Noschen

AI-Powered Research Note-Taking App

Noschen is a privacy-first, local-first research note-taking application that provides live AI feedback to support academic researchers, consultants, and students. The AI analyzes your notes in real-time and suggests improvements, identifies gaps, and enhances research quality.

Features

Core Features

Local-First Privacy: All notes are stored locally on your machine. No cloud sync, no data leaving your device.
AI-Powered Feedback: Real-time intelligent suggestions powered by local LLM (Ollama) or Mistral API.
Hierarchical Note Structure: Organize research with H1 headings (main topic) and H2 headings (sub-questions/aspects).
Dark Mode Interface: Clean, distraction-free dark theme designed for extended research sessions.
Auto-Save: Never lose your work—notes save automatically after 1 second of inactivity.
Full-Text Search: Instantly search across all your notes.

AI Analysis Types

Type	Description
MECE Validation	Checks if your structure is Mutually Exclusive and Collectively Exhaustive. Suggests missing categories or better groupings.
Gap Identification	Flags aspects, considerations, or perspectives not yet addressed in your research.
Source Suggestions	Recommends types of literature or domains to explore (e.g., "Consider literature on institutional economics").
Structural Improvements	Proposes reorganization for clearer argumentation or logical flow.

Additional Features

Accept/Reject workflow for AI suggestions
Review previously rejected suggestions
Contextual feedback aware of your document hierarchy
Section exclusion from AI feedback
Keyboard-driven workflow with Apple-appropriate shortcuts

Screenshots

┌─────────────────────────────────────────────────────────────────────────┐
│  Noschen                              │                    AI Connected │
├───────────────────┬─────────────────────────────────────────────────────┤
│                   │                                                     │
│  🔍 Search notes  │  # Impact of Remote Work on Productivity           │
│                   │  ─────────────────────────────────────────────────  │
│  ┌─────────────┐  │                                                     │
│  │ Research    │  │  ## Communication Patterns                         │
│  │ Paper Draft │  │                                                     │
│  │ Jan 19      │  │  Remote work has fundamentally changed how teams   │
│  └─────────────┘  │  communicate. Asynchronous communication has       │
│                   │  become the norm, replacing spontaneous office     │
│  ┌─────────────┐  │  conversations...                                  │
│  │ Literature  │  │                                                     │
│  │ Review      │  │  ┌──────────────────────────────────────────────┐  │
│  │ Jan 18      │  │  │ AI Suggestions (2)                           │  │
│  └─────────────┘  │  │                                              │  │
│                   │  │ [GAP] Consider addressing time zone          │  │
│  + New Note       │  │ challenges in distributed teams    [✓] [✗]  │  │
│                   │  │                                              │  │
│  ⚙ Settings       │  │ [MECE] Your analysis covers sync vs async   │  │
│                   │  │ but misses hybrid approaches      [✓] [✗]  │  │
│                   │  └──────────────────────────────────────────────┘  │
└───────────────────┴─────────────────────────────────────────────────────┘

Installation

Prerequisites

Node.js 18 or higher
npm 9 or higher
Ollama (recommended) or Mistral API key

Quick Start

# Clone the repository
git clone https://github.com/maxpolwin/Noschen.git
cd Noschen

# Switch to the development branch
git checkout claude/noschen-research-notes-app-SCU8k

# Install dependencies
npm install

# Start in development mode
npm run dev

Build for Production

# Build the application
npm run build

# Package as a desktop app (macOS, Windows, Linux)
npm run package

Installing Ollama (Recommended)

For the best privacy-first experience, install Ollama to run AI models locally:

Download Ollama from ollama.ai
Install and launch Ollama
Pull a recommended model:

# Recommended for M2 MacBook (good balance of speed and quality)
ollama pull llama3.2

# Alternative lightweight models
ollama pull phi3
ollama pull mistral

Configuration

AI Provider Setup

Launch Noschen and click the Settings icon in the sidebar to configure your AI provider.

Option 1: Ollama (Local LLM) — Recommended

Best for privacy and offline use. Runs entirely on your machine.

Setting	Default	Description
Ollama URL	`http://localhost:11434`	Ollama server address
Model	`llama3.2`	The model to use for analysis

Recommended models by hardware:

Hardware	Recommended Model	Notes
M2/M3 MacBook	`llama3.2`, `phi3`	Fast, good quality
M1 MacBook	`phi3`, `mistral`	Lighter weight
16GB+ RAM	`llama3.2:8b`	Higher quality
8GB RAM	`phi3`	Optimized for lower memory

Option 2: Mistral API (Cloud)

Best for users without powerful local hardware or who prefer cloud processing.

Create an account at console.mistral.ai
Generate an API key
Enter the key in Noschen settings

Setting	Description
API Key	Your Mistral API key (starts with `sk-`)

Usage Guide

Creating Your First Note

Click + New Note in the sidebar
Start with an H1 heading for your main research topic:
- Type # followed by your topic
- Example: # The Impact of AI on Creative Industries
Add H2 headings for sub-questions or aspects:
- Type ## followed by the sub-topic
- Example: ## Economic Disruption
Write your content under each heading
AI feedback will appear automatically after 2 seconds of inactivity

Document Structure Best Practices

# Main Research Question (H1)
Your overarching research topic or thesis question.

## Sub-Question 1 (H2)
First major aspect to investigate.

### Supporting Point (H3)
Detailed evidence or argument.

## Sub-Question 2 (H2)
Second major aspect to investigate.

## Sub-Question 3 (H2)
Third major aspect to investigate.

Why this structure matters:

The AI uses H1 to understand your overall research intent
H2 headings define the scope of analysis
Feedback relates primarily to the current H2 section while considering the broader context

Working with AI Suggestions

When AI feedback appears:

Review the suggestion in the feedback panel below your text
Accept (✓) to add a TODO reminder in your text
Reject (✗) to hide the suggestion
Reconsider previously rejected suggestions by clicking "Show rejected"

Excluding Sections from AI Feedback

If you want to exclude certain sections from AI analysis (e.g., personal notes, drafts):

Add [no-ai] tag after your H2 heading
Example: ## Personal Notes [no-ai]

AI Feedback System

How It Works

Trigger: Feedback generates after 2 seconds of typing inactivity
Context: AI considers:
- The current H2 section you're working in
- Relationship to other H2 sections
- The overarching H1 topic
Display: Colored inline badges appear in the feedback panel

Feedback Types Explained

MECE Validation (Purple)

Checks if your research structure is:

Mutually Exclusive: No overlapping categories
Collectively Exhaustive: No missing categories

Example feedback:

"Your analysis of market segments overlaps between 'Enterprise' and 'Large Business'. Consider consolidating or clarifying the distinction."

Gap Identification (Blue)

Identifies missing perspectives, considerations, or aspects.

Example feedback:

"Consider addressing the ethical implications of this technology, which is absent from your current analysis."

Source Suggestions (Green)

Recommends types of literature or domains to explore.

Example feedback:

"Consider literature on behavioral economics, particularly work by Kahneman and Thaler, for the decision-making section."

Structural Improvements (Orange)

Suggests reorganization for clarity and logical flow.

Example feedback:

"Consider moving the 'Historical Context' section before 'Current State' for better chronological flow."

Feedback Quality Tips

For best AI feedback:

Write clear, descriptive H1 and H2 headings
Include enough content (50+ characters) before expecting feedback
Be specific in your writing—vague text produces vague feedback
Use consistent terminology throughout your document

Keyboard Shortcuts

Global Shortcuts

Shortcut	Action
`Cmd + S`	Force save current note
`Cmd + N`	Create new note
`Cmd + F`	Focus search

Feedback Shortcuts

Shortcut	Action
`Cmd + Enter`	Accept first active suggestion
`Cmd + Delete`	Reject first active suggestion

Editor Shortcuts

Shortcut	Action
`Cmd + B`	Bold text
`Cmd + I`	Italic text
`Cmd + Shift + 1`	Heading 1
`Cmd + Shift + 2`	Heading 2
`Cmd + Shift + 3`	Heading 3
`Cmd + Z`	Undo
`Cmd + Shift + Z`	Redo

Architecture

Project Structure

noschen/
├── src/
│   ├── main/                    # Electron main process
│   │   ├── main.ts              # App entry, window management, IPC handlers
│   │   └── preload.ts           # Context bridge for renderer
│   │
│   ├── renderer/                # React frontend
│   │   ├── components/
│   │   │   ├── Editor.tsx       # TipTap editor with feedback integration
│   │   │   ├── Sidebar.tsx      # Note list and search
│   │   │   ├── FeedbackPanel.tsx # AI suggestions display
│   │   │   ├── SettingsModal.tsx # AI configuration
│   │   │   └── EmptyState.tsx   # Welcome screen
│   │   │
│   │   ├── styles/
│   │   │   └── global.css       # Dark theme and all styles
│   │   │
│   │   ├── App.tsx              # Main app component
│   │   ├── main.tsx             # React entry point
│   │   └── index.html           # HTML template
│   │
│   └── shared/
│       └── types.ts             # Shared TypeScript interfaces
│
├── package.json
├── tsconfig.json
├── vite.config.ts               # Vite bundler configuration
└── README.md

Technology Stack

Layer	Technology
Framework	Electron 28
Frontend	React 18 + TypeScript
Editor	TipTap (ProseMirror)
Bundler	Vite 5
Styling	CSS Custom Properties
Icons	Lucide React
Local AI	Ollama API
Cloud AI	Mistral API

Data Flow

┌─────────────┐     ┌─────────────┐     ┌─────────────┐
│   Editor    │────▶│  Debounce   │────▶│  AI Engine  │
│  (TipTap)   │     │  (2 sec)    │     │  (Ollama/   │
└─────────────┘     └─────────────┘     │   Mistral)  │
                                        └──────┬──────┘
                                               │
┌─────────────┐     ┌─────────────┐            │
│  Feedback   │◀────│   Parse &   │◀───────────┘
│   Panel     │     │   Display   │
└─────────────┘     └─────────────┘

Data Storage

Storage Location

Notes are stored as JSON files in your system's application data folder:

OS	Path
macOS	`~/Library/Application Support/noschen/notes/`
Windows	`%APPDATA%/noschen/notes/`
Linux	`~/.config/noschen/notes/`

Note Format

Each note is stored as a separate JSON file:

{
  "id": "note_1705678900000",
  "title": "Research on AI Ethics",
  "content": "<h1>Research on AI Ethics</h1><p>Content here...</p>",
  "createdAt": "2024-01-19T12:00:00.000Z",
  "updatedAt": "2024-01-19T14:30:00.000Z",
  "excludedSections": ["Personal Notes"]
}

Settings Storage

AI settings are stored in:

OS	Path
macOS	`~/Library/Application Support/noschen/settings.json`
Windows	`%APPDATA%/noschen/settings.json`
Linux	`~/.config/noschen/settings.json`

Backup

To backup your notes, simply copy the entire noschen folder from your application data directory.

Troubleshooting

Common Issues

"AI Disconnected" Status

Cause: Cannot connect to Ollama or Mistral API.

Solutions:

For Ollama:
- Ensure Ollama is running: ollama serve
- Check the URL in settings (default: http://localhost:11434)
- Verify a model is installed: ollama list
For Mistral:
- Verify your API key is correct
- Check your internet connection
- Ensure you have API credits

No Feedback Appearing

Possible causes:

Content is too short (minimum ~50 characters needed)
AI provider not configured
Still within the 2-second debounce period

Solutions:

Write more content
Check AI connection status
Wait for the debounce timer

Electron Won't Start

Solution:

# Clear node_modules and reinstall
rm -rf node_modules
npm install

# If Electron download fails, use a mirror
ELECTRON_MIRROR="https://npmmirror.com/mirrors/electron/" npm install

High Memory Usage

Solution: Use a lighter Ollama model:

ollama pull phi3

Then update the model in Noschen settings.

Getting Help

GitHub Issues: github.com/maxpolwin/Noschen/issues
Discussions: github.com/maxpolwin/Noschen/discussions

Development

Development Mode

# Start with hot reload
npm run dev

Building

# Build renderer (Vite)
npm run build:renderer

# Build main process (TypeScript)
npm run build:main

# Build everything
npm run build

Type Checking

npx tsc --noEmit

Project Scripts

Script	Description
`npm run dev`	Start development server with hot reload
`npm run build`	Build for production
`npm run package`	Package as desktop app
`npm run lint`	Run ESLint
`npm run typecheck`	Run TypeScript type checking

Roadmap

Planned Features

Export to Markdown and PDF
Web search integration for concrete source recommendations
AI learning from rejected suggestions
Cross-referencing across multiple notes
Counter-arguments and alternative perspectives
Terminology clarification prompts
Collaborative editing (optional cloud sync)

License

MIT License

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

Made with care for researchers, by researchers.

Name		Name	Last commit message	Last commit date
Latest commit History 62 Commits
.github		.github
build		build
docs		docs
models		models
scripts		scripts
src		src
.gitignore		.gitignore
PACKAGE_VERSION_ANALYSIS.md		PACKAGE_VERSION_ANALYSIS.md
README.md		README.md
package-lock.json		package-lock.json
package.json		package.json
tsconfig.json		tsconfig.json
tsconfig.main.json		tsconfig.main.json
tsconfig.node.json		tsconfig.node.json
vite.config.ts		vite.config.ts

Folders and files

Latest commit

History

Repository files navigation

Noschen

Table of Contents

Features

Core Features

AI Analysis Types

Additional Features

Screenshots

Installation

Prerequisites

Quick Start

Build for Production

Installing Ollama (Recommended)

Configuration

AI Provider Setup

Option 1: Ollama (Local LLM) — Recommended

Option 2: Mistral API (Cloud)

Usage Guide

Creating Your First Note

Document Structure Best Practices

Working with AI Suggestions

Excluding Sections from AI Feedback

AI Feedback System

How It Works

Feedback Types Explained

MECE Validation (Purple)

Gap Identification (Blue)

Source Suggestions (Green)

Structural Improvements (Orange)

Feedback Quality Tips

Keyboard Shortcuts

Global Shortcuts

Feedback Shortcuts

Editor Shortcuts

Architecture

Project Structure

Technology Stack

Data Flow

Data Storage

Storage Location

Note Format

Settings Storage

Backup

Troubleshooting

Common Issues

"AI Disconnected" Status

No Feedback Appearing

Electron Won't Start

High Memory Usage

Getting Help

Development

Development Mode

Building

Type Checking

Project Scripts

Roadmap

Planned Features

License

About

Resources

Uh oh!

Stars

Watchers

Forks

Releases

Packages 0

Uh oh!

Uh oh!

Contributors

Uh oh!

Languages

Packages