This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass
What I Built
We spend so much time online looking for something interesting to do that we sometimes forget there’s an entire world waiting outside our doors.
That thought inspired Grounded, a real-world adventure discovery platform I'm building.
💡 The idea
Grounded helps people break out of repetitive routines and discover small, meaningful experiences based on their interests, available time, and budget.
Not every adventure needs to be a mountain trek or an expensive trip. It could be a ten-minute walk, finding an unfamiliar street, photographing something beautiful, discovering a quiet café, or watching the sunset.
The goal is simple: make going outside feel exciting, accessible, and rewarding.
🌱 What I'm building
Personalized discovery: Find activities that match your mood, interests, time, and budget.
Adventure planning: Turn a free afternoon into a practical itinerary.
Explore your world: Discover nature, photography spots, hidden gems, and local experiences.
Community adventures: Find activities to share with friends.
Gamified exploration: Earn XP and unlock achievements as you complete adventures.
🛠️ The technical direction
I'm building the frontend with React and TypeScript, with a focus on interactive discovery, thoughtful UI design, and a smooth user experience.
The longer-term direction includes AI-assisted adventure planning and an architecture that can connect to an open-weight language model and backend services.
I'm developing the experience incrementally, starting with a functional frontend and sample adventure data before connecting real AI services.
Demo
Code
https://github.com/PragyaTandi/grounded-ai
How I Built It
I built Grounded, a real-world adventure discovery platform, using React, TypeScript, and Vite. The idea is to help people break out of repetitive routines and discover small, meaningful experiences outside their homes, whether that means exploring a nearby park, trying street photography, finding a hidden café, or taking an unplanned walk.
The frontend focuses on interactive, mood-based discovery, personalized activity planning, and a gamified exploration experience. I'm designing the architecture so that a dedicated recommendation service can eventually connect the interface to an open-weight language model for more personalized adventure suggestions.
My approach is to start with a functional application and structured adventure data, then introduce AI capabilities as modular services rather than making the entire product dependent on a single model or provider. This makes the project easier to test, extend, and adapt to different inference environments.
Why Does Open Innovation Matter?
Open innovation matters for Grounded because I want the product to remain adaptable, accessible, and independent of a single AI provider.
Using open-weight models could allow me to experiment with local inference, customize recommendation workflows, and evaluate different models for generating useful, context-aware adventure plans. It could also make it possible to run certain AI features on a user's own hardware, reducing dependence on paid, closed APIs and giving developers greater control over data handling.
More importantly, open innovation lets developers learn from existing tools, build on community contributions, and share improvements rather than reinventing everything from scratch. For a project like Grounded, that flexibility creates room to experiment with personalized recommendations while keeping the core experience usable even when an AI service is unavailable.
I see open innovation not just as a way to reduce costs, but as a way to make AI-powered experiences more transparent, adaptable, and accessible to more people.
Top comments (0)