ZOVO.life

Hero screen — AI assistant greets the user and offers “Start Assessment” or sign-in.
Chat & navigation — profile drawer, WhatsApp-style chat, and one-tap progress panel.
12-question onboarding quiz captures metrics, goals, allergies, and age for personalisation.
Core flow — photo analysis returns calories/macros and updates the goal dashboard in seconds.

Inspiration

Many nutrition buffs still pay human coaches in WhatsApp chats:
snap a meal photo → wait hours → get macros and costly advice.
I’d done it myself (and so had my friends) until the bill hurt.
So I asked: “What if AI could answer instantly, for almost zero cost?”
That pain point sparked ZOVO.life — a WhatsApp-style chat where an AI coach delivers real-time calorie tracking and personalised guidance.

What it does

Chat like WhatsApp — I type or drop a meal photo.
Instant analysis — GPT-4o Vision estimates calories & macros.
Personal coaching — The AI factors a 12-question onboarding quiz and my meal history to send tailored tips for the day.
Progress dashboard — I see daily / weekly macro targets and streaks.

Result: the value of a human coach, 24 / 7, with zero overwhelm.

How I built it

Layer	Stack & Tools
Code scaffold	Bolt.new one-click template → Expo / React Native
Front-end	Expo Router, TypeScript, NativeWind (Tailwind-like)
Back-end	Supabase (Postgres, Auth, Edge Functions, Storage)
AI layer	OpenAI GPT-4o Vision endpoint for meal photos
Hosting	Netlify for the web build (https://zovo.life)
E-mail	Resend for password reset & onboarding drip
DevOps	GitHub Actions → Netlify CI/CD, Supabase migrations

Challenges I ran into

Zero dev background. My last coding was 15 years ago (plain HTML/CSS). React Native, Supabase—even Git—were all brand-new.
Five complete restarts. I deleted the repo and re-scaffolded Bolt.new five times because things kept breaking faster than I could fix them.
Architecture fog. Screens, services, Edge Functions—everything was a black box; I peppered Bolt with hundreds of micro-questions to inch forward.
Web build nightmare. Mobile ran in Expo Go, but the browser build refused to load image-upload functions; hours of CORS-header detective work finally solved it.

Accomplishments I’m proud of

Working full-stack MVP in a handful of late-night sessions, from blank screen to live site at zovo.life.
Solved every blocker solo, using Bolt for ~95 % of prompts and a pinch of ChatGPT near the end.
First real users—friends who used to pay human WhatsApp coaches—now log meals in ZOVO instead and love the instant feedback.

What I learned

Clear Bolt prompts = speed. Short, precise requests beat hours on forums.
Modern stack mental model. Expo ➜ Router ➜ React components ➜ Supabase Edge ➜ Netlify CDN finally “clicked.”
Guardrails matter. Pre-deploy env checks and strict TypeScript types save days of silent failure.
Biggest insight: if a marketing guy can ship AI photo analysis in React Native, thousands of non-coders can build their ideas too.