Tracing
Langfuse tracing for LLM calls, tool use, retrieval, and multi-step agent workflows.
We implement Langfuse tracing, evaluations, and production monitoring so teams can understand how their LLM applications behave and catch regressions before release.
“The whole pipeline to improve our chat agent happens entirely in Langfuse. We screen conversations, review flagged sessions, turn clusters of unwanted behavior into datasets, and run those evals as regression tests before every release.”
Langfuse tracing for LLM calls, tool use, retrieval, and multi-step agent workflows.
Eval harnesses and scorecards that catch regressions before users do.
Observability setups that connect traces to human review and release gates.
We implement Langfuse for teams putting AI into production and need visibility from day one.
We partner on observability and eval design with ClickHouse and Langfuse field motions.
We operate the trace and eval loop as models, prompts, and agents evolve.
Langfuse is the open-source platform teams use to trace, evaluate, and improve LLM apps and agents. It joined ClickHouse in January 2026. These figures are Langfuse's own.
In production at Canva · Khan Academy · Ramp · Twilio · Intuit · Samsara · Hugging Face · SumUp · Merck Group
Sources: About Langfuse ↗ · langfuse.com ↗ · Langfuse customers ↗
“We've seen a 10-20% decrease in token usage, 15% fewer tool calls, and 30% faster end-to-end session times.”David TrainaData Platform, RampRead the story on langfuse.com ↗ (opens in a new tab)Slite · Software12xhigher experimentation velocityfrom a test every few months to one a week
“Now, once per week we can try something. We changed a few variables, we run the test suite, and we already know if it's worth it to pursue or not.”Charley DavidPrincipal Engineer, SliteRead the story on langfuse.com ↗ (opens in a new tab)SumUp · Payments50%of support conversations kept from humans300,000 merchant requests a month
“Building on Langfuse we have achieved a 50% deflection rate of conversations that don't go to humans.”Ana CasadoHead of Operations Data and AI, SumUpRead the story on langfuse.com ↗ (opens in a new tab)Evolve · Sleep health50%reduction in memory misses
“Everything that has to do with evals and error analysis, we completely rely on the fact that we have Langfuse as a repository for traces and prompts.”Ignacio PerezCTO, EvolveRead the story on langfuse.com ↗ (opens in a new tab)
Sound like you? Book a call ↗
Field notes from our own builds, and guides for choosing your stack.
Talks from teams in production.
We use Langfuse to trace, evaluate, and improve LLM apps in production.
More partnerships
Press Esc to close