PostHog

Analyze LLM usage behavior in product analytics workflows

PostHog broadcast emits LLM analytics events so you can tie model spend and quality to feature adoption, user cohorts, and funnel performance.

broadcast-posthog-config.json

{
  "destination": "posthog",
  "enabled": true,
  "api_key": "phc_...",
  "endpoint": "https://us.i.posthog.com",
  "sampling_rate": 1
}

cURL

curl https://api.therouter.ai/v1/chat/completions   -H "Authorization: Bearer $THEROUTER_API_KEY"   -H "Content-Type: application/json"   -H "X-TheRouter.ai-Broadcast: true"   -d '{
    "model":"openai/gpt-4.1-mini",
    "messages":[{"role":"user","content":"Recommend an upgrade plan"}],
    "user":"user_88",
    "trace":{"trace_name":"PostHog Event Test","feature":"upsell-assistant"}
  }'

Verification

In PostHog LLM analytics, filter by `$ai_user` or your custom `feature` property to validate event flow.