Skip to main content
New in Beta: Our Realtime Assistants API lets you create voice-enabled AI agents that can have natural conversations with users.

Quick Overview

Realtime Assistants combine UpliftAI’s speech technology with conversational AI to create voice agents that can:
  • 🎯 Respond in ~1 second end-to-end
  • 🗣️ Handle interruptions naturally
  • 🛠️ Execute custom tools on the user’s device
  • 🌍 Support multiple languages including Urdu, Sindhi, and Balochi
  • 🔄 Update behavior dynamically during conversations

Perfect for Voice Applications

If you’re using our Orator TTS API, Realtime Assistants take it to the next level by adding:
  • Full conversational capabilities
  • Context awareness
  • Tool execution
  • WebRTC streaming

Example Use Cases

  • Customer Support - 24/7 voice agents
  • E-commerce - Voice shopping assistants
  • Healthcare - Patient intake systems
  • Education - Interactive tutors

Get Started

View Full Documentation

Explore the complete Realtime Assistants documentation including tutorials, API reference, and SDK guides.

Quick Example

Create a voice assistant in minutes:
// 1. Create an assistant (via API)
const assistant = await createAssistant({
  name: 'Shopping Assistant',
  config: {
    agent: {
      instructions: 'Help users shop for products',
      initialGreeting: true
    },
    stt: { default: { provider: 'groq', model: 'whisper-large-v3' }},
    tts: { default: { provider: 'upliftai', voiceId: 'v_meklc281' }},
    llm: { default: { provider: 'groq', model: 'openai/gpt-oss-120b' }}
  }
});

// 2. Connect from your app
import { UpliftAIRoom } from '@upliftai/assistants-react';

<UpliftAIRoom token={sessionToken} serverUrl={wsUrl}>
  <YourAssistantUI />
</UpliftAIRoom>

Integration with Orator

Realtime Assistants use the same high-quality TTS voices from Orator, but add:
  • Automatic speech recognition (STT)
  • Conversation management
  • Tool calling capabilities
  • Real-time streaming via WebRTC
I