llm-analytics-setup

Instrument LLM analytics across providers by wrapping your existing LLM SDK with PostHog instrumentation to automatically emit trace and generation events without hand-wiring each provider.

PostHog LLM analytics instrumentation wraps your LLM SDK calls to automatically emit standard events like $ai_generation and $ai_trace, enabling you to track and analyze AI generations within PostHog dashboards.

Yes, PostHog instrumentation works with Anthropic, Google Gemini, Cohere, Mistral, OpenAI, and Azure OpenAI, providing dedicated reference guides for each provider to standardize SDK integration.

Verify LLM traces and generations in PostHog by following the included verification workflow guidance, which directs you to check PostHog dashboards to confirm that automatic generation and trace events are properly captured.

No, you do not need to hand-wire tracing for every provider, because PostHog offers provider-agnostic instrumentation that wraps your LLM SDKs to standardize analytics across OpenAI, Anthropic, Cohere, and others.

The best way to standardize AI generation tracking across different LLM SDKs is to wrap them with PostHog's provider-agnostic instrumentation, which automatically emits consistent trace and generation events for all supported providers.