Telemetry — Observability for AI and LLM apps

Monitor AI model calls, agent steps, and retrieval with token tracking, cost analysis, and latency metrics.