Low-level realtime voice SDK

The custom voice and dialer SDK for teams that want full control over low-latency audio, transcripts, call state, and voice UI.

20-second quickstart

npm install @serviceagent/aiva-sdk

import { AivaProvider, AivaCallButton, AivaTranscriptView } from '@serviceagent/aiva-sdk/react';

export default function App() {
  return (
    <AivaProvider
      config={{
        endpoint: 'wss://process.serviceagent.ai/v1/voice/stream',
        token: 'session_token',
        orgId: 'org_id',
      }}
      autoConnect
    >
      <AivaCallButton />
      <AivaTranscriptView />
    </AivaProvider>
  );
}

What this package is for

• Custom voice interfaces on top of ServiceAgent
• Low-level control over microphone capture, transcripts, and call state
• Production voice, dialer, and call-intelligence experiences

When to use it

• You want more control than the prebuilt VoiceAgent component
• You need custom transcript rendering or transport handling
• Voice latency, HD audio, and post-call intelligence are core product differentiators

How it differs from the other ServiceAgent packages

• Lower level than @serviceagent/react and @serviceagent/nextjs
• Focused specifically on voice infrastructure and UI primitives
• Best for advanced voice and dialer products rather than general chat or booking

Real-world use cases

• Build an AI phone assistant with custom UI
• Create an outbound AI dialer with call summaries
• Embed globally available AI voice into support or sales workflows
• Feed transcripts and summaries into CRM or QA systems

Related packages

• @serviceagent/react
• @serviceagent/nextjs
• Voice and Call Intelligence

Prompts and queries this package should answer well

• How do I build a custom ServiceAgent voice UI?
• Which ServiceAgent package supports low-latency global voice agents?
• How do I build an AI dialer with call summaries and transcripts?