# Evaluating and improving agents with Langfuse

Canonical: https://brew.new/templates/langfuse/evaluating-and-improving-agents-with-langfuse

Brand: langfuse.com
Category: onboarding

![Preview of Evaluating and improving agents with Langfuse](https://cdn.brew.new/email-preview-ce3e407069f5252a-tracking_r57p5bpyyc9xh406b47pppkmtx8h0jrw-1793031748724.png)

## Email content

Hey Alex Reed,

Using evaluators to find interesting traces shows you where to look. Building out a suite of evals to reliably test your agent system allows you to continuously improve and check for regressions before you ship, and through that, close the loop.

Create your first dataset

A dataset is a set of test cases you want to repeatedly evaluate your agent against. The first dataset can be small. 15-30 examples that represent what your application handles with clear expectations on the output.

Run your first experiment

Experiments allow you to test variants of your system. A different prompt, a cheaper model, another harness. Each run gets evaluated and you can compare the different runs of your system.

Run experiments in your CI/CD

Some happy paths of your application need to always work. Check those as part of your CI/CD by including an experiment run as a deployment gate.

If anything is not covered here, just hit reply, we read every message. Or book a call and we'll look at your setup together.

Best,

Marc

Co-founder, Langfuse (part of Clickhouse)

Links: Roadmap, Docs, Talk to us, Example Project, Langfuse Academy

You are receiving this email because you opted in to receive email from the list Onboarding Journey from Langfuse

Langfuse, Langfuse GmbH, Oranienburger Strasse 91, 10178, Berlin, Germany

Unsubscribe

[Open and remix this design](https://brew.new/templates/langfuse/evaluating-and-improving-agents-with-langfuse)

[Browse email designs](https://brew.new/browse/templates)