Skip to main content

​
Overview

This guide shows you how to integrate Latitude Telemetry into an application that uses Vercel AI SDK v7.
Using Vercel AI SDK v6? See the Vercel AI SDK guide instead — its telemetry setup is different.
In v7, OpenTelemetry collection moved out of the ai package into the separate @ai-sdk/otel package, and it is now opt-out: once you register a telemetry integration, every AI SDK call emits telemetry by default — there is no per-call experimental_telemetry flag anymore. Latitude ingests those spans without a provider-specific instrumentation entry.
You’ll keep calling the Vercel AI SDK exactly as you do today. Telemetry simply observes and enriches those calls.
The Vercel AI SDK integration is TypeScript only.

​
Requirements

  • A Latitude account and API key
  • A Latitude project slug
  • A Node.js project that uses Vercel AI SDK v7 (ai@7) and the matching @ai-sdk/otel

​
Steps

1

Install

Install Latitude Telemetry and the AI SDK v7 OpenTelemetry package.
npm install @latitude-data/telemetry @ai-sdk/otel
2

Initialize and use

Initialize Latitude without an instrumentations array. Then register the AI SDK OpenTelemetry integration once, after constructing Latitude — it uses the global tracer provider that Latitude registered, so spans flow to Latitude automatically. No per-call flag is needed.
import { Latitude, capture } from "@latitude-data/telemetry"
import { generateText, registerTelemetry } from "ai"
import { OpenTelemetry } from "@ai-sdk/otel"
import { openai } from "@ai-sdk/openai"

const latitude = new Latitude({
  apiKey: process.env.LATITUDE_API_KEY!,
  project: process.env.LATITUDE_PROJECT_SLUG!,
})

// Register once, after Latitude. All AI SDK calls now emit telemetry (opt-out).
registerTelemetry(new OpenTelemetry())

await capture("generate-support-reply", async () => {
  const { text } = await generateText({
    model: openai("gpt-4o"),
    prompt: "Hello",
  })
  return text
})

await latitude.shutdown()
The recommended integration is OpenTelemetry, which emits standard OpenTelemetry GenAI semantic-convention spans. @ai-sdk/otel also exports LegacyOpenTelemetry, which emits the older ai.* spans (same as v6). Latitude ingests both.

​
Next.js

Register the integration in your instrumentation.ts, alongside your OpenTelemetry provider setup:
instrumentation.ts
import { registerOTel } from "@vercel/otel"
import { registerTelemetry } from "ai"
import { OpenTelemetry } from "@ai-sdk/otel"

export function register() {
  registerOTel({ serviceName: "my-ai-app" })
  registerTelemetry(new OpenTelemetry())
}

​
Opting out

Telemetry is opt-out. To disable it for a specific call, set telemetry: { isEnabled: false }. To disable it globally, don’t register any telemetry integration.

​
Seeing Your Traces

Once connected, traces appear automatically in Latitude:
  1. Open your project in the Latitude dashboard
  2. Each execution shows input/output messages, model, token usage, latency, and errors