Skip to main content

​
Vercel AI SDK Integration

The TypeScript SDK supports Vercel AI SDK via OpenTelemetry spans created by the AI SDK when you pass a tracer. aiTelemetry() is a convenience helper that returns a pre-configured telemetry object compatible with the AI SDK’s experimental_telemetry option. It binds the current Lucidic session context so your generateText/streamText calls are captured as Lucidic events without additional setup.

​
Setup

import { init, aiTelemetry } from 'lucidicai';
import { generateText, streamText, tool } from 'ai';
import { openai } from '@ai-sdk/openai';
import { z } from 'zod';

// Initialize Lucidic (non-global provider)
await init({ sessionName: 'Vercel AI SDK Demo' });

// Regular chat
const res = await generateText({
  model: openai('gpt-4o-mini'),
  prompt: 'Say "hello from vercel ai"',
  experimental_telemetry: aiTelemetry(),
});
console.log(res.text);

​
Streaming

const { textStream } = await streamText({
  model: openai('gpt-4o-mini'),
  prompt: 'Count to 5:',
  experimental_telemetry: aiTelemetry(),
});
let full = '';
for await (const chunk of textStream) full += chunk;
console.log('Stream:', full);

​
Tools

const tools = {
  add: tool({
    description: 'Add two numbers',
    inputSchema: z.object({ a: z.number(), b: z.number() }),
    execute: async ({ a, b }: { a: number; b: number }) => a + b,
  }),
} as const;

const res = await generateText({
  model: openai('gpt-4o-mini'),
  system: 'You can use tools when needed.',
  prompt: 'What is 2+3? Use the add tool.',
  tools,
  experimental_telemetry: aiTelemetry(),
});
console.log(res.text);

​
Notes

  • Pass experimental_telemetry: aiTelemetry() to every AI SDK call to route spans to Lucidic (recommended for most use cases)
  • Use init({ sessionName }) first; session context is attached via AsyncLocalStorage
  • Images in messages are handled by the AI SDK and extracted by Lucidic

​
See Also