# Jofit — ML wellness coach

> A fitness & wellness PWA that turns water intake, training load and blood-work into one ranked daily suggestion — each tagged by urgency tier and linked to its source paper.

- **Source:** https://amplifiedcreations.com/lab/jofit
- **Publisher:** amplified creations (https://amplifiedcreations.com/)
- **Published:** 2026-05-23
- **Language:** en
- **Stack:** Next.js, PWA, Machine Learning, Drizzle, Recharts, TypeScript
- **Live demo:** https://ptthe13.github.io/amplifiedlabs/jofit/
- **Source code:** https://github.com/PTthe13/amplifiedlabs/tree/main/jofit

---

Most wellness apps hand you a number. Jofit hands you a move.

It pulls together what you've actually done — water logged, kilocalories in and out, training sessions, the last blood panel — and runs each candidate intervention through a small ranker. Three axes only: **evidence weight** (RCT > cohort > case), **personal delta** (how far you are from the clinical target), and **adherence cost** (how much friction the suggestion adds to your day).

The top-scoring nudge surfaces as today's move. Anything in the **URGENT** tier — a vitamin D in the severe-deficiency band, a sleep debt that's now metabolic — jumps the queue and always shows first, with the source paper one tap away.

The interface is deliberately quiet. Editorial dark, three surface elevations, tabular numerals, and a single lime accent reserved for adherence and confirmation. A five-colour semantic tier palette does the rest. The preview in the labs repo strips out the backend so you can open the whole design system in a browser and read it like a notebook.

---

Published by amplified creations, an independent technology studio in Leiria & Lisboa, Portugal. Contact: info@amplifiedcreations.com · More: https://amplifiedcreations.com/llms.txt
