For teams building realtime voice assistants, AI Gate issues ephemeral WebRTC credentials, runs TURN globally, and keeps your OpenAI keys sealed in our vault so you can focus on product experience.
We are in private beta and review every request within 48–72 hours to co-design the roadmap with early adopters.
Free sandbox stays available while we approve paid plans • No credit card required to start
We are co-building AI Gate with teams shipping realtime voice AI. Join the private beta to influence the roadmap, secure grandfathered pricing, and get white-glove support while we scale.
Starter ($29/mo) and Growth ($99/mo) unlock before public launch with grandfathered pricing.
You talk directly with the team shipping AI Gate to keep your roadmap in sync with ours.
Hands-on onboarding, billing checks, and integration reviews with 24–48 hour response time.
We approve applications manually to protect capacity and make sure your use case gets the support it needs.
We publish our plans so you know exactly what is live, what ships next, and how we reach general availability.
Everything here is running today for our sandbox and beta partners—no aspirational screenshots.
Web frontend and API are deployed on Cloudflare Workers with mirrored staging environments.
Keys stay encrypted in D1 + KV; worker-to-worker routing keeps traffic off shared infrastructure.
v1.2.0-beta landed Oct 9 with worker fixes, beta application flow, and Stripe environment parity.
Try the sandbox, review the docs, and follow our changelog while we bring more teams into the beta.
The same managed infrastructure we operate every day for our beta partners
Skip the boilerplate. We manage session lifecycle, WebRTC proxying, and TURN configuration for every connection.
Bring your OpenAI keys—our vault keeps them encrypted at rest and never exposes them to shared infrastructure.
Get temporary credentials in milliseconds. Each session gets isolated, time-limited access.
TypeScript and Python SDKs with full type safety. 5-minute quickstart to production.
Deployed on Cloudflare Workers. Low latency worldwide with automatic scaling.
Battle-tested infrastructure with monitoring, rate limiting, and quota management built-in.
Three simple steps to production
Sign up and add your OpenAI API key. It is encrypted in our vault and never leaves our infrastructure.
Add our SDK to your app. TypeScript or Python - takes less than 5 minutes.
Create sessions, connect via WebRTC, and ship your voice AI app. We handle the rest.
Full-featured SDKs for TypeScript and Python. Get type safety, auto-completion, and excellent documentation out of the box.
import { AIGateClient } from '@ai-gate/sdk';
const client = new AIGateClient({
projectId: 'your-project-id',
});
// Create a voice session
const session = await client.createSession({
voice: 'echo',
instructions: 'You are a helpful assistant',
});
// Connect with WebRTC
const pc = new RTCPeerConnection({
iceServers: session.turn_credentials.iceServers
});
// Start talking! 🎉Built in the open with our early access community
We never see your OpenAI keys or audio data. Everything is encrypted in transit and at rest.
Your security is our top priority.
AES-256 encryption at rest for all API keys
We can't read your keys or access your audio
Security audit in progress (Q2 2025)
EU hosting available on request
Join the beta, help shape the roadmap, and become one of the first teams we feature here.
We are onboarding a focused group of teams and will publish their results once they are live in production.
Every feature—including vault-encrypted BYOK and latency benchmarks—is documented as it ships, not before.
Design partners get direct access to the engineers building AI Gate for feedback and integration help.
We keep feedback loops short so you never wonder what is happening with your voice stack.
support@ai-gate.dev • responses within 24–48 hours during beta.
Apply at /beta; we review every request manually to keep capacity in check.
We publish CURRENT_STATUS.md after each deploy so you always know what shipped.
View live statusTell us about your voice assistant and we will help you launch without rebuilding infrastructure from scratch.
We review beta requests within 48–72 hours • Free sandbox stays open while we scale paid plans