This is a submission for the Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady

What I Built

I built SlothDesk — an offline, zero-cloud desktop file cleaner and auto-organizer designed to reward total laziness.

Everyone dreads sorting their overflowing Downloads or Desktop folder. Messy downloads with cryptic names like receipt_9921_oct.pdf, assignment_final_v2.docx, and temp_scraper_v2.py sit around for weeks because manually inspecting, renaming, and filing them is tedious cognitive work.

SlothDesk lets you channel your inner sloth. It inspects file headers and text previews, asks a local open-weight model to semantically categorize and standardize the filenames, and automatically moves them into structured directories (such as Invoices & Finance, Coursework & Notes, Code & Dev, and Media). You kick back, take a nap, and let local open-source AI do all the digital chores in the background.

Demo

• Live Interactive Demo: https://sandipsam2025.github.io/SlothDesk/

The web demo showcases the complete pipeline in action: an animated sloth companion, drag-and-drop clutter intake, real-time simulated offline inference streams, and instant reorganization into structured vaults.

Code

• GitHub Repository: https://github.com/SandipSAM2025/SlothDesk

SandipSAM2025 / SlothDesk

SlothDesk 🦥

Slow & Steady Offline AI File Tidier
Channel your inner sloth — let open-source AI clean your digital clutter while you nap.

---

📌 Overview

SlothDesk is a zero-cloud, local desktop utility and interactive showcase designed for the "Slow and Steady" Open-Source AI Challenge (celebrating International Sloth Day).

Instead of wasting cognitive energy manually inspecting, renaming, and filing messy downloads (receipt_9921_oct.pdf, assignment_final_v2.docx, temp_scraper_v2.py), SlothDesk utilizes a compact local open-weight model (Llama 3.2 3B via Ollama) to semantically inspect file headers, generate descriptive names, and organize files into dedicated directory vaults automatically.

---

💡 Why Open Innovation Matters

Advantage	Why It Matters for SlothDesk
🔒 100% Zero-Leak Privacy	Invoices, medical records, tax receipts, and proprietary source code never leave your device. Model inference runs strictly against local memory with zero external telemetry.
⚡ True Offline Independence	Works reliably on airplanes, rural retreats, and during internet

…

View on GitHub

How I Built It

SlothDesk is built around open-source AI and local inference:

• Local Inference Engine: Powered by Ollama running open-weight models locally (optimized for Llama 3.2 3B or Phi-3.5-mini).
• Core Automation Daemon: Written in Python using the official ollama client library, os, and shutil. It reads a lightweight 200-byte preview snippet from each file and queries the local model with strict JSON schema constraints:

json
  {
    "category": "Invoices_and_Finance",
    "new_name": "Electricity_Bill_Oct2026.pdf"
  }