Skip to main content

Overview

This example uses InferenceRuntime directly and shows the Langfuse connection inline. It is useful when you want visibility into the raw LLM call path without the additional StructuredOutput layer. Key concepts:
  • explicit LangfuseConfig / LangfuseHttpTransport / LangfuseExporter setup
  • InferenceRuntime: direct Polyglot runtime for inference calls
  • PolyglotTelemetryProjector: maps inference lifecycle events
  • HttpClientTelemetryProjector: captures transport spans
  • Telemetry::flush(): pushes the final batch to Langfuse

Example