Offscreen: Local AI That Gets You Off Your Phone and Outside

Sourav Aarav — Sun, 11 Oct 2026 18:32:08 +0000

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass

What I Built

I built Offscreen, an open-source project that helps people spend less time scrolling and more time outside.

The idea came from a simple problem: we spend hours on our phones, sometimes without even knowing what else to do. Offscreen gives you a reason to put the phone down. It suggests outdoor missions, lets you set a timer, and helps you track completed activities in a personal journal.

It has an Android app and a browser version, with local AI options for generating tasks on supported devices. There's also a Streamlit prototype.

It's still a work in progress, and I'm building it as a solo developer. Some devices struggle with larger models, and browser generation can be slow, so there's plenty left to improve.

Demo

Live web app: https://ouuwz.github.io/OffScreen/

GitHub repository: https://github.com/ouuwz/OffScreen

The web version runs best on devices and browsers that support the required WebGPU features. Android model compatibility and performance can vary by device.

Code

Source code: https://github.com/ouuwz/OffScreen

The repository includes the Flutter Android app, browser app, Streamlit prototype, and GitHub Actions workflows for testing, deployment, and release builds.

How I Built It

I built Offscreen with Flutter for Android and JavaScript with Vite for the browser version. I also kept the original Streamlit prototype.

For local AI, the Android version uses a native llama.cpp integration to load compatible GGUF models. The browser version uses WebLLM with WebGPU on supported devices. The prototype uses Ollama.

The goal is to let users generate outdoor activities locally instead of depending entirely on a hosted AI API. Model files are downloaded separately, so users need to choose a model that their device can handle.

There are trade-offs. Large models can exceed Android's available memory and crash the app, while browser inference can temporarily freeze or slow down some devices. These are known issues I'm working on.

Why Does Open Innovation Matter?

Open innovation matters because people should be able to inspect, modify, and run the tools they use. With local AI, developers can experiment without making every request depend on a paid, closed API.

For a project like Offscreen, that also means users can run supported models on their own hardware and keep their journal on their device. They aren't required to send every activity or prompt to a hosted AI service.

Open source also means other developers can help improve model compatibility, performance, accessibility, and the overall experience. I have a lot left to learn and improve, and making the project public gives others a chance to contribute.

Images and Demo

1. Offscreen: A Little More Outside

2. Plan Your Next Outdoor Mission

3. A Private Nature Journal

4. Track Your Progress

5. Local AI and Privacy Settings

My Agent Session

I don't have a public DevRelay session to share yet.

About Me

I'm @ouuwz, a solo student developer building Offscreen in the open. This is an early release, not a finished product. I'm sharing it now so people can try it, report bugs, and help make it better.

GitHub: https://github.com/ouuwz
Project: https://github.com/ouuwz/OffScreen

DEV Community: Sourav Aarav

[Boost]