This is a submission for the Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady
What I Built
Meet SlothAI 🦥 — a local, privacy-first AI automation suite built around the timeless sloth philosophy: move slow, conserve energy, and let automation do the heavy lifting while you rest.
Modern life buries us in daily digital busywork: endless group chat notifications, chaotic inboxes, scattered download directories, and evening decision fatigue. Most "AI productivity apps" demand monthly subscriptions, cloud accounts, and access to your most intimate personal messages.
SlothAI does the exact opposite. It runs 100% locally on your machine using open-weight models via Ollama. It handles tedious digital chores with zero cloud dependencies, zero recurring fees, and zero data leakage:
- 💬 Group Chat Morning Brief: Ingests hundreds of noisy messages from WhatsApp exports, Slack, or Discord and produces an actionable 2-line TL;DR with key updates.
- 📧 Inbox Tamer: Prioritizes urgent emails, tags routine noise, and drafts polite, context-aware replies automatically.
- 🌙 Offline Bedtime Storyteller: Crafts calming, screen-free stories tailored to any mood—working even when your Wi-Fi is shut off for the night.
- 📁 Smart File & Download Sorter: Inspects messy
~/Downloadsfolders and organizes files semantically instead of relying purely on file extensions. - 🥗 Fridge-to-Recipe Planner: Identifies leftover ingredients and suggests fast, zero-effort meals so you never have to think "what's for dinner?" Who gets to be lazier? Anyone overwhelmed by digital clutter—developers, students, and busy families who want their mornings and weekends back without handing their private life to big tech servers. --- ## Demo Experience SlothAI through both an interactive web simulator and a fast terminal CLI:
-
Interactive Web App & Live Simulator: Open
index.htmldirectly in any web browser to test simulated local model inferences across all 4 modes (Chat Summarizer, Inbox Assistant, Storyteller, and File Organizer) with an aesthetic dark glassmorphism UI. -
Local Live Server: Run
python -m http.server 3000inside the repo and navigate tohttp://localhost:3000. ### Terminal CLI in Action
bash
# 1. Summarize 400 noisy group chat messages in 3 seconds:
$ python slothai.py summarize ./family_chat.txt --model llama3.2:3b
[🦥 SlothAI] Reading local chat log...
[🦥 SlothAI] Running local inference via Ollama (llama3.2:3b)...
------------------------------------------------------------
📋 TL;DR: Mom's birthday dinner is confirmed for Saturday 7 PM at Bistro 9.
📌 Actions: Sarah is picking up the cake; John needs to RSVP by Thursday.
------------------------------------------------------------
✨ 0 bytes sent to cloud. Elapsed: 2.1s. Back to napping.
bash
# 2. Tell a cozy bedtime story fully offline:
$ python slothai.py story --mood calming --theme "a sleepy forest in autumn"
Code
The complete source code, documentation, and web interface are open-source on GitHub:
🦥 SlothAI — Your Lazy Local AI Assistant
"Move slow and automate everything."
Built for Hacktoberfest 2024 · Week 3 Challenge — Slow and Steady
What is SlothAI?
SlothAI is a fully local, open-source AI automation assistant that does your tedious digital chores while you relax — inspired by International Sloth Day (October 20th 🦥).
It runs 100% on your machine using Ollama and open-weight LLMs like:
-
LLaMA 3.2 (3B / 8B)
-
Mistral (7B)
- Phi-3 Mini
-
Gemma 2 (2B / 9B)
-
Qwen 2.5 (3B)
-
LLaVA (vision tasks)
Zero cloud calls. Zero API costs. Zero data leaving your machine.
Features
Feature
Description
Model
💬 Chat Summarizer
Summarizes group chats (WhatsApp/Slack/Discord) to 2-line briefs
llama3.2
📧 Inbox Tamer
Categorizes, prioritizes, and drafts email replies
mistral
🌙 Bedtime Storyteller
Generates calming bedtime stories entirely offline
llama3.1
📁 File Organizer
Renames and sorts chaotic Downloads folders
phi3
🖼️ Fridge → Grocery
Photo-to-grocery-list using local
👉 Direct link: https://github.com/sarthaklole/hacktoberfest-week-3
Repository Highlights
slothai.py: Lightweight Python CLI harness orchestrating local models via Ollama's REST API.
index.html / styles.css / app.js: Zero-dependency, modern dark glassmorphic web dashboard with interactive tabbed simulators.
requirements.txt: Minimal dependencies (requests>=2.31.0).
LICENSE: Permissive MIT License.
How I Built It
SlothAI was built from the ground up around open-weight foundation models and open-source local inference infrastructure:
┌────────────────────────────────────────────────────────┐
│ Your Local Data (Chats / Files) │
└──────────────────────────┬─────────────────────────────┘
│ (Local Filesystem only)
▼
┌────────────────────────────────────────────────────────┐
│ SlothAI Engine (Python CLI / Web) │
└──────────────────────────┬─────────────────────────────┘
│ (http://localhost:11434)
▼
┌────────────────────────────────────────────────────────┐
│ Ollama Runtime │
└──────────────────────────┬─────────────────────────────┘
│
┌───────────────────┼───────────────────┐
▼ ▼ ▼
LLaMA 3.2 3B Mistral 7B Phi-3 Mini
(Chat Summaries) (Inbox & Drafts) (File Sorting)
Local Inference Runtime: Built on Ollama, an open-source framework that handles model quantization, GPU/CPU memory allocation, and provides a clean local REST API (localhost:11434).
Open-Weight Models Tailored per Task:
LLaMA 3.2 (3B): Meta's open-weight model optimized for rapid edge inference. Perfect for high-speed chat summarization on standard 8GB laptops.
Mistral 7B (Instruct): Apache 2.0 licensed workhorse for structured tone-matching, email triage, and response drafting.
Phi-3 Mini (3.8B): Microsoft's MIT-licensed compact model for fast JSON parsing and file metadata categorization.
LLaVA 7B: Open-source multimodal vision model for inspecting fridge snapshots locally.
No Heavy Framework Overhead: Built with pure, clean Python and vanilla web technologies to eliminate bloated dependency chains.
Why Does Open Innovation Matter?
Open innovation is the single reason SlothAI can exist without compromising user values. Here is where open-source AI fundamentally beats closed APIs:
Total Data Sovereignty & Privacy: Group chat backups, private emails, and personal documents contain sensitive, intimate data. Routing those through proprietary cloud APIs creates permanent logs on external servers and risks third-party model training. With local open-weight weights, the data never leaves your computer's RAM.
True $0.00 Operating Cost: Closed APIs charge per token. A daily cron job summarizing 50k tokens of group chats and emails quickly turns into an expensive monthly subscription. With open weights, you pay once in disk space (~2GB) and run inference indefinitely for free.
Unstoppable Offline Resilience: SlothAI works on an airplane, in a remote cabin, or during home internet outages. Closed APIs turn into useless error toasts the second your Wi-Fi drops.
Model Freedom & Swappability: Closed platforms can arbitrarily deprecate endpoints, change safety filters, or raise prices. With open-source models, you own your weights. You can swap LLaMA 3.2 for Mistral, fine-tune custom system prompts, or run quantized 4-bit models tailored to your exact hardware.
My Agent Session
This project was conceived and built with the assistance of an AI pair programmer in Antigravity:
Designed the modular CLI harness in slothai.py connecting directly to Ollama.
Crafted the responsive, dark glassmorphism web client with custom micro-interactions in index.html, styles.css, and app.js.
Configured Git version control and pushed the complete open-source codebase to GitHub.
The Lazy Payoff
Here is the real-world sloth math from using SlothAI:
~45 minutes saved every morning: Skipping through 300+ unread family and developer chat messages; reading a 3-bullet summary over morning coffee instead.
~30 minutes saved each workday: Batch-categorizing unread inbox clutter and auto-generating draft replies for confirmation.
100% cognitive relief: No context switching, no decision fatigue, and zero anxiety over cloud API keys or billing quotas.
You configure it once, create a simple scheduled shortcut, and let your machine work while you nap.
Prize Categories
Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady (Primary Category)
Top comments (0)