Langfuse is an open-source LLM engineering platform that helps teams collaboratively debug, analyze, and iterate on their LLM applications.

---

We use Langfuse for tracing and prompt management in Zammad-AI, RIS-KI and the Dienstleistungsfinder KI-Suche. MUCGPT supports optional observability with Langfuse.

Everything runs on open standards: all traces are OpenTelemetry spans with the GenAI semantic conventions (model, inputs, outputs, token usage), ingested via the open OTLP protocol. Since v4, Langfuse ingests OTLP chunks natively without a reverse proxy — traces and observations now arrive via OTLP instead of the legacy ingestion API, which still handles other event types such as scores and SDK logs. This keeps our observability vendor-neutral: identical telemetry can also be consumed by any other OTLP-capable backend we may want to use in the future.