Core

OrgaAI static class for SDK initialization and configuration

OrgaAI

Import

import { OrgaAI } from '@orga-ai/react';

Props

Option	Type	Required	Default	Description
`enableTranscriptions`	`boolean`	No	`false`	Enables or disables transcription of conversation items, which are returned to the client via `useOrgaAI` hook.
`logLevel`	`"debug" \| "info" \| "warn" \| "error" \| "none"`	No	`”none”`	Logging verbosity.
`modalities`	`Modality[]`	No	`[“audio”]`	Array of allowed interaction modalities ("audio", "video") for the session. Controls availability of audio and video streams in `useOrgaAI`
`timeout`	`number`	No	`30000`	Timeout for requests, in milliseconds.
`sessionConfigEndpoint`	`string`	Yes*	-	URL to your backend endpoint for fetching ephemeral tokens and ICE servers.
`fetchSessionConfig`	`() => Promise<{ ephemeralToken: string; iceServers: RTCIceServer[] }>`	Yes*	-	Custom function to fetch ephemeral token and ICE servers.
`model`	`OrgaAIModel`	No	`orga-1-beta`	Model to use (see Types for available models).
`voice`	`OrgaAIVoice`	No	`alloy`	Voice to use (see Types for available voices).
`temperature`	`number`	No	`0.5`	Sampling temperature (randomness). Must be between allowed min/max.
`maxTokens`	`number`	No	-	Maximum tokens for responses. Must be between 100 and 1000.
`instructions`	`string`	No	-	Custom instructions to modify OrgaAI’s behavior, such as response tone or context. Example: 'Respond in a friendly tone.'
`history`	`boolean`	No	`true`

Either sessionConfigEndpoint or fetchSessionConfig is required

Usage

import { OrgaAI } from '@orga-ai/react';

OrgaAI.init({
  fetchSessionConfig: async () => {
    const response = await fetch('https://your-backend.com/api/orga-client-secrets');
    const { ephemeralToken, iceServers } = await response.json();
    return { ephemeralToken, iceServers };
  },
  model: 'orga-1-beta',
  voice: 'alloy',
  temperature: 0.7,
  enableTranscriptions: true,
  modalities: ['audio', 'video'],
  instructions: 'Respond in a friendly tone.',
  logLevel: 'debug',
});