Vercel AI LLM analytics installation

  1. Install the PostHog SDK

    Required

    Setting up analytics starts with installing the PostHog SDK.

    Terminal
    npm install @posthog/ai posthog-node

  2. Install the Vercel AI SDK

    Required

    Install the Vercel AI SDK:

    Terminal
    npm install ai @ai-sdk/openai
    Proxy note

    These SDKs do not proxy your calls. They only fire off an async call to PostHog in the background to send the data.

    You can also use LLM analytics with other SDKs or our API, but you will need to capture the data in the right format. See the schema in the manual capture section for more details.

  3. Initialize PostHog and Vercel AI

    Required

    Initialize PostHog with your project API key and host from your project settings, then pass the Vercel AI OpenAI client and the PostHog client to the withTracing wrapper.

    TypeScript
    import { PostHog } from "posthog-node";
    import { withTracing } from "@posthog/ai"
    import { generateText } from "ai"
    import { createOpenAI } from "@ai-sdk/openai"
    

    const phClient = new PostHog(
      '<ph_project_api_key>',
      { host: 'https://us.i.posthog.com' }
    );
    

    const openaiClient = createOpenAI({
      apiKey: 'your_openai_api_key',
      compatibility: 'strict'
    });
    

    const model = withTracing(openaiClient("gpt-4-turbo"), phClient, {
      posthogDistinctId: "user_123", // optional
      posthogTraceId: "trace_123", // optional
      posthogProperties: { conversationId: "abc123", paid: true }, // optional
      posthogPrivacyMode: false, // optional
      posthogGroups: { company: "companyIdInYourDb" }, // optional
    });
    

    phClient.shutdown()

    You can enrich LLM events with additional data by passing parameters such as the trace ID, distinct ID, custom properties, groups, and privacy mode options.

  4. Call Vercel AI

    Required

    Now, when you use the Vercel AI SDK to call LLMs, PostHog automatically captures an $ai_generation event.

    This works for both text and image message types.

    TypeScript
    const { text } = await generateText({
      model: model,
      prompt: message
    });
    

    console.log(text)

    Note: If you want to capture LLM events anonymously, don't pass a distinct ID to the request. See our docs on anonymous vs identified events to learn more.

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

    PropertyDescription
    $ai_modelThe specific model, like gpt-5-mini or claude-4-sonnet
    $ai_latencyThe latency of the LLM call in seconds
    $ai_toolsTools and functions available to the LLM
    $ai_inputList of messages sent to the LLM
    $ai_input_tokensThe number of tokens in the input (often found in response.usage)
    $ai_output_choicesList of response choices from the LLM
    $ai_output_tokensThe number of tokens in the output (often found in response.usage)
    $ai_total_cost_usdThe total cost in USD (input + output)
    ...See full list of properties

  5. Verify traces and generations

    Checkpoint
    Confirm LLM events are being sent to PostHog

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


    LLM generations in PostHog
    Check for LLM events in PostHog

Community questions

Was this page useful?

Questions about this page? or post a community question.