NodeTool - 🚀 Build AI Workflows Locally

An open-source, privacy-first, no-code platform for rapidly building and automating local AI workflows.

Nodetool Interface

🔒 Privacy-First & Fully Local

Local
  • Run AI models locally with zero data transmission
  • Use in restricted environments
  • Process sensitive data locally

🎨 Fast AI Prototyping

Prototyping
  • Fastest way to prototype AI workflows
  • Visual workflow editor, no coding required`
  • Drag-and-drop LLMs, Diffusion Models, & more

💻 Comprehensive System Integration

Integration
  • Control apps, clipboard & browser with AI
  • Custom shortcuts to trigger AI workflows
  • Access your knowledge base and local files

🌩️ Optional Cloud Providers

Cloud
  • Scale with GPU-powered cloud providers
  • Connect to OpenAI, Anthropic, Replicate & Fal.AI
  • Hybrid workflows, only share what you need

Note: Requires Nvidia GPU or Apple Silicon M1+ and at least 20GB of free space for model downloads.

Features

A Complete AI Toolbox

Visual Workflow Editor

Design complex AI workflows visually with drag-and-drop simplicity—no coding required. Easily create, understand, and modify AI workflows with our intuitive interface.

Visual Workflow Editor
Local AI Models

Run powerful open-source models from Hugging Face and Ollama directly on your hardware, ensuring complete privacy and offline capability.

Local AI Models
Global Chat Overlay

Access and trigger AI workflows from anywhere on your desktop through a unified, always-ready chat interface.

Global Chat Overlay
Asset Management

Effortlessly import, organize, and manage your images, audio, video, and other media assets within your AI workflows.

Asset Management
ComfyUI Integration

Directly import and execute your existing ComfyUI workflows, enhancing your creative AI projects.

ComfyUI Integration
Vector Storage & RAG

Built-in ChromaDB integration for storing and querying embeddings. Create powerful Retrieval-Augmented Generation workflows with any supported LLM.

Vector Storage & RAG

Examples

Discover what you can build

Explore our collection of pre-built workflows to jumpstart your AI journey

Categorize Mails

Categorize Mails

This workflow automates email categorization using AI to organize your inbox efficiently. It connects to your Gmail account, retrieves recent emails, and uses a Llama 3.2 model to intelligently classify each message into predefined categories (Personal, Work, Promotions, Newsletters, Finance, or Support) based on content analysis. The workflow then automatically applies the appropriate label to each email in Gmail, creating a well-organized inbox without manual sorting. Perfect for managing email overload and maintaining inbox zero with minimal effort. The AI analyzes subject lines, sender information, and email content to make accurate categorization decisions, even with a lightweight 3B parameter model.

emailllm
Daily Digest using Deepseek R1

Daily Digest using Deepseek R1

This workflow automatically searches your Gmail inbox for emails with '[AINews]' in the subject line from the past week, then uses the Deepseek R1 7B model to generate a concise, well-organized summary. The workflow extracts key information from each email including sender details, main topics, action items, and deadlines, then combines them into a comprehensive digest with categorized bullet points. Perfect for staying on top of important AI news without reading through multiple emails.

emailllm
Personal ChatGPT

Personal ChatGPT

A simple chat interface powered by OpenAI's models. This workflow allows you to have conversations with an AI assistant using your OpenAI API key.

chatopenai
Summarize Audio

Summarize Audio

Transcribe an audio file and summarize the text. This workflow takes an audio recording as input, converts speech to text using Whisper, and then generates a concise summary using an LLM. Perfect for quickly extracting key information from lectures, meetings, or voice notes without having to listen to the entire recording. Simply upload your audio file, run the workflow, and get a clear, organized summary of the content.

audiostart
Pokemon Maker

Pokemon Maker

Create unique and imaginative Pokemon with detailed descriptions and visually stunning images using AI. Input animal inspirations and get custom Pokemon with types, abilities, and personalities.

stable diffusionstart
Movie Posters

Movie Posters

Create cinematic movie posters using AI image generation. This workflow uses Ollama to generate movie concepts and character descriptions based on your chosen genre, then creates realistic movie poster images using Stable Diffusion. Simply enter a film genre (like 'sci-fi', 'western', or 'romance') and the system will generate three unique movie concepts with corresponding high-quality poster images. Perfect for creative inspiration, visual storytelling exercises, or exploring AI-generated cinematic art.

startimage
Style Transfer

Style Transfer

Transform your images by applying artistic styles from reference images. This workflow uses IP-Adapter to transfer visual styles while ControlNet preserves the original structure. Perfect for creating artistic variations of portraits or other images.

huggingfaceimagestart
Simple Chat

Simple Chat

A straightforward chat interface powered by the Deepseek R1 7B model. This example demonstrates how to create a basic conversational AI application with message history tracking. Users can interact with the AI through a simple chat interface, making it ideal for beginners to understand the fundamentals of building conversational workflows. The workflow processes user inputs, maintains conversation context, and generates natural language responses using a lightweight but capable language model.

chatstart
Image Enhance

Image Enhance

Improve image quality with basic enhancement tools like sharpening, contrast and color adjustment

imagestart
Simple RAG

Simple RAG

A retrieval-augmented generation (RAG) workflow that searches a document collection based on user queries. The system retrieves relevant information from the 'papers' collection, formats the results into a prompt, and uses the Deepseek-R1 7B model to generate concise, informative answers based on the retrieved content.

chat
Transcribe Audio

Transcribe Audio

Convert speech to text using Whisper model with word-level timestamps. This example demonstrates how to transcribe spoken audio into written text using OpenAI's Whisper small model. Simply record your voice or upload an audio file, and the workflow will automatically detect the language and generate an accurate text transcription. The model includes word-level timestamps for precise alignment between audio and text. Perfect for creating subtitles, transcribing interviews, meetings, or any spoken content.

startaudio
Object Detection

Object Detection

Detect objects in an image and visualize the detections

huggingfacestart
Segment Audio

Segment Audio

This example demonstrates how to automatically segment an audio file based on detected onsets (sudden changes in audio intensity). The workflow first loads an audio file, then uses librosa's onset detection to identify points where new sounds begin. These onset points are used to slice the original audio into separate segments, which are then saved as individual files. This technique is useful for breaking down longer audio recordings into meaningful chunks for further processing, analysis, or organization. Common applications include splitting music into individual notes, separating speech into words or phrases, or isolating sound effects from a continuous recording.

audio
Sklearn Classifier

Sklearn Classifier

A machine learning workflow that demonstrates a complete classification pipeline using scikit-learn. This example loads the Iris dataset, splits it into training and test sets, applies standard scaling for feature normalization, trains a Random Forest classifier, and evaluates the model performance with accuracy metrics and decision boundary visualization.

sklearn
Stable Diffusion XL in Comfy

Stable Diffusion XL in Comfy

Generate high-quality images using Stable Diffusion XL with advanced settings and refinement. This workflow demonstrates how to use the Realistic Vision V5.1 model to create detailed, high-resolution images at 1024x1024. You can customize the positive prompt to describe what you want to see (currently set to 'a butterfly in rainbow colors') and the negative prompt to specify what to avoid (currently set to 'deformed'). The workflow uses the DDIM sampler with 20 steps and a CFG scale of 8 for balanced results. Experiment with different models, prompts, and sampling settings to achieve your desired artistic style.

comfyimage

Join our Discord Community

Connect with developers, get help, and share your AI workflows

Nodetool Discord

Contact

Get in Touch

Got ideas, suggestions, or just want to say hi?
We'd love to hear from you!

[email protected]

Need help or want to chat with other developers?

Nodetool Forum
GitHub

Contribute to our open-source project

GitHub

Matthias Georgi: [email protected]

David Bührer: [email protected]