​
Overview

The Vercel AI SDK provides a consistent API for interacting with various AI providers. Gentrace integrates seamlessly to provide automatic tracing and error analysis for your AI application.

​
Prerequisites

This guide assumes you have already setup an interaction as shown in our quickstart.

​
Instrumentation

The AI SDK natively exports to OpenTelemetry, and Gentrace can consume this telemetry automatically. To enable telemetry, set the experimental_telemetry flag to true in your AI SDK calls. For example:
The AI SDK call must be inside the callstack of an interaction for Gentrace to capture the trace.
const { text } = await generateText({
  // Enable telemetry
  experimental_telemetry: {
    isEnabled: true,
  },
  model: openai('gpt-4.1-nano'),
  prompt: `Write a haiku about ${topic}`,
});

​
Full example

.env
GENTRACE_API_KEY=your-gentrace-api-key
GENTRACE_PIPELINE_ID=your-pipeline-id
OPENAI_API_KEY=your-openai-api-key

import { init, interaction } from 'gentrace';
import { openai } from '@ai-sdk/openai';
import { generateText } from 'ai';

init({
  apiKey: process.env.GENTRACE_API_KEY,
});

const writePoem = interaction('write-poem', async (topic: string) => {
  const { text } = await generateText({
    model: openai('gpt-4.1-nano'),
    prompt: `Write a haiku about ${topic}`,
    experimental_telemetry: {
      isEnabled: true,
    },
  });
  return text;
});

const main = async () => {
  const result = await writePoem('quantum computing');
  console.log(result);
};

main();
For an example in our SDK, see the OpenAI AI SDK example on GitHub.