What is Langfuse?

Langfuse is an open-source LLM engineering platform that provides observability, prompt management, evaluation, and experimentation tools for AI agents and applications. It enables teams to trace, monitor, and debug LLM calls, manage prompts, run evaluations, and improve AI systems from prototype to production scale. The platform supports self-hosting and cloud deployment with a focus on open standards and data portability.

How to use Langfuse?

To get started with Langfuse, create an account or self-host the platform, then obtain API credentials from project settings. Install the Langfuse SDK (Python, JS/TS, etc.) or use OpenTelemetry instrumentation to ingest traces. Choose your framework (OpenAI SDK, LangChain, Vercel AI SDK, etc.) and follow the integration guide to automatically record model calls. Use the dashboard to monitor traces, manage prompts, run evaluations, and set up experiments.

Core features of Langfuse

LLM Observability: hierarchical traces capturing every LLM call, tool invocation, and retrieval step
Evaluation: LLM-as-a-judge, heuristic functions, and human review with production data
Prompt Management: separate prompts from code with one-click deployments and rollbacks
Playground: test prompts on real production inputs and compare models side-by-side
Experiments: define test cases and compare results side by side
Human Annotation: collaborative human-in-the-loop workflows for review and golden datasets
Cost & Latency Monitoring: dashboards and automated alerts for cost, latency, and quality
Open platform: MIT license, self-hostable, REST APIs, and 100+ integrations