Centerbeam AI

Developers

Integrate Intelligent Video or Voice Agents in days, not quarters

Nova SDK gives you everything needed to embed Intelligent Video or Voice Agents into any website or app, with full security, high-quality realistic voices, full integration stacks, and training, and realtime streaming.

1. Install SDK

npm install @centerbeam/nova-sdk

2. Create Agent

import { Agent } from '@centerbeam/nova-sdk'

const agent = new Agent({
  sdb: {
    persona: 'enterprise-support',
    guardrails: ['pii-filter', 'compliance'],
    tools: ['knowledge-base', 'ticket-system'],
    policies: { escalation: 'human-handoff' }
  }
})

3. Connect Knowledge

agent.connect({
  rag: {
    source: 'your-vector-store',
    memory: 'threaded',
    workflows: ['intake', 'escalation']
  },
  relay: 'websocket' // or 'webrtc'
})

4. Embed Voice Widget

// React component
<NovaVoiceWidget
  agent={agent}
  theme="dark"
  position="bottom-right"
  onConnect={() => console.log('Voice active')}
/>

5. Or ship natively on Apple platforms (iOS 17+ / macOS 14+)

// Add the NovaNanoBot Swift Package to your Xcode project:
//   File > Add Package Dependencies > Add Local
//   Point at: public/nanobot/SwiftUI/

import SwiftUI
import NovaNanoBot

struct ContentView: View {
  var body: some View {
    NanoBotView(config: NanoBotConfig(
      novaHost:    "https://<your-nova-host>",
      relayHost:   "wss://<your-relay-host>",
      assistantId: "<your-assistant-id>"
    ))
  }
}

// For custom layouts, compose NanoBotMicButton and
// NanoBotTranscriptPanel yourself instead of the bundled
// NanoBotView overlay. See public/nanobot/SwiftUI/README.md
// for Info.plist microphone setup and AEC notes.

Try It In-Browser

Interactive Postman wizard

Exercise the Nova API end-to-end from your browser — step through auth, session creation, tool wiring, and realtime events without leaving the docs. The wizard runs live calls against a sandbox assistant so you can validate your integration shape before writing any code.

WebSocket Relay

Bidirectional realtime communication with automatic reconnection and Safari compatibility.

Threaded Memory

Context-aware conversations with threaded memory architecture. Agents remember across sessions.

Realtime Streaming

Voice-to-voice streaming with sub-200ms latency. Interrupt handling and turn-taking built in.

Ready to deploy intelligent video or voice experiences?