OutsideIn: Turning Open-Weight AI Into Tiny Outdoor Adventures 🌿

Dhruv Dagar — Mon, 12 Oct 2026 06:33:15 +0000

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

What I Built

OutsideIn — Local AI Micro-Adventures is a small web app built around a simple idea: sometimes we don't need another reason to stay online. We need a little inspiration to step outside.

OutsideIn helps people turn a few spare minutes into a small outdoor adventure. Choose how much time you have, what interests you, where you can explore, and the pace that feels right. The app generates a personalized mission with practical steps, an observation goal, a reflection question, and a safety note.

Whether it's noticing birds, observing leaves, taking a mindful break, or exploring a nearby green space, the goal is to make spending time outdoors feel easy and approachable.

Demo

Try OutsideIn: https://outside-in-beta.vercel.app

Source code: https://github.com/dhruv-dagar/OutsideIn

Note: The deployed demo uses curated sample missions. For local AI generation, run OutsideIn with Ollama and the Gemma 3 1B model.

Code

The project is open source: https://github.com/dhruv-dagar/OutsideIn

Contributions, ideas, bug reports, and improvements are welcome.

How I Built It

OutsideIn uses:

React + Vite for the interactive frontend.
Express and Node.js for the local API.
Gemma 3 1B through Ollama for local, open-weight AI inference.
localStorage to keep the user's adventure history in their browser.
Vercel to host the public demo.

The local setup is designed to generate adventures using a model running on the user's own machine rather than requiring a paid, hosted LLM API. The public deployment currently provides a curated demo experience.

Why Does Open Innovation Matter?

Outdoor routines are personal. People have different schedules, interests, energy levels, and places they can access. Open-weight models make it possible to experiment with personalized experiences without making a proprietary AI service the core dependency.

Running Gemma locally also gives users more control over where inference happens and can avoid per-request model API costs. The code is available to inspect, adapt, and extend, so others can experiment with new types of outdoor missions, models, and accessibility features.

For me, the point of this project is that AI should not always try to keep us engaged with a screen. Sometimes the best result is a small suggestion that helps us put the device away and pay attention to the world around us.

Prize Categories

Best Use of Gemma — using Gemma 3 1B for local mission generation.

Thanks for checking out OutsideIn. 🌱