For teams rolling out Langfuse

Langfuse training, built on your use case.

A two-hour private session, taught on a workspace we set up around what you are shipping. Up to 10 people, hands on.

Book Training Enquire about training

The session

Format
2 hours, live, remote
Attendees
Up to 10
Workspace
Built around your use case
Follow-up
1 hour Q&A, within two weeks
Recording
Yours to keep

01

Built on your use case

We set up the workspace around your use case, before the session.

02

Small and hands on

Up to 10 people, so everyone works in the workspace and every question gets answered.

03

Post training follow-up

We include an hour of expert time to review your project progress and answer questions.

04

Taught by operators

We run Langfuse in production across forty-plus agents. We teach what that taught us.

How it works

The workshop.

  1. 01

    The tour

    A real Langfuse deployment, running. How projects and environments split, what a trace tree looks like when an agent calls five tools, where cost and latency show up.

  2. 02

    Setup

    Project, keys, region, first trace out of a real service. SDK or OpenTelemetry. Plus the three mistakes that produce an empty project.

  3. 03

    Tracing that holds up

    The three layers, in the order that works. Sessions, users and tags, so a support ticket becomes a trace in one search. Cost that is actually right, including Bedrock and Vertex. Masking PII before it leaves your process.

  4. 04

    Prompts and evals

    Prompts out of the repo into versioned labels, so a change is a label move and a rollback is the same move backwards. Datasets built from real traces, including the failures. Run it, score every item, change one thing, compare side by side.

  5. 05

    Monitoring and controls

    LLM-as-judge evaluators in the UI. Annotation queues for what a model cannot settle. Alerts into Slack or a webhook. SSO, RBAC, retention, residency. What actually drives your bill: trace depth, not user count.

  6. 06

    Q&A on your stack

    Usually the most useful part. Bring the argument your team is currently having.

Book Training Enquire about training

After you book

Slot
You choose, after payment
Invite
Sent when you pick the slot
Workspace
Built before the session
Reschedule
72 hours' notice

Before you book

What people ask on the enquiry call.

  • Six 20-minute blocks. A tour of a running deployment and setup of your first trace. Tracing that holds up in production. Prompts and evals, so you can prove a change helped. Monitoring and controls, including what drives your bill. Then Q&A on your stack. The full breakdown is above.

  • The workspace, the recording, and a short runbook: instrumentation checklist, starter evaluators, retention plan.

  • It is free, good, and built on a sample app. Do it. This one is built on your use case, covers what the Enterprise tier adds, and is taught by people who run Langfuse in production.

  • Up to 10, from one company. That keeps it hands on: everyone works in the workspace and gets their questions answered. More than 10? Talk to us first.

  • Two short answers on the form: what you are building, and what to focus on. No access to your systems or data.

  • Yes. Tracing, prompts and evaluation are the same either way. We also cover which Enterprise features you need, including when the answer is none.

  • No. The workspace is ours, built with synthetic traffic, so there is nothing to clear with security. If you want us inside your systems, that is a deployment engagement: the scopes are here.

  • Free with 72 hours' notice. Inside that, the workspace is already built and the session is used up.

  • Take the enquiry call and we will invoice you.

  • Enter it on the form and it carries to checkout. A full-value code takes the total to zero.

  • The engineers who run it in production, and who deploy the ClickHouse and Langfuse stack for other teams.

Something not answered here? contact@bonnevilledata.com →