Meals live, data layers next
Personal Health OS
One place for what I eat, how I sleep, how I train and what my bloodwork says, starting with the question I ask most: what should I eat?
Spicy tofu Mexican bowl
Lunch · 20 min · uses the rice and black beans you have
- Protein
- 30g
- Carbs
- 60g
- Fat
- 18g
7h 48m
Score 86 · last 7 nights
HRV above your baseline
9,412 steps
Two lifts, one run this week
ApoB 72 mg/dL
In range · retest in March
- Role
- Product, engineering
- Live
- WhatToEat at eat.ravishah.work
- Inputs
- Pantry, taste, goals, a daily check-in
- Next
- Oura, Eight Sleep, Apple Health, labs
- Built with
- TypeScript, Next.js, Supabase, Structured LLM output
It starts with food because food is the decision I make most often and think about least. WhatToEat takes what’s in the pantry, what I like, what I’m aiming for and a quick check-in, then suggests something I’d actually make — a spicy tofu Mexican bowl, a paneer tikka wrap, lentil dal with rice.
The rest of my health data lives in separate apps: sleep in Eight Sleep, readiness in Oura, activity in Apple Health, bloodwork in Superpower. The plan is to bring them into one model so each answer knows about the others. A bad night of sleep should change lunch. A lab result should change the grocery list.
The goal is fewer decisions, not more charts.