🌿WindQuest-Explore more. Scroll less.

A developer built WildQuest, an open-source outdoor adventure app that uses local Gemma inference via Ollama to generate personalized nature quests with missions and safety guidance. The React, TypeScript, and Node.js app keeps prompts and journal entries on-device, with a fallback quest generator for when local inference is unavailable.

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass https://dev.to/challenges/hacktoberfest-week1-2026-10-05 WildQuest is an open-source, local-AI outdoor adventure app that turns an ordinary walk into a personalized nature quest. The idea is simple: instead of spending more time on your phone, use AI to discover reasons to put it away and explore the world around you. Users can choose their preferred activity, duration, difficulty, and accessibility preferences. Gemma generates a nature quest with actionable missions and safety guidance. Users then head outdoors, complete the missions, and record their observations in a private nature journal. What WildQuest offers: WildQuest is designed to make the screen a starting point for exploration, not the destination. https://dev-to-uploads.s3.us-east-2.amazonaws.com/uploads/articles/xh3ru4lhn84su4h49ape.png https://dev-to-uploads.s3.us-east-2.amazonaws.com/uploads/articles/xh3ru4lhn84su4h49ape.png https://dev-to-uploads.s3.us-east-2.amazonaws.com/uploads/articles/u1ya1kq0y9piyz8rizba.png https://dev-to-uploads.s3.us-east-2.amazonaws.com/uploads/articles/u1ya1kq0y9piyz8rizba.png The demo will show the complete journey: configuring a quest, generating it with local Gemma inference, completing missions outdoors, and saving observations in the nature journal. GitHub repository: https://github.com/yashr5120/wildquest https://github.com/yashr5120/wildquest Built with React, TypeScript, Vite, Tailwind CSS, Node.js, Express, Zod, Ollama, and Gemma. The repository includes the frontend, local AI integration, quest-generation logic, browser-side journal, PWA configuration, and automated tests. I built WildQuest around Gemma 4 E2B gemma4:e2b , using Ollama to run the open-weight model locally on my computer. The application follows a straightforward architecture: I also included a fallback quest generator so users have an alternative when local AI inference is unavailable. Fallback quests are separate from model-generated content. The focus was to keep the architecture simple, avoid unnecessary cloud services, and make the core experience practical for a small open-source project. For WildQuest, open innovation is about giving people more control over the AI they use and the data they create. Privacy by design: Running Gemma locally means quest-generation prompts do not need to be sent to a hosted AI provider. Journal entries remain in the browser by default. More independence: Once the model is installed and the local runtime is available, the application can generate quests without relying on a paid, per-request cloud AI API. Freedom to experiment: Developers can modify the prompts, experiment with model configurations, and adapt quests for different environments and communities. Community-driven improvement: An open-source implementation makes it easier for contributors to improve quest quality, accessibility, validation, and support for additional models. Most importantly, local AI makes the experience more personal without requiring users to hand their observations to a third-party service. For a project about getting people outside, I wanted the technology to be useful without making people more dependent on their screens.