Guides

Capture explicit and implicit user feedback as scores on your traces, and use it to surface the responses worth reviewing.

Review traces to classify issues like hallucinations, irrelevance, and formatting errors. Turn raw logs into actionable insights.

Automated evaluators give you a scalable way to measure and monitor your application’s failure modes.

Build a testing foundation with deterministic checks and probabilistic ones using LLM judges to prevent regressions.

Run Langfuse experiments in CI/CD pipelines to catch quality regressions before they reach production.

Scale your test coverage with LLM-generated synthetic data. Create diverse inputs and adversarial cases without waiting for users.

Run experiments to compare prompts, models, or pipelines. Measure metrics like accuracy and speed to quantify progress.

Specialized metrics for retrieval-augmented generation pipelines including retrieval relevance, answer faithfulness, and context completeness.

Evaluate coherence, memory, and resolution in conversational apps that maintain context across multiple turns.

Generate and test user-AI exchanges safely before production. Build realistic dialogue scenarios for robust evaluation.

Assess end-to-end trajectories for agents with tool use and planning. Evaluate action selection and task completion.

Build a custom annotation UI in minutes using Langfuse as data source

Use the Langfuse skill to analyze trace feedback and iteratively improve your prompts.

Test whether your judge prompt agrees with how you would label cases, and iterate until it does.

Capture human judgments from your own review tooling and ingest them into Langfuse as structured scores, validated against a score config.