Langfuse

Connect GridWork HQ to Langfuse and trace all pipeline LLM runs. View token usage, latency, and model outputs in your own Langfuse project.

Prerequisites

• Langfuse account (cloud or self-hosted) with a project created
• Langfuse Public Key, Secret Key, and Host URL (if self-hosted)

Connect to Langfuse

1. In Langfuse, go to your project settings.
2. Copy your Public Key and Secret Key.
3. Note your Host URL (cloud: https://cloud.langfuse.com, self-hosted: your instance URL).
4. In GridWork HQ, go to Settings > Integrations > Langfuse.
5. Paste your Public Key, Secret Key, and Host URL.
6. Click Connect.

What It Enables

Once connected, all pipeline LLM runs are automatically traced in Langfuse. You can:

• View token usage and cost per pipeline run.
• Monitor latency and identify bottlenecks.
• Inspect model inputs and outputs.
• Track performance over time with dashboards and alerts.
• Debug LLM behavior with detailed traces.

Disconnect from Langfuse

1. Go to Settings > Integrations > Langfuse.
2. Click Remove or Disconnect.
3. New pipeline runs will no longer be traced to Langfuse.