NodeTool: Visual AI Workflow Builder

Build AI Workflows Visually. Locally.

NodeTool is an open-source visual environment for building, testing, and deploying AI workflows. Connect powerful models running locally or in the cloud. No API keys needed for local inference. Your data stays with you.

Core Principles

• Local First: Run entirely on your machine. No cloud dependency.
• Open Source: AGPL-3.0 licensed. Inspect, modify, and self-host.
• Private: No telemetry. No tracking. Your data is yours.
• Portable: Workflows run anywhere, from laptop to cloud.

Features

• Visual Workflow Engine: Drag-and-drop interface with real-time execution, type safety, and multimodal support (text, image, audio, video).
• Universal Inference: Run SOTA models (LLMs, Flux, SDXL) natively on Apple Silicon (MLX), NVIDIA GPUs, or CPU (llama.cpp).
• HuggingFace Integration: Download and run thousands of models directly from the Hub.
• Flexible Providers: Mix local models with APIs from OpenAI, Anthropic, Gemini, Replicate, and more.
• One-Click Deployment: Scale from local dev to serverless GPUs with nodetool deploy.
• Asset Management: Built-in manager for all your media files.

Use Cases

• Smart Assistants: Chat with local documents.
• Agentic Workflows: Search, classify, and automate.
• Content Creation: Generate text, images, video, and music.
• Data Analysis: Process data and extract insights.

Quick Start

Platform	Download	Requirements
Windows	Download Installer	Nvidia GPU recommended, 20GB free space
macOS	Download Installer	M1+ Apple Silicon
Linux	Download AppImage	Nvidia GPU recommended

Hardware for Local Inference:

• Apple Silicon: 16GB+ RAM (LLM/TTS), 24GB+ (Image Gen)
• Windows/Linux: 4GB+ VRAM (LLM/TTS), 8GB+ (Image), 12GB+ (Video)
• Cloud Only: No GPU required.

---

Documentation

• Getting Started
• Node Packs
• Custom Nodes
• Deployment
• API Reference

---

🛠️ Development Setup

For core library development, see nodetool-core.

Prerequisites

• Python 3.11, Conda, Node.js LTS

Quick Start

# 1. Setup Environment
conda env update -f environment.yml --prune
conda activate nodetool

# 2. Install Core
uv pip install git+https://github.com/nodetool-ai/nodetool-core git+https://github.com/nodetool-ai/nodetool-base

# 3. Run Backend & Frontend
nodetool serve --reload &
cd web && npm install && npm start

Desktop App (Electron)

Configure settings.yaml with your Conda path, then:

# Build
cd web && npm install && npm run build
cd ../apps && npm install && npm run build
cd ../electron && npm install && npm run build

# Start
cd ../electron && npm start

Testing & Quality

# Python
pytest -q

# Web
cd web && npm test && npm run lint

# Pre-commit
pip install pre-commit && pre-commit install && pre-commit run --all-files

Contributing

We welcome contributions!

• Bug reports & Features: Help us improve.
• Code: Fix bugs or add features.
• Nodes: Create new nodes.

Pull requests are welcome! Open an issue for major changes.

License

AGPL-3.0 license.

Get in Touch

• General: hello@nodetool.ai
• Team: matti@nodetool.ai, david@nodetool.ai

GitHub • Discord