🌱 GreenBuddy AI: An AI Nature Companion That Encourages People to Touch Grass

A developer built GreenBuddy AI, a Streamlit-based nature companion that recommends outdoor activities based on a user's available time, interests, and location. The initial prototype ran Ollama with the Qwen2.5:3b model locally, but the public cloud deployment switched to a hosted AI API because the local Ollama service could not be reached by the cloud app, a lesson the developer cites about model hosting, API security, cost, and data persistence.

Live Demo: https://greenbuddy-ai-kxqx3wshurh5ggzjx9nb4v.streamlit.app/ https://greenbuddy-ai-kxqx3wshurh5ggzjx9nb4v.streamlit.app/ GitHub Repository: https://github.com/kavin553/GreenBuddy-AI https://github.com/kavin553/GreenBuddy-AI We spend a lot of time looking at screens, even when we have free time to enjoy the world around us. Sometimes, we want to go outside or try something new, but we don't know what activity to choose. I built GreenBuddy AI , a nature companion that recommends practical activities based on a person's available time, interests, and surroundings. Users can select their available time, interests, and preferred location to receive a personalised activity recommendation. Key features include: Building GreenBuddy AI taught me how to connect an AI model to a user-facing application, design a simple interaction flow, handle API errors, and deploy a Python application online. My initial local prototype used Ollama with the Qwen2.5:3b model. For the public cloud deployment, I switched to a hosted AI API because the local Ollama service on my laptop could not be accessed directly by the cloud app. This was an important lesson: deploying an AI application requires decisions about model hosting, API security, cost, and data persistence—not just building the interface. Open models and open-source tools give developers opportunities to experiment, learn, and build applications they can adapt to their needs. My initial local-model experiment helped me understand the benefits and limitations of running AI on a personal device. I want to explore a deployment that supports local open-weight inference alongside a hosted option. Source Code: https://github.com/kavin553/GreenBuddy-AI https://github.com/kavin553/GreenBuddy-AI I'd love to hear your suggestions: what small activity would help you spend less time on screens and more time exploring the world? Let's make technology a companion to real-world experiences, not a replacement for them. 🌿