A single API to create, deploy, and scale voice agents.
Ship conversational AI that sounds human — in minutes, not months.
import VoiceForge from 'voiceforge';
const agent = await VoiceForge.create({
name: 'Support Agent',
voice: 'nova',
model: 'gpt-4o',
prompt: 'You are a friendly support agent for Acme Inc...',
tools: [lookupOrder, transferCall],
onEvent: (e) => console.log(e),
});
// Deploy to a phone number instantly
await agent.deploy({ phone: '+1-800-555-0199' });One npm install, one API key. Create a production-ready voice agent in five minutes. No telephony expertise needed.
Plug in GPT-4o, Claude, Gemini, or your own fine-tuned model. Swap providers with a single config change — zero code rewrite.
Give your agent real power. Connect databases, CRMs, calendars, or any REST / GraphQL API as callable tools mid-conversation.
Voice runs on our worldwide edge for sub-500 ms round-trip latency. Your callers never notice they're talking to an AI.
Full transcription, sentiment analysis, and conversation metrics streamed in real time. Debug and improve from day one.
SOC 2 Type II, HIPAA-ready, end-to-end encryption. Your data never trains our models. Audit logs for every call.
24/7 phone support that resolves 80% of tickets without a human.
Voice agents that check availability and book appointments in real time.
Let customers check, modify, or cancel orders with a single phone call.
HIPAA-compliant agents that collect symptoms and route patients.
Secure voice agents for balance checks, fraud alerts, and loan apps.
Conduct phone surveys with natural conversation and real-time analytics.
Twilio, SIP trunks, WebRTC, SRTP — we abstract it all. You write business logic; we handle the voice infra so you never touch a codec.
Handle 1 call or 10,000 concurrent calls with the same code. Our edge auto-scales so you never provision servers or worry about capacity.
Full transcripts, recordings, and metadata. Export everything. No vendor lock-in — bring your own models, your own numbers, your own stack.
Real-time transcription, latency graphs, cost breakdowns, and conversation replays — all in one dashboard. Know exactly what your agents are saying and how they're performing.
Most developers ship their first agent in under 5 minutes. Install the SDK, pass your prompt and tools, and call .deploy(). That's it.
We support 30+ languages and 50+ voices out of the box. You can also clone a custom voice or bring your own TTS provider.
Yes. Bring any OpenAI-compatible API, or host your own model. We're model-agnostic — swap providers with a one-line config change.
Yes. Our Enterprise plan includes BAA, SOC 2 Type II certification, end-to-end encryption, and audit logging for every call.
We'll notify you before you hit your limit. You can upgrade instantly — no downtime, no data loss. Overages on Pro are billed at the same per-minute rate.
Absolutely. No contracts, no commitments. Cancel from the dashboard and you won't be charged again.
We got tired of stitching together STT, LLM, and TTS pipelines just to answer a phone call. VoiceForge is the API we wished existed — one call to create an intelligent voice agent, with the reliability and latency production demands.
Join hundreds of startups building the next generation of voice experiences.