# 🦥 SlothAI: Work Less, Automate More, Stay Slow > Source: > Published: 2026-10-20 19:38:38+00:00 This is a submission for the Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady. What I Built Meet SlothAI — a local-first, open-weight AI productivity workspace designed to take the repetitive work off your plate. We spend too much time summarizing documents, digging through notes for action items, renaming files, and turning scattered information into useful updates. SlothAI brings these everyday tasks together in one workspace, so you can spend less time on digital busywork and more time on what matters. 🦥 What can SlothAI do? AI Document Summarizer: Turn long documents into short, balanced, or detailed summaries, with key takeaways and notable terms. Task Extractor: Identify actionable tasks from your notes, edit them, track completion, and export them. Deadlines are grounded in the source text rather than invented. Intelligent File Organizer: Generate cleaner filenames and destination suggestions, review proposed changes, and approve individual operations before execution. Daily Briefing Compiler: Turn notes and documents into concise summaries, key points, action items, deadlines, and follow-up questions. Local Model Management: Check model availability, inspect backend health, and choose or download compatible local models. The idea is simple: let AI handle the repetitive work without giving up control of your files or workflow. Demo 🚀 Try SlothAI: [https://sloth-ai-lemon.vercel.app/](https://sloth-ai-lemon.vercel.app/) Explore the deployed interface and discover the SlothAI workflow. For local model inference, use the full application with its FastAPI backend and Ollama runtime, as described in the repository's setup guide. Code 💻 GitHub Repository: [https://github.com/SOUMYADEEPDEY1217/Sloth-AI](https://github.com/SOUMYADEEPDEY1217/Sloth-AI) The project is released under the MIT License. Explore the source code, run it locally, experiment with compatible models, and contribute improvements. How I Built It SlothAI combines a modern web interface with a Python backend and local language-model inference. Tech stack Frontend: React 19, TypeScript, Vite, Tailwind CSS Backend: Python, FastAPI, Uvicorn, and Pydantic AI runtime: Ollama Default suggested model: Qwen3 4B (qwen3:4b) Document processing: Python libraries for formats such as PDF, DOCX, TXT, Markdown, CSV, and JSON Testing: Python unit tests and TypeScript/build checks How the architecture works The React frontend provides the workspace for summarizing documents, extracting tasks, organizing files, and compiling briefings. The FastAPI backend validates requests, processes supported documents, and coordinates application logic. Ollama runs a compatible open-weight language model locally and supplies AI-generated results. The application structures outputs and applies additional checks, including source-grounded deadlines and controlled file operations. Safety is part of the design SlothAI is designed to suggest changes before executing them. File organization includes an approval workflow and directory restrictions, with safeguards against path traversal, accidental overwrites, and unsafe file extensions. The goal isn't to make an agent that does everything without asking. It's to make an assistant that saves effort while keeping the user in control. Why Does Open Innovation Matter? For SlothAI, open innovation is not just a technology choice. It shapes how the application can be used, customized, and trusted. 🔒 Privacy through local inference When configured and run locally, SlothAI can process documents through a local FastAPI backend and Ollama rather than sending their contents to a hosted LLM API. This matters when working with personal notes, internal documents, or other information that shouldn't automatically leave your machine. 📴 Useful without a constant internet connection After installing the dependencies and downloading a compatible model, local inference can run offline. This makes supported AI workflows available even when an internet connection is unavailable. The hosted demo and the local application have different requirements: offline inference applies to the locally configured application, not the Vercel-hosted website. 🔧 Freedom to change models The application is designed to support configurable local models. Qwen3 4B is the suggested default, but compatible alternatives can be selected according to hardware, speed, and output quality. This flexibility lets developers experiment with different models instead of being locked into one provider's API. 💸 Fewer dependencies on paid APIs Local inference avoids per-request charges from a hosted LLM provider. There are still costs associated with hardware, electricity, storage, and setup, but running the model locally can make repeated use more predictable. 🛠️ Transparent, modifiable software The MIT-licensed codebase gives developers the freedom to inspect, adapt, and extend the application. The backend's validation and file-operation safeguards can also be reviewed and improved by the community. For this project, the advantage of open AI is control: control over where inference runs, which compatible model is used, how the workflow behaves, and how the code evolves. The Lazy Payoff The biggest payoff is reducing the number of small, repetitive tasks that interrupt a productive day. Instead of manually condensing a long document, searching through notes for action items, or deciding how to rename a batch of files, users can start with AI-generated suggestions and review the results. The file organizer adds another useful detail: it stages proposed changes for review rather than silently moving files. I haven't published a measured time-saving benchmark yet, so I won't invent numbers. The practical goal is to reduce repetitive effort while keeping important decisions and file operations under user control. Why Slow and Steady? A useful AI assistant doesn't have to be a giant cloud service or an agent with unlimited permissions. Sometimes, the better solution is a small, customizable tool that runs on your own machine, handles the boring work, and lets you decide what happens next. SlothAI is my exploration of that idea: less busywork, more control, and open AI that works at your pace. I'd love to hear what repetitive task you'd automate first.