Visual Blocks

Visual Blocks is a Google open-source framework for building machine learning pipelines visually, with a no-code graph editor instead of scripts. Hosted on GitHub with about 1,400 stars under the Apache-2.0 license and still updated in 2026, it lets you prototype ML workflows by connecting drag-and-drop components for models, inputs, processors and visualizations.

Overview

Known from Google’s research as the system behind the Rapsai paper at CHI 2023, Visual Blocks for ML aims to accelerate many stages of the ML product cycle: pipeline authoring, model evaluation, data pipelining and experimentation. It ships as a JavaScript front-end library for low/no-code editing plus a separate library for embedding the editor experience into your own platform or application. Users drag nodes onto a canvas and wire them together — for example chaining a camera input into an ML model and then into a visualization — and can run and tweak the pipeline interactively, which makes demos, teaching and rapid prototyping far more approachable than code-first tooling.

Key Features

• No-code graph editor for building ML pipelines by drag and drop
• Reusable nodes for models, user inputs, processors and visualizations
• JavaScript front-end library that platforms can embed in their own apps
• Speeds up pipeline authoring, evaluation and experimentation cycles
• Interactive runs so users can adjust workflows and see results immediately
• From Google, based on the published Rapsai research system
• Apache-2.0 licensed and free to integrate

💳 Pricing

Visual Blocks is free and open source under the Apache-2.0 license. There is no paid tier or hosted plan from the project; you integrate the JavaScript libraries into your own product and run everything yourself.

Who Is It For

Platform builders, educators and product teams who want to give non-programmers a friendly visual way to assemble and experiment with ML pipelines, and researchers prototyping multimedia AI applications quickly.