xAI ships a no-code voice agent builder
xAI's Voice Agent Builder beta configures production phone agents on Grok Voice at $0.05/min, with telephony, knowledge retrieval, tools, and MCP support.
xAI released Voice Agent Builder in beta on July 1: a no-code platform for configuring production voice agents on Grok Voice. Agents are billed at the API rate of $0.05 per minute of audio with voices included; telephony on a provisioned number adds $0.01 per minute. Each account gets a free phone number, existing numbers connect over SIP, and custom clients can attach over WebSocket.
The pitch is consolidation. Most voice stacks stitch together three APIs for speech-to-text, a language model, and text-to-speech, often from different vendors, and every hop adds latency and a failure mode. Voice Agent Builder runs one speech-to-speech path coupled to the Grok Voice model, with telephony, knowledge retrieval, tools, guardrails, MCP servers, and observability in one interface.
Setup is a plain-language description of how calls should flow, plus attached documents organized into shareable collections. Agents can act during calls through connectors, scheduling in Google Calendar or Outlook and sending confirmations by email. Voices come from an 80-plus library or a clone built from about two minutes of audio. On xAI’s own tau-voice Bench, Grok Voice Think Fast 1.0 scores 67.3% against 43.8% for Gemini 3.1 Flash Live and 35.3% for GPT Realtime 1.5; treat vendor benchmarks as a starting point, not a verdict.
Independent evaluation is exactly the gap Coval raised $28M to fill. For the wider voice stack picture, see Cerebras and Hugging Face’s open Gemma 4 voice pipeline, and read the customer support guide before you put an agent on a phone line.