July 1, 2024

Dify x Langfuse: Built in observability & analytics

Popular LLM development framework Dify.AI now integrates with Langfuse in one click.

We are excited to announce that Dify.AI (opens in a new tab), the highly popular, open source llm app development platform is now deeply integrated with Langfuse. The integration comes out of the box with version 0.6.12 and can be activated from within Dify in just one click.

Introducing the Langfuse - Dify Integration

Why Dify?

We have been impressed with the Dify project for a while now. Dify is an open source project that allows its users to build a wide array of powerful LLM application by providing tools and templates that enable the deployment of AI applications with no to little code. Three things we like about Dify:

  • Open Source: Dify is open source and like Langfuse it can easily be self-hosted. It has garned >35k stars on GitHub (opens in a new tab) and is pushed forward by its committed core team and community.

  • Ease of Use: Visual workflows, templated applications and the Dify Orchestration studio make deploying an LLM app a breeze. Dify works for both technical and non-technical, hobbyists and enterprise users.

  • Breadth of use cases: With Dify, users can build a large variety of applications such as agents, knowledge bases, RAG apps and workflows. You can use huge variety of models and tools to dream up new types of applications.

Bonus: Dify let's you embed your apps or use their APIs to easily integrate your apps with your existing systems.

Using Langfuse with Dify

Using Langfuse with Dify allows users to become even more methodical and data driven about tracking and improving their applications.

The core features of Langfuse we think will be appealing to Dify users:

  • Core Tracing: Inspect and analyze every step of your Dify app in Detail
  • Cost Tracking: Understand the cost and token usage incurred by your app and slice it by users, features and more
  • Dashboards: Don't we all love stats and graphs? The Langfuse dashboards allow you to get a brids-eye view of your app over time.
  • Evaluations: Dify users can take full advantage of Langfuse's powerful model-based evaluation feature to run fully configurable evals on any new incoming traces.
  • Datasets: More advanced users can start adding traces from Dify to their datasets and prepare for running tests against their 'golden data'.

How to get started?

Integrating Langfuse with Dify

Setting up tracing in Langfuse for Dify users is seriously simple. When editing your Dify app in the Orchestration Studio, click the 'Monitoring' tab in the menu and configure Langfuse by entering your Public and Secret Keys you can find in your Langfuse Project Settings.

Now enable monitoring and you will start to find all incoming traces in your Langfuse Tables and Dashboards. If you're having trouble with this, have a look at our Troubleshooting FAQ.

Any questions?

For thoughts, suggestions and questions regarding the Langfuse - Dify integration, please head to our GitHub Discussions (opens in a new tab).

Was this page useful?

Questions? We're here to help

Subscribe to updates