🦥SlothAI: A Local-First AI Assistant That Automates Repetitive Tasks with Ollama

Ayush Kumar — Sun, 25 Oct 2026 18:06:45 +0000

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

What I Built

🦥 SlothAI — Work Less. Automate More. Stay in Control.

What if you could spend less time organizing files, reading lengthy documents, and manually creating tasks—and more time doing the work that actually matters?

That's the idea behind SlothAI, a local-first AI productivity assistant I built for the Hacktoberfest Open-Source AI Challenge.

SlothAI uses Ollama-compatible local AI models, deterministic fallback algorithms, and SQLite to simplify repetitive productivity tasks while keeping users in control of their data and actions.

🎯 The Problem

Developers, students, and knowledge workers frequently spend time on repetitive tasks:

Reading lengthy meeting notes to find the important points.

Copying action items from documents into a task manager.

Organizing cluttered Downloads and project folders.

Identifying overdue tasks and deciding what to prioritize each morning.

These tasks aren't necessarily difficult, but they consume time and attention that could be spent building, learning, or solving problems.

✨ What SlothAI Can Do

📄 1. Document Summarization

Upload supported .txt, .md, .csv, .json, or .log files and generate executive summaries, key takeaways, and candidate action items.

Long documents are processed in manageable chunks to support summarization.

✅ 2. AI-Assisted Task Extraction

Turn meeting notes into structured tasks with priority labels and due dates.

Extracted tasks appear in a review panel, allowing users to choose which items to save instead of automatically accepting every AI suggestion.

📁 3. Preview-First File Organizer

Scan a selected workspace and preview proposed file categories:

Documents

Spreadsheets

Images

Videos

Audio

ayushkumarak018-stack / SlothAI-Your-Local-AI-Productivity-Assistant

Work Less. Automate More. Stay in Control.

SlothAI 🦥

Your Local AI Productivity Assistant

"Work Less. Automate More. Stay in Control."

A submission for the Hacktoberfest Open-Source AI Challenge, Week 3: "Slow and Steady" (https://dev.to/challenges/hacktoberfest-week3-2026-10-19).

1. Problem Statement

Modern productivity tools demand constant manual interaction, charge expensive monthly per-token subscriptions, and upload proprietary corporate or personal documents to commercial cloud AI vendors. Users are caught between tedious repetitive digital work (sorting downloads, extracting to-dos, summarizing notes) and surrendering their private data to hosted AI providers.

2. The Solution

SlothAI is a privacy-first, desktop-friendly web application powered by open-weight language models running locally via Ollama. It acts like a patient personal assistant that prepares your day, organizes your digital clutter, and extracts actionable tasks—all while keeping 100% of your data on your local device.

3. Key Features

🦥 Local AI Assistant: Chat with an assistant that maps natural language requests to safe, allowlisted productivity…

View on GitHub

Explore the source code, inspect the implementation, and share feedback or suggestions.

How I Built It

SlothAI combines a modern web interface, local AI inference, structured persistence, and safety-focused automation.

🧠 Local AI with Ollama

SlothAI communicates with Ollama through its local HTTP API.

The integration supports:

Discovering locally installed models.

Checking Ollama connectivity and model availability.

Sending inference requests for supported AI workflows.

Validating structured AI responses before using them in application logic.

The application distinguishes between backend availability, Ollama connectivity, model availability, and the active inference mode.

If Ollama is unavailable or the selected model is missing, supported workflows can use a deterministic local fallback instead of silently sending private content to a hosted AI service.

This makes the application's behavior more transparent.

🧩 Model Choice and Licensing

SlothAI is designed to work with lightweight models available through Ollama, including examples such as:

qwen2.5:3b

mistral:7b

phi3:mini

llama3.2:3b

Actual hardware requirements depend on model size, quantization, context length, and other applications running on the machine.

One important distinction: open-weight and open-source are not always interchangeable terms. Each model has its own licensing conditions, which users should review before adopting it.

SlothAI is intended to make local model use more accessible without assuming every model has the same license.

🛠️ Technology Stack

Layer

Technology

Frontend

React, TypeScript, Vite

Styling

Tailwind CSS

Icons

Lucide React

Python backend

FastAPI, Pydantic

Additional server integration

Node.js, Express

Database

SQLite

Local AI

Ollama

Testing

Python unittest, TypeScript node:test

The interface uses a calm, modern design with warm off-white backgrounds, forest-green accents, clear data tables, and responsive layouts.

🔐 Safety and Reliability

Automation should not mean giving up control.

SlothAI incorporates safeguards such as:

Validating filesystem paths and workspace boundaries.

Checking for filename collisions before moving files.

Previewing proposed changes before approval.

Recording file-operation manifests to support undo.

Validating structured AI output.

Keeping activity records focused on operational metadata rather than raw document contents.

Avoiding silent fallback to commercial cloud AI APIs.

These measures are designed to make automation more predictable and transparent.

🧪 Testing

The project includes automated unit tests for backend and frontend logic.

The current test suite is documented as containing:

19 Python unit tests

11 TypeScript unit tests

That is 30 automated unit tests in total. Verify that both suites pass against the version being submitted before publishing this count.

Why Does Open Innovation Matter?

I believe open innovation matters because developers should have more choices in how they build, run, and improve AI-powered applications.

SlothAI explores this idea through four principles.

🔒 Privacy and Data Control

Meeting notes, personal documents, and local file structures can contain sensitive information.

Local inference provides an alternative to sending supported content to a remote AI provider. SlothAI is designed to process these workflows locally when the required components are installed and configured.

🌱 Accessibility

Local models make experimentation possible without necessarily paying recurring per-request AI API fees.

There are still costs associated with hardware, memory, processing power, and electricity, but users have greater control over where inference happens.

🔧 Freedom to Experiment

Developers can inspect the implementation, try compatible models, adapt prompts, and build new workflows around their needs.

That flexibility is valuable for students and independent developers who want to learn by building rather than treating AI as a black box.

🌍 Community-Driven Development

Open-source software creates opportunities for people to improve projects through code contributions, testing, documentation, accessibility improvements, and new ideas.

SlothAI is my attempt to explore how open AI technologies can support practical productivity tools without making every task dependent on a closed cloud service.

My Agent Session

I used an AI-assisted software development workflow while building SlothAI, focusing on application structure, database persistence, document processing, filesystem safety, and automated testing.

The development process involved separating reusable logic into testable modules and designing workflows around explicit user approval.

You can explore the source code and implementation on GitHub:

🔗 SlothAI — GitHub Repository

The Lazy Payoff

The best kind of automation is the kind that removes repetitive work without taking away your ability to make decisions.

Imagine starting your Monday with a long retrospective, a cluttered Downloads folder, and several overdue tasks.

With SlothAI, the intended workflow is straightforward.

📄 For Documents

Upload your notes → Generate a summary → Review extracted tasks → Save the items you approve.

📁 For File Organization

Scan a workspace → Review proposed categories → Approve the changes → Undo supported moves if necessary.

🌅 For Daily Planning

Generate a morning briefing → Identify overdue and high-priority tasks → Decide what to tackle first.

Instead of repeatedly performing the same administrative chores, you can review suggestions and focus on the next meaningful action.

I haven't presented estimated time savings as measured results. The actual benefit depends on the user's workflow, document length, task volume, and hardware.

For me, the core idea behind SlothAI is simple:

Let automation handle repetitive work while humans stay in control of the decisions.

Prize Categories

Hacktoberfest Open-Source AI Challenge — Week 3: Slow and Steady

SlothAI explores local-first productivity and workflow automation using Ollama-compatible open-weight models and open-source development tools.

💭 Final Thoughts

Building SlothAI has been an opportunity to explore a practical question:

Can AI make everyday work easier while giving users more control over their data and actions?

My approach combines local inference, deterministic fallback behavior, preview-first file operations, and human-approved task workflows.

The goal isn't to automate everything blindly. It's to reduce the small, repetitive tasks that interrupt our work every day.

Because sometimes, the smartest productivity improvement is doing less unnecessary work.

🦥 SlothAI — Work Less. Automate More. Stay in Control.

🔗 Explore the Project

GitHub: SlothAI Repository

Live Demo: Try SlothAI

Challenge: Hacktoberfest Week 3: Slow and Steady

Hacktoberfest #OpenSource #AI #Productivity

DEV Community: Ayush Kumar