Vercel AI SDK LLM analytics installation
Contents
- 1
Install dependencies
RequiredInstall the PostHog AI package, the Vercel AI SDK, and the OpenTelemetry SDK.
No proxyThese SDKs do not proxy your calls. They only send analytics data to PostHog in the background.
- 2
Set up the OpenTelemetry exporter
RequiredInitialize the OpenTelemetry SDK with PostHog's
PostHogTraceExporter. This sendsgen_ai.*spans directly to PostHog's OTLP ingestion endpoint. PostHog converts these into$ai_generationevents automatically. - 3
Call Vercel AI with telemetry enabled
RequiredPass
experimental_telemetryto your Vercel AI SDK calls. Theposthog_distinct_idmetadata field links events to a specific user in PostHog.Note: If you want to capture LLM events anonymously, omit the
posthog_distinct_idmetadata field. See our docs on anonymous vs identified events to learn more.You can expect captured
$ai_generationevents to have the following properties:Property Description $ai_modelThe specific model, like gpt-5-miniorclaude-4-sonnet$ai_latencyThe latency of the LLM call in seconds $ai_time_to_first_tokenTime to first token in seconds (streaming only) $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 - 4
Next steps
RecommendedNow that you're capturing AI conversations, continue with the resources below to learn what else LLM Analytics enables within the PostHog platform.
Resource Description Basics Learn the basics of how LLM calls become events in PostHog. Generations Read about the $ai_generationevent and its properties.Traces Explore the trace hierarchy and how to use it to debug LLM calls. Spans Review spans and their role in representing individual operations. Anaylze LLM performance Learn how to create dashboards to analyze LLM performance.

