Vercel AI

2

Set up the OpenTelemetry exporter

Required

Initialize the OpenTelemetry SDK with PostHog’s PostHogSpanProcessor. This sends gen_ai.* spans directly to PostHog’s OTLP ingestion endpoint. PostHog converts these into $ai_generation events automatically.

import { NodeSDK } from '@opentelemetry/sdk-node'
import { resourceFromAttributes } from '@opentelemetry/resources'
import { PostHogSpanProcessor } from '@posthog/ai/otel'
const sdk = new NodeSDK({
  resource: resourceFromAttributes({
    'service.name': 'my-app',
  }),
  spanProcessors: [
    new PostHogSpanProcessor({
      apiKey: '<ph_project_token>',
      host: 'https://us.i.posthog.com',
    }),
  ],
})
sdk.start()

3

Call Vercel AI with telemetry enabled

Required

Pass experimental_telemetry to your Vercel AI SDK calls. The posthog_distinct_id metadata field links events to a specific user in PostHog.

import { generateText } from 'ai'
import { openai } from '@ai-sdk/openai'
const result = await generateText({
  model: openai('gpt-5-mini'),
  prompt: 'Tell me a fun fact about hedgehogs.',
  experimental_telemetry: {
    isEnabled: true,
    functionId: 'my-ai-function',
    metadata: {
      posthog_distinct_id: 'user_123', // optional
    },
  },
})
console.log(result.text)

Note: If you want to capture LLM events anonymously, omit the posthog_distinct_id metadata field. See our docs on anonymous vs identified events (opens in a new tab) to learn more.

You can expect captured $ai_generation events to have the following properties:

Verify traces and generations

Recommended

Confirm LLM events are being sent to PostHog

Let’s make sure LLM events are being captured and sent to PostHog. Under AI Observability, you should see rows of data appear in the Traces and Generations tabs.

Check for LLM events in PostHog (opens in a new tab)

4

Next steps

Recommended

Now that you’re capturing AI conversations, continue with the resources below to learn what else AI Observability enables within the PostHog platform.