Downloads folders that have become archaeological sites. Project directories where every file is labeled 'final_v2_REAL_final.docx'. Thousands of photos organized by the camera date rather than by anything meaningful. If this sounds familiar, File Organizer was built for you.
File Organizer is a cross-platform desktop app (macOS, Windows, Linux) that scans a directory, uses an AI model to classify each file into meaningful categories and subcategories, and then organizes them into a folder structure you control. Crucially, by default the classification runs entirely on your machine — no files leave your disk.
How It Works
The app reads each file's name, extension, and optionally its content, then sends that information to an LLM for classification. The model returns a suggested category and subcategory. You review the proposed organization before applying anything — File Organizer shows you what it plans to do and waits for approval.
LLM Provider Options
- Managed Local LLM: the app downloads and manages a local model automatically — no manual setup required
- LM Studio: connect to a locally-running LM Studio server (default: http://localhost:1234)
- Ollama: connect to Ollama running a model like llama2 or mistral (default: http://localhost:11434)
- OpenAI: cloud classification using GPT models with your API key
- Anthropic (Claude): cloud classification via Claude API
- Groq: fast cloud inference via Groq API
- Google Gemini: classification via Gemini
- Custom: any OpenAI-compatible API endpoint
Scan Control
For large directories, you can pause a scan mid-way and resume from where you left off. Stopping a scan early shows a preview of all files processed so far and sends partial results for category optimization. This is useful when you want to inspect the model's classification decisions before committing to the full directory. A 'New Scan' button resets all state for a fresh start.
Built With Tauri (Rust + React)
File Organizer uses Tauri — a framework that pairs a Rust backend with a React frontend — to deliver a small, fast, native binary without Electron's memory overhead. The result is a lean desktop app with a real system tray icon, native file dialogs, and cross-platform installers (DMG for macOS, MSI/EXE for Windows, DEB/AppImage for Linux).