<aside> 💡

This document explains how Nebuly connects to LangFuse to ingest chat data into the Nebuly Analytics platform.

</aside>

There are two main ways to connect LangFuse with the Nebuly platform:

  1. Full integration – Nebuly directly connects to your LangFuse account and automatically ingests the data.
  2. Local integration – Nebuly provides open-source scripts that read data from the LangFuse APIs and send it to Nebuly’s endpoint. This approach allows customers to fully manage the ingestion process on their end, without storing LangFuse credentials in their Nebuly account.

Full integration

Nebuly requires two pieces of information to retrieve your data from LangFuse:

  1. The LangFuse public API key
  2. The LangFuse private API key

Both keys are securely stored in our encrypted secret storage, hosted on Azure Key Vault (or an equivalent solution for self-hosted environments). This ensures that the secrets are accessible only to the components that need them.

By default, data ingestion occurs once per day (this interval can be configured). The Nebuly platform uses the provided public and private API keys to access the LangFuse data, which it then automatically converts into the internal format used within Nebuly.

<aside> ⚠️

All information about the end-user, like geography, location, role, etc. that can be interesting to look at when analysing the data MUST be stored on LangFuse as tags for the trace object!

</aside>

Local integration

The local integration allows customers to fully manage their LangFuse integration without sharing any additional information with Nebuly servers.

We provide open-source scripts that, once configured with LangFuse API keys, extract data from LangFuse, convert it into the Nebuly-supported format, and send it to the Nebuly endpoint. These scripts are available on our GitHub page, where you can download and fully customize them.

This approach gives you finer control over the data sent to the Nebuly platform and is particularly useful when you want to enrich the chatbot interactions stored in LangFuse with additional information—such as user or customer-related data—that developers typically cannot access.

https://github.com/nebuly-ai/nebuly-integration-examples/blob/main/langfuse/langfuse_api.py