PostHog

The PostHog integration lets you visualize Muster (langfuse) LLM metrics inside PostHog dashboards — and correlate LLM activity with your wider product analytics.

How it works

Muster batches aggregated events to PostHog hourly. Three event types are sent:

• Trace events — cost, latency, observation counts.
• Generation events — model usage, token counts, performance metrics.
• Score events — user feedback and quality assessments.

Users are matched between platforms via Muster user_id ↔ PostHog $distinct_id.

Setup

1. Enable the integration in your Muster project settings — provide your PostHog hostname and API key.
2. Choose the export source — new integrations default to "Enriched observations (recommended)" for best performance. Legacy configurations can migrate by switching sources incrementally.
3. Build dashboards in PostHog using the pre-built "AI Metrics" template, which includes common Muster (Langfuse) metrics out of the box.

Use cases

• Correlate LLM feature usage with retention and conversion.
• Slice cost and latency by user cohorts defined in PostHog.
• Trigger PostHog alerts on Muster score thresholds.

See also

• Upstream Langfuse PostHog docs
• Mixpanel