Vercel AI SDK

XeroML integrates with the Vercel AI SDK through OpenTelemetry. Enable experimental_telemetry on your AI SDK calls and XeroML captures all model interactions automatically.

Installation

npm install ai @ai-sdk/openai @xeroml/tracing @xeroml/otel @opentelemetry/sdk-node

Setup

1. Initialize OpenTelemetry:

   import { NodeSDK } from "@opentelemetry/sdk-node";
   import { XeroMLSpanProcessor } from "@xeroml/otel";
   
   const sdk = new NodeSDK({
     spanProcessors: [new XeroMLSpanProcessor()],
   });
   sdk.start();

2. Enable telemetry on AI SDK calls:

   import { generateText } from "ai";
   import { openai } from "@ai-sdk/openai";
   
   const { text } = await generateText({
     model: openai("gpt-4o"),
     prompt: "Explain XeroML in one sentence.",
     experimental_telemetry: { isEnabled: true },
   });

With streamText

import { streamText } from "ai";
import { openai } from "@ai-sdk/openai";

const result = await streamText({
  model: openai("gpt-4o"),
  prompt: "Write a poem about observability.",
  experimental_telemetry: { isEnabled: true },
});

for await (const chunk of result.textStream) {
  process.stdout.write(chunk);
}

Adding Context

Wrap the AI SDK call in propagateAttributes() to attach user and session context:

import { propagateAttributes } from "@xeroml/tracing";
import { generateText } from "ai";
import { openai } from "@ai-sdk/openai";

async function handleRequest(prompt: string, userId: string, sessionId: string) {
  return await propagateAttributes({ userId, sessionId }, async () => {
    return await generateText({
      model: openai("gpt-4o"),
      prompt,
      experimental_telemetry: { isEnabled: true },
    });
  });
}

Next.js Integration

For Next.js App Router, set up XeroML in instrumentation.ts:

instrumentation.ts

export async function register() {
  if (process.env.NEXT_RUNTIME === "nodejs") {
    const { NodeSDK } = await import("@opentelemetry/sdk-node");
    const { XeroMLSpanProcessor } = await import("@xeroml/otel");

    const sdk = new NodeSDK({
      spanProcessors: [new XeroMLSpanProcessor()],
    });
    sdk.start();
  }
}

Then add experimental_telemetry: { isEnabled: true } to your AI SDK calls in route handlers or server actions.