Try Langfuse yourself with a view-only project. Alternatively:
Get access to demo project by creating a free account (no credit card required).
Create demo account
Interact with chatbot below to see new traces and user feedback in Langfuse.
👋 Do you have any questions about Langfuse? Ask me!
⚠️ Warning: Do not enter sensitive information. All chat messages can be viewed in the live demo. Humans (the founders) are available via the chat widget.
Interested in implementation details? Checkout the blog post (code is fully open source).
- Monitor usage (cost control)
- Track latency, quality (based on user feedback) and OpenAI errors
Each response is based on the following steps which can go wrong, be slow or expensive:
- Embedding of user request
- Embedding similarity search in Postgres
- Summary of docs as markdown
- Generation of response based on retrieved context and chat history
This is how a single trace looks like in Langfuse:
In this example, we can see how we do:
- Collection of feedback using the Langfuse Web SDK
Negative, Langchain not included in response
- Browsing of feedback
- Identification of the root cause of the low-quality response
Docs on Langchain integration are not included in embedding similarity search
We are currently running an alpha with more prebuilt analytics capabilities based on the ingested production data. All users of Langfuse Cloud get early access.
Why is this an alpha program? We rely on some closed source software to make this work in order to move fast and focus on what's most interesting for most of you. We'll soon bring these features to all users since we currently plan a migration to an OLAP database and flexible modeling structure to power these features.