This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass TrailQuest is an AI-powered (Gemma) outdoor activity idea generator designed to help people spend less time online and more time exploring the world around them.
Instead of endlessly consuming content on a screen, users can generate simple, personalized outdoor missions based on their interests, available time, preferred pace, mobility, surroundings and whether they want to explore alone or with others.
Some missions might encourage a short nature walk, observing birds, exploring a nearby area, or trying a small outdoor challenge. The goal is to make stepping outside feel approachable, enjoyable and easy to fit into everyday life.
TrailQuest also learns from user feedback. After attempting a mission, users can record whether they completed it, how difficult they found it, how much they enjoyed it, and what they would like to do differently next time. This feedback can help personalize subsequent suggestions.
https://github.com/Veerhan-glitch/trailquest/ TrailQuest uses Gemma 4 E2B, an open-weight AI model, running locally through Ollama. The model acts as the mission-generation engine, turning users’ interests, available time, preferred activity pace, mobility needs, and social preferences into personalized outdoor missions.
The application passes these preferences, along with relevant feedback from previous missions, to the model to guide its suggestions. After completing a mission, users can record their experience, helping inform future recommendations.
By running inference locally, TrailQuest can generate missions without relying on a proprietary, hosted AI API. The model is the core of the experience: it transforms user preferences into actionable ideas designed to get people away from their screens and outdoors.
Open innovation makes experimentation more accessible.
For a small project like TrailQuest, open-weight models and tools such as Ollama make it possible to explore AI-powered experiences without building an entire machine-learning stack from scratch or requiring a paid proprietary inference API. It also makes the architecture easier to inspect, modify, and extend. Developers can experiment with different prompts, model configurations, and personalization approaches while learning how AI systems behave in a practical application.
Most importantly, this project explores a different use of AI: rather than maximizing engagement with a device, can AI help people disengage from their screens and reconnect with their surroundings?
I see open innovation as an opportunity to build useful, adaptable tools that more people can experiment with, learn from, and improve.
Gemma