Engineering Resources
In-depth, engineering-focused articles on LLM observability, evaluation, and tooling.
Migrations
- Migrate a Promptfoo eval suite to Langfuse
Step-by-step guide to migrate a Promptfoo eval suite to Langfuse: turn tests into dataset items, port assertions to evaluators, move prompts, and keep failing CI on regressions.
Last updated Sep 14, 2026
- Migrate from Arize AX to Langfuse
Migrate from Arize AX to Langfuse: keep OpenInference, switch the exporter, export datasets, and re-run experiments without importing old scores.
Last updated Sep 12, 2026
- Migrate from Arize Phoenix to Langfuse
Step-by-step guide to migrating from Arize Phoenix to Langfuse: keep OpenInference instrumentation, replace phoenix.otel.register(), and recreate datasets, prompts, and experiments.
Last updated Sep 12, 2026
- Migrate from Braintrust to Langfuse
Step-by-step guide to migrate from Braintrust to Langfuse: swap SDK instrumentation, export datasets via API, keep autoevals scorers, and re-run experiments.
Last updated Sep 14, 2026
- Migrate from Helicone to Langfuse
Step-by-step guide to migrate from Helicone to Langfuse: switch tracing off the gateway, map users, sessions, and properties, migrate prompts with versions, and export request history.
Last updated Sep 14, 2026
- Migrate from LangSmith to Langfuse
Migrate from LangSmith to Langfuse: fan out traces via OpenTelemetry, copy datasets and prompts, re-run experiments, and optionally reingest trace history.
Last updated Sep 12, 2026
Articles
- 🪢 Knot emoji: meaning, Unicode, and why it's the Langfuse logo
What the 🪢 knot emoji means, its Unicode codepoint, how to copy and use it, and why an LLM engineering platform picked it as a logo.
Last updated Sep 7, 2026
- 10 code evaluator examples for AI application evaluation
Ten copy-paste code evaluators for LLM applications: output validation, PII screening, RAG citation checks, numeric tolerance, refusal detection, and more.
Last updated Jul 6, 2026
- AI agent evaluation: trajectory, tool calls, and task completion
AI agent evaluation explained: how to measure trajectory, tool use, task completion, and multi-turn quality, with offline and online evaluation patterns.
Last updated Sep 14, 2026
- AI safety research with Langfuse: tracing and evaluation
Use Langfuse for AI safety research: trace agent behavior, review failures, evaluate safety monitors, and compare interventions with open-source infrastructure.
Last updated Sep 9, 2026
- Answer relevance evaluation for LLM applications
Answer relevance evaluation for LLM apps: what the metric measures, judge prompt design, a runnable Langfuse experiment, and how Ragas answer relevancy fits.
Last updated Jul 16, 2026
- Chatbot analytics: analyzing what users ask your AI chatbot
Four working patterns for chatbot analytics on LLM traces: offline intent classification, LLM-as-a-judge detection, score dashboards, and agent-driven analysis.
Last updated Aug 12, 2026
- Clarifications
Langfuse is an open-source AI engineering platform for tracing, prompt management, evals, and experiments. September 2026 clarifications of what Langfuse includes.
Last updated Sep 12, 2026
- Evaluate an existing application
Reuse Python or TypeScript application functions and deterministic graders, compare versions on a fixed dataset, and catch regressions in CI.
Last updated Sep 11, 2026
- Evaluation of RAG pipelines with Ragas
Use RAGAS to evaluate your RAG pipelines traced with Langfuse to measure the quality of your retrieval and sythesis.
Last updated Aug 12, 2026
- Golden dataset evaluation: build and maintain LLM test sets
Golden dataset evaluation for LLM apps: build test sets from production traces, keep them fresh, and compare prompt versions over time with experiments.
Last updated Aug 12, 2026
- Hallucination detection for LLM apps
Hallucination detection for LLM apps: taxonomy, LLM-as-a-judge evaluators on production traces, offline experiments with references, and code pre-screens.
Last updated Jul 31, 2026
- How Langfuse runs ClickHouse at agent scale
Why Langfuse stores agent traces in ClickHouse and how we operate it at scale: the data model, the write path, read isolation, and the trade-offs we engineer around.
Last updated Jul 24, 2026
- How to build an LLM evaluation strategy
How to build an LLM evaluation strategy: quality dimensions from failure modes, evaluator selection, CI release gates, production monitoring, and human review.
Last updated Sep 8, 2026
- How to evaluate RAG faithfulness with LLM-as-a-judge
RAG faithfulness evaluation explained: what the faithfulness metric measures, how to design an LLM-as-a-judge for it, and runnable Langfuse Python SDK code.
Last updated Jul 16, 2026
- How to evaluate sessions and conversations
How to evaluate entire sessions in Langfuse rather than individual traces: session scores, LLM-as-a-judge on conversations, annotation, and SDK patterns.
Last updated Aug 17, 2026
- How to organize, version, and test hundreds of prompts
We manage hundreds of prompts across several LLM features. How should we organize, version, and test them? Folders, labels, prompt experiments, and UI edits.
Last updated Sep 4, 2026
- Langfuse for EU teams: data residency, GDPR, and compliance
Where Langfuse Cloud's EU region stores data and backups, what the DPA and DSAR process cover, and which controls and certifications apply to EU teams.
Last updated Sep 4, 2026
- Langfuse for product teams: review conversations without code
How product teams use Langfuse without writing code: session replay, annotation queues, user feedback, dashboards, and natural language questions over traces.
Last updated Sep 10, 2026
- Langfuse SDK Performance Test
Benchmark of the latency impact of the Langfuse Python SDK and its OpenAI, Langchain, and LlamaIndex integrations.
Last updated Aug 10, 2026
- LLM cost management: how to track, attribute, and control spend
LLM cost management: track token spend accurately, attribute it to users and features, and control it with alerts. Pitfalls and how to solve them in Langfuse.
Last updated Sep 7, 2026
- LLM regression testing: fail CI before regressions ship
Set up LLM regression testing that fails CI: golden datasets, experiment thresholds, RegressionError, and GitHub Actions via langfuse/experiment-action.
Last updated Sep 11, 2026
- PII masking patterns for LLM applications
Where to mask PII in an LLM application: client-side SDK masking, server-side ingestion masking, gateway-level filtering, and detection with evaluators. Patterns and trade-offs.
Last updated Jul 6, 2026
- Prompt CI/CD: version, gate, and roll out prompts like code
Learn how to build a prompt CI/CD pipeline: version prompts with labels, validate on datasets, gate promotion in CI, roll out gradually, and roll back fast.
Last updated Aug 21, 2026
- Prompt Management Performance Benchmark
Performance benchmark on Langfuse Prompt Management measuring latency of retrieving and compiling prompts.
Last updated Aug 10, 2026
- Tracing coding agents: Claude Code, Codex, Copilot & more
How to trace AI coding agents with Langfuse: Claude Code, OpenAI Codex, GitHub Copilot, Cursor, and five more. Setup patterns, cost tracking, and team governance.
Last updated Sep 11, 2026
- Turn user feedback on AI responses into evaluation datasets
How to collect user feedback on AI responses and turn it into evaluation datasets: capture ratings as scores, filter the failures, build dataset items, gate CI.
Last updated Sep 14, 2026
- Use Langfuse from Go, Java, C#, and Ruby via OpenTelemetry
Langfuse works from any language with an OpenTelemetry SDK: point the OTLP exporter at the Langfuse endpoint. Setup for Go, Java, C#/.NET, and Ruby.
Last updated Sep 14, 2026
- Using DeepEval with Langfuse: LLM evaluation guide
Use DeepEval with Langfuse to evaluate LLM applications: run DeepEval metrics in your Python pipeline and store results as scores on traces and experiments.
Last updated Aug 12, 2026
- Webinar: Traceability and Observability in Multi-Step LLM Systems
In this session, we talked about advanced techniques and best practices for implementing traceability and observability in multi-step LLM systems.
Last updated Sep 7, 2026
- What is an LLM gateway? When you need one (and when you don't)
An LLM gateway is a proxy layer between your application and model providers: one API, failover, caching, cost controls. How gateways work and how to pick one.
Last updated Sep 7, 2026