🌿 TouchGrass AI: What If We Used AI to Spend Less Time Online?

A developer built TouchGrass AI, a Python and Gradio web app that uses the open-weight NVIDIA Nemotron 3 Ultra model via OpenRouter to generate personalized outdoor activity suggestions. The app combines user preferences (city, available time, energy level, interests) with current weather data from Open-Meteo to produce weather-aware missions, with prompts instructing the model to avoid fabricated local facts and prioritize safety. The prototype is hosted on Render and its source is available on GitHub.

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass https://dev.to/challenges/hacktoberfest-week1-2026-10-05 We spend so much time using technology that sometimes we forget to step away from it. We open our phones to check one thing, and suddenly we have spent much longer scrolling than we intended. I started thinking about a small question: What if we used AI to help us spend less time online and more time experiencing the world around us? That question became the idea behind TouchGrass AI. 🌱 < -- TouchGrass AI is a small AI-powered web application that turns your preferences into a personalized outdoor activity mission. The idea is simple. You enter your city, choose how much time you have, select your energy level, and tell the app what interests you. It retrieves current weather information and sends those details to an AI model, which generates an outdoor activity suggestion tailored to your situation. For example, someone with only 10 minutes and low energy should not receive the same suggestion as someone with an hour and plenty of energy I wanted the experience to feel approachable rather than overwhelming. You do not need an elaborate plan, expensive equipment, or an entire free afternoon. Sometimes, a short walk, a little photography, or a few minutes observing nature is enough to get started. I also wanted the suggestions to be weather-conscious. The application provides weather information to the model and instructs it to consider unsuitable conditions, avoid inventing local facts, and prioritize personal safety. < -- Try TouchGrass AI: https://touchgrass-ai-w61h.onrender.com https://touchgrass-ai-w61h.onrender.com Video demo link: Please note that the prototype currently depends on external weather and AI services. The weather provider has recently rate-limited some requests, so mission generation may occasionally be unavailable while I work on improving its reliability. 🌱 GitHub repository: https://github.com/mounab3/touchgrass-ai https://github.com/mounab3/touchgrass-ai The repository contains the Python application, dependencies, setup instructions, and project documentation. I wanted to keep the project accessible to people who might want to explore the implementation, learn from it, or contribute improvements. I built TouchGrass AI with Python and Gradio. I chose Gradio because it allowed me to create an interactive web interface without having to build a separate frontend from scratch. Here is how the main components work together: Python handles the application logic and connects the different services. Gradio provides the interface where users enter their preferences and receive their missions. Open-Meteo provides city geocoding and current weather information. OpenRouter connects the application to the selected AI model. NVIDIA Nemotron 3 Ultra generates the personalized outdoor activity suggestions. Render hosts the application so other people can access it online. One important part of the project is the use of an open-weight AI model. Rather than writing every possible activity suggestion manually, I provide the model with a structured prompt containing the user's preferences, available time, energy level, and retrieved weather data. The prompt also includes instructions about safety, practicality, and avoiding fabricated local recommendations. I used OpenRouter to access the model, so inference takes place through a hosted service rather than on my own computer. The project therefore explores using an open-weight model in a practical application, rather than running the model locally. This project also taught me that building an AI application is an important process. External APIs can fail, rate limits can interrupt requests, and an application needs to handle those situations gracefully. Making the demo more resilient is one of the improvements I want to continue working on. For me, open innovation matters because it makes experimentation more accessible. As a computer science student, I am still learning how to turn ideas into complete applications. Access to open-weight models, open-source tools, documentation, and public code repositories gives me opportunities to experiment, learn from existing work, and build something of my own. Open-weight models are interesting because they offer more flexibility to developers who want to explore how AI behaves, customize prompts, and build applications around their own ideas. Depending on the model and its license, they can also offer options for different deployment arrangements. I would not claim that a hosted open-weight model automatically provides every advantage of running a model locally. My project currently relies on an external inference service, and that comes with its own limitations. However, I can still explore how an open-weight model can be used in a real application without needing to host a large model myself. Most importantly, open innovation encourages people to build, share, learn, and improve ideas together. TouchGrass AI is a small project, but it represents something I find exciting about this ecosystem: a student with an idea can combine existing open tools and AI capabilities to create something practical, share it publicly, and continue improving it through feedback. And perhaps that is the part I appreciate most. Best Use of Render — I deployed TouchGrass AI on Render and made the application publicly accessible. Thank you for organizing this challenge It gave me a reason to explore open-weight AI through a project with a purpose I genuinely care about. My hope is that TouchGrass AI reminds us that technology does not always have to ask for more of our attention. Sometimes, it can help us give that attention back to the world around us. Less scrolling. More exploring. See you outside ☀️🌱