta release, xAI is making a clear push beyond chatbot-style AI and into enterprise voice automation, where speed, cost, integrations, compliance controls, and human-like conversation quality are all becoming competitive battlegrounds.
