Walking Adventure

Ayesha Rahman — Fri, 09 Oct 2026 18:10:21 +0000

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

What I Built

What if AI could help us put our phones away instead of keeping us glued to them?

I built Offline Adventure Generator, a web app that turns an ordinary walk into a personalized outdoor mission.

Choose your available time, walking difficulty, mood, and company. The app generates a mission with creative challenges designed to help you notice the world around you.

For example, a mission might ask you to discover different textures, observe unusual shapes in nearby buildings, or find something you've never noticed before.

The app also includes an interactive checklist, progress tracking, reflection prompts, and adventure history.

The goal is simple: make stepping outside feel like an adventure, even when you have no destination in mind.

Demo

Code

https://github.com/ayesha-rahman01/adventure-walking

How I Built It

The app uses Ollama with the qwen2.5:3b model to generate personalized outdoor missions locally.

The main components are:

-Local AI: Generates missions based on the user's selected preferences.

-Structured output: Requests structured JSON so missions can be validated and rendered consistently.

-Offline fallback: Uses built-in adventure templates if local AI generation fails.

-Service worker: Caches app assets to support offline access.

-Local storage: Saves adventure history, progress, and reflections on the device.

I tested real browser-based generation with the local model and verified the mission checklist, reflection saving, and adventure history.

Why Does Open Innovation Matter?

For an app designed to get people away from their screens, depending entirely on a cloud AI API would undermine the idea.

Using a locally running, open-weight model makes it possible to generate personalized missions without sending prompts to a third-party AI service.

It also makes experimentation more accessible. Developers can inspect the integration, change the prompts, experiment with different models, and run inference on their own hardware.

The offline fallback adds another layer of resilience: the app can still offer adventures when AI generation is unavailable.