I Built TidyByte to Make Document Organization Less Tedious

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 3: Slow and Steady. What I Built Have you ever downloaded a document, saved it with a confusing filename, and promised yourself you'd organize it later? That small task can turn into a pile of forgotten files. I built TidyByte to make document organization less tedious, with reviewable suggestions and a one-click ZIP export. The idea is simple: let software handle repetitive housekeeping while keeping the user in control. TidyByte is a privacy-conscious document organizer that lets users upload TXT and PDF files, review suggested filenames and categories, edit suggestions, approve the results, and export everything into a neatly structured ZIP archive. What it can do Upload TXT and PDF documents in batches. Suggest filenames, categories, and descriptions. Review, edit, and approve suggestions before exporting. Search and filter documents. Export approved files into a ZIP containing category folders, original files, and a README. Download the generated archive again when needed. Work without an external AI API key in the current demo. Demo Live application: TidyByte on Vercel Source code: TidyByte on GitHub I tested the deployed workflow with text files and PDFs, including a PDF larger than 30 MB. I also verified that the generated ZIP archive contained valid files and the expected folder structure. Code TidyByte is built with React, Vite, and TypeScript. The repository includes the source code, tests, and setup instructions: Explore TidyByte on GitHub How I Built It I wanted to build something useful without creating another application that depends on a paid API or sends every document to a third-party service. I used Antigravity as my development environment to build the application, refine the interface, troubleshoot upload and export behavior, and validate the project. I then pushed the code to GitHub and deployed the frontend using Vercel. The current version uses deterministic, rule-based suggestions, including filename heuristics. It does not yet run an open-weight language model. The current demo focuses on delivering a working document-management workflow, while model-powered classification and descriptions remain future work. I also paid attention to practical details, such as handling duplicate uploads, validating file sizes, preserving access to original files, and making ZIP exports easy to retrieve again. Why Does Open Innovation Matter? For a tool that handles personal documents, privacy should be part of the design rather than an afterthought. Open-weight models could eventually let TidyByte analyze documents without relying on a closed AI API. Depending on the model and deployment setup, inference could run locally or through a hosted provider. Open implementations would also give developers more freedom to change models, inspect behavior, and avoid being locked into a single provider. Cost matters too. I want to explore an approach that remains usable without per-request inference fees, rather than building a feature that becomes expensive as usage grows. For now, I have deliberately kept the demo independent of an AI service. My next step is to integrate a suitable open-weight model through a provider or deployment option that fits a strict ₹0 inference budget, while being transparent about any data sent to a hosted service. Open innovation matters because it gives developers the freedom to experiment, adapt tools to real needs, and make thoughtful choices about cost and privacy. My Agent Session I used Antigravity to develop and refine TidyByte, test its functionality, and work through implementation issues. I haven't attached a DevRelay session link. You can explore the implementation and project setup through the public GitHub repository. What's Next? My next goals are to integrate an open-weight model for more useful document classification and descriptions, improve handling of scanned PDFs, and keep the review-and-approve workflow under the user's control. TidyByte is a small step toward making everyday digital housekeeping less tedious—one document at a time.