Touch Grass Sports AI – Offline AI That Gets You Outside

This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass

What I Built

Touch Grass Sports AI is a fully offline open-source AI that generates personalized outdoor sports challenges.

Its only mission: get you off the screen and into the real world.

You pick your fitness level, sport, available time, and location. It replies with a concrete challenge that must be done outside and always includes a real “touch grass” moment.

Supports many sports: running, basketball, cycling, yoga, swimming, climbing, parkour, volleyball, tennis, skateboarding, martial arts, hiking, and more.

Works in two modes:

• Local LLM (Ollama) for creative challenges
• Pure rule-based mode (no model needed)

Demo

Web-UI

CLI

Code

GitHub: https://github.com/gabutersproject/touch-grass-sports-ai

How I Built It

Built entirely around open-source AI and local inference:

• Ollama for running open-weight models locally (Llama 3.2, Phi-3, Gemma 2, etc.)
• Strict system prompts that force every output to be outdoor-only + include a touch-grass action
• Multi-language support (English + Indonesian)
• Gradio web UI + pure CLI
• Full rule-based fallback so it works even without any LLM

Why Does Open Innovation Matter?

A closed cloud API would have killed the whole idea.

• Runs completely offline (useful on trails and parks with no signal)
• Zero data leaves the device
• Free forever, no API costs
• Fully customizable (swap models, edit prompts, expand challenges)
• Aligns with the “touch grass” philosophy — less dopamine, more real life Open-weight models + local inference made this possible.

Thanks for reading!

Go outside. Touch grass. Move your body. 🏃‍♂️🌿