v0.1.0 — now available for Windows, macOS, and Linux

Build data pipelines
visually.

Drag, connect, and execute Python workflows without writing boilerplate. Full pandas power, zero friction.

Download Learn more
App screenshot — coming soon

Everything you need for data work

A visual ETL tool that generates real Python code under the hood.

Visual Node Editor

Drag tools from the sidebar, connect them with edges, and watch your pipeline take shape. No boilerplate, no guessing what connects to what.

One-Click Execution

Run your entire pipeline or just a specific node. See results instantly in the built-in data preview — tables, scalars, and errors, all inline.

Real Python Code

Every node generates actual pandas code. No proprietary format — inspect it, learn from it, or export it. What you see is what Python runs.

Extensible Tool Libraries

40+ built-in transforms — filter, join, group, pivot, and more. Import community libraries or create your own with a simple JSON format.

Bundled Python Runtime

No Python installation required. Confusion Flow ships with Python 3.13 and auto-installs missing packages at runtime via uv.

AI Tool Integration

Built-in MCP server lets Claude and other AI tools interact with your pipelines directly. Connect via Claude Desktop or Claude Code.

How it works

1

Drag tools onto the canvas

Choose from transforms like Read CSV, Filter, Join, Group By, and more from the sidebar. Each becomes a node on your canvas.

2

Connect and configure

Draw edges between nodes to define your data flow. Click any node to configure its parameters — column names, conditions, formulas.

3

Execute and explore

Hit Run and watch your pipeline execute node by node. Click any node to preview its output data in a built-in table view.

4

Save, share, iterate

Save your workflow as a JSON file. Re-open it later, share it with teammates, or load one of the included sample workflows to learn.

Download Confusion Flow

Available for all major platforms. Free and open to use.

Windows macOS Linux