Daggr: AI Workflow Visualization Tool Released

Daggr is a new open-source Python library that helps you build complex AI pipelines by connecting Gradio apps, ML models, and custom functions.

Unlike existing GUI-based node editors, it adopts a code-first approach, where you define workflows in Python code and a visual canvas is automatically generated. This provides both version-control-friendly code and intuitive visual inspection at the same time.

Key features:

• Step-by-step inspection and re-execution: You can check the intermediate output of a specific step or modify its input and re-run just that step, without needing to re-run the entire pipeline, making debugging highly efficient.
• Seamless integration with Gradio: Since it's built by the Gradio team, public/private Spaces deployed on Hugging Face Spaces can be used directly as nodes without any additional adapters.
• State Persistence: The workflow's state, input values, cached results, canvas position, and more are automatically saved, allowing you to pick up your work right where you left off.

Users can leverage GradioNode, FunctionNode, CustomNode, and more to quickly build and visualize complex AI agents or data processing pipelines.