AI Classification
AI-powered file organization running entirely on your device. No API keys, no cloud accounts — Talyx uses open-source models to classify files by content. Semantic classification sorts invoices, contracts, reports, and photos intelligently. Choose your preset: Light (fast), Balanced (default), or Power (best accuracy).
How It Works
Talyx scans your files
Select or drag any folder into Talyx. It scans every file and extracts signals — extensions, metadata, text snippets, and image content.
Local multi-model AI pipeline
Four specialized models analyze your files on-device: text extraction identifies document content, SmolVLM understands images and video frames, Florence-2 performs OCR on scanned documents, and GLiNER extracts named entities. A coordinator LLM (Qwen2.5) combines all signals into semantic classifications — invoices go to Documents/Invoices, vacation photos to Photos/Travel, even if the filename is IMG_0042.jpg.
Preview and sort with one click
Review the proposed organization before applying. Choose to copy or move files, then sort with one click. Every operation is fully undoable.
Key Benefits
Semantic understanding
AI reads content, not just extensions. Invoices, contracts, reports, and photos are classified by what they actually contain.
3 presets, swappable models
Light for fast sorting, Balanced for everyday use, Power for maximum accuracy. Swap in any compatible model from HuggingFace.
No API keys or accounts
All AI runs locally via open-source models. No cloud subscriptions, no API keys, no accounts. Download models once and classify forever.
Fully offline after setup
After the one-time model download (~2GB from HuggingFace), Talyx works completely offline. No internet required for any classification task.
Ready to organize your files?
Join the waitlist to be the first to try Talyx.