Streaming Chat

The SDK provides an async iterator pattern for consuming SSE streams — the most ergonomic way to handle real-time AI responses.

Async Iterator (recommended)

import { Senda } from '@senda/sdk';

const senda = new Senda({
  apiKey: 'sk_live_YOUR_KEY',
  baseUrl: 'https://senda.telar.ai/api'
});

for await (const event of senda.chat.stream('Explain quantum computing')) {
  switch (event.type) {
    case 'text':
      process.stdout.write(event.text!);
      break;
    case 'component':
      console.log('UI Widget:', event.component_name, event.props);
      break;
    case 'agent_info':
      console.log('Routed to agent:', event.agent_id);
      break;
    case 'done':
      console.log('\n[Stream complete]');
      break;
  }
}

Non-Streaming (simple)

If you don't need real-time display:

const response = await senda.chat.send('What is 2+2?');
console.log(response); // "2 + 2 = 4"

Continue a Conversation

Pass a chatId to maintain context:

const chatId = 'chat_abc123';

const response1 = await senda.chat.send('My name is Eddie', { chatId });
const response2 = await senda.chat.send('What is my name?', { chatId });
// → "Your name is Eddie"

Event Types

Type	Fields	Description
text	text	Streaming text token
agent_info	agent_id	Which agent was selected
component	component_name, props	Generative UI widget data
done	—	Stream complete
error	text	Error message