{
  "id": 15356691,
  "title": "Folio: Organize your files using AI on-device",
  "url": "https://urgent.news/2026/10/18/folio-organize-your-files-using-ai-on-device",
  "topic": "world",
  "section": "World",
  "published": "2026-10-18T05:30:02.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/aravindsathesh/folio-organize-your-files-using-ai-on-device-3421"
  },
  "original_language": "en",
  "account": "This article is about a new macOS menu bar app called Folio, which uses local AI to organize downloaded files. Users often accumulate invoices, receipts, screenshots, lecture notes, and random PDFs in their downloads folder. Folio aims to automatically organize these files without interrupting the user's workflow. When a file arrives in a watched folder, Folio employs a local Gemma model to suggest a filename and destination. For instance, an invoice (3).pdf might be renamed to Finance/Invoices/Work/Cobalt Cloud Hosting Invoice 2026-10.pdf.\n\nFolio also features a Restructure mode, which allows users to review proposed changes before applying them. If Folio is unsure where a file belongs, it leaves it alone and adds it to a review list. The app does not delete or overwrite files and logs every move before execution.\n\nThe development of Folio faced the challenge of ensuring the safe use of AI to modify the filesystem. Folio utilizes two models through Ollama: Gemma (gemma4:12b) for suggesting filenames, destination folders, and restructuring plans, and mxbai-embed-large for retrieving similar past corrections to adapt to the user's filing preferences. The application manages text extraction, folder discovery, and validation.\n\nTo test Folio's models, a 50-file evaluation set containing various types of files was created. The results showed that the 12B model achieved the best accuracy and processing time, with 98% of auto-moved files correct, only 2% incorrect across all files, and an average time of 7.2 seconds per file. The smaller model was less accurate but had faster processing times, averaging 1.9 seconds per file. The 12B model was chosen for its accuracy and speed, and the thinking mode added significant latency, which was disabled to reduce processing time.\n\nFolio is designed for use by freelancers and small business owners who regularly handle sensitive documents. The app is built using Swift 6 and SwiftUI, relying on Apple's native frameworks and avoiding third-party packages. With a one-time license price of $19.99, Folio aims to make file organization a once setup-and-forget solution, operating in the background without sending sensitive documents to the cloud.",
  "summary": "This is a submission for the Hacktoberfest Open-Source AI Challenge Week 2: Side Hustle What I Built I built Folio, a macOS menu bar app that organises downloaded files using local AI. Downloads folders tend to accumulate invoices, receipts, screenshots, lecture notes, and random PDFs. I wanted to build something that could organise these files automatically without requiring me to stop what I’m…",
  "key_points": [
    "Folio is a macOS menu bar app using local AI to organize files.",
    "Gemma model suggests filenames and destination folders for files.",
    "Restructure mode allows users to review proposed changes before applying."
  ],
  "editors_take": "By processing sensitive documents through local models instead of the cloud, Folio offers small business owners a privacy-focused method to automate the tedious task of digital file management.",
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}