Skip to main content
Back

Google's Visual Blocks Lets Anyone Build AI Pipelines Without Writing Code

Google's open-source Visual Blocks lets you wire up ML pipelines by dragging nodes — no code required, and now with LLM-powered pipeline generation via InstructPipe.

  • Visual Blocks is Google's free, open-source no-code framework for building ML pipelines by connecting drag-and-drop nodes in a browser editor.
  • The library now has 70+ nodes covering models, I/O (camera, mic, image), processors, and visualizations, plus support for custom nodes tied to any API.
  • InstructPipe (CHI 2025 Honorable Mention) lets you describe a pipeline in plain text and have an LLM generate the node graph automatically.
  • Hugging Face integration (launched at Google I/O 2024) adds nodes for text-to-image, chat completion, image classification, and more, running in-browser via Transformers.js.
  • A Python package (pip install visualblocks) embeds the visual editor directly inside Google Colab notebooks, wrapping your own inference functions as nodes.
  • The project is Apache 2.0 licensed and actively maintained; try it at visualblocks.withgoogle.com or on GitHub.

Visual Blocks for ML is Google's open-source framework for building machine learning pipelines without writing code. Instead of scripting data flows and model calls by hand, you drag nodes onto a canvas, connect them, and watch results update in real time. It's been quietly maturing since its debut at CHI 2023, and its recent Hugging Face integration and InstructPipe AI assistant make it worth a serious look right now.

Visual Blocks provides a no-code graph building experience through its node-graph editor, where users can create and connect different components to rapidly build an ML pipeline and see the results in real time. The framework is mainly written in JavaScript and leverages TensorFlow.js and TensorFlow Lite for ML capabilities, and three.js for graphics rendering.

What's actually inside the editor

The interface gives users three coordinated views: a Nodes Library containing over 30 nodes (such as Image Processing, Body Segmentation, and Image Comparison) with a search bar for filtering; a Node-graph Editor for building and adjusting pipelines by dragging nodes; and a Preview Panel that visualizes the pipeline's input and output, lets you alter intermediate results, and visually compare different models.

Visual Blocks node graph editor showing connected ML pipeline nodes

Visual Blocks now provides more than 70 nodes you can arrange or connect together, and recently launched support for defining your own custom nodes that integrate with any service, model, or code. The custom node system is where things get genuinely interesting for teams that want to plug in their own APIs or proprietary models.

The node types span the full pipeline lifecycle:

  • Models , pre-built ML models for vision, NLP, and more
  • I/O , live camera, image upload, microphone, text input
  • Processors , image transforms, text processors, data augmentation