This is a submission for the Hacktoberfest Open-Source AI Challenge Week 1: Touch Grass.
What I Built
Ritmo Claro is a post-run progress tracker for runners. You choose a goal and log runs; the app calculates pace, distance, pulse, perceived effort, and outing frequency, then shows how those numbers change over time.
The outdoor part is the run itself. Ritmo Claro is designed for a short review after a run, not something to watch while moving. It does not plan routes or workouts, and it cannot make someone go outside on its own. I built and deployed the app during this challenge week.
Demo
Code
How I Built It
The interface is plain JavaScript, HTML, and CSS, with a Cloudflare Worker handling the AI request. The app calculates every metric in code; the model only explains the supplied results. It uses the open-weight Qwen 3.8 27B model through Cloudflare Workers AI by default.
The AI reading is user-triggered. The Worker receives the calculated analysis and recent run notes, then returns one short global reading. If the model is unavailable or out of quota, the app has a calculated local fallback. Google sign-in stores a runner's goal and runs in one Drive file; guest mode uses clearly fictional sample runs.
Why Does Open Innovation Matter?
Using an open-weight model gives this project a path to change models or providers without handing over the metric calculations. The numbers and comparisons stay in readable application code, so the model can explain the analysis without deciding what counts as progress.
The deployed version still sends an AI request to Cloudflare Workers AI, and Google users keep their records in Drive. This is not offline inference or a privacy claim. The value of the open model here is that the project can keep the explanation layer replaceable while its calculations remain inspectable and independent of any one model provider.
Prize Categories
I am entering the overall challenge only; I am not claiming a partner category.
Top comments (0)