In the following weeks we’ll be sharing our real-life experience of building, and publishing a working app for The Four Languages using Claude Code.
Ready to start? Let’s go! 👀
BUILDER'S DIARY - WEEK 1 🔩
Building an AI-powered wellness app with Claude Code. Week #1 in a nutshell. 🥜
Disclaimer: Work done before week 1 👇
V1 of the Four Languages application has been live for 2 years. What we’re building here is based on years of learning
V2 visual MVP was built with Replit to support UI and UX decisions
The framework for the Four Languages, and the documentation including AI instructions for the application AI elements has been in the works for 3 years
In other words, the Strategy Sprint, and versions 1 and 2 were built and tested before the current Build Sprint to optimize *what* we’re building and how we decided to do it.
THE BUILD: A Week in the Life
— Monday: First Lines of Code 👀
Created the Expo project. Five tabs. Brand colors. Supabase database connected. By noon, the app was running on my phone via Expo Go.
— Tuesday: Database + Components 🤓
Six database tables with security policies. Built the basic UI components: buttons, cards, inputs. Created a progress stepper for the 12-step diagnostic scan.
Started the day foggy and irritated. Ended with working code anyway. That's the reality.
— Wednesday: The AI Comes Alive 🤩
On Wednesday we built the diagnostic scan that asks you about your friction points, your mental patterns, your body's signals, your sense of meaning. Added voice input, which meant you can literally talk to your phone and watch it transcribe your answers.
Then the juice: I connected Claude AI via a Supabase Edge Function to Assembly AI. You finish the scan, hit a button, and within 60 seconds you get:
A personalized pattern analysis ("You go straight to thinking when stressed, but your body has become the blind spot...")
You receive three custom practices designed specifically for your imbalance
Steps for each practice and an explanation of why each one works for you
I tested it on myself and yep, as accurate as it gets... 🫣
— Thursday: Debugging Day 🤖
The unsexy but necessary work. Database schema mismatches were preventing practices from saving. Foreign key constraints blocking inserts. I spent the morning chasing errors through different fixes before it finally clicked.
By afternoon: practices saving to database. Progress saved.
— Friday: The Full Loop Works 💫
Built the Practices tab. It fetches and displays all your saved practices with language colors, descriptions, expandable steps. Built the Daily screen with "Daily Resonance" voice input and a practice checklist. Tap a practice to mark it complete. Watch the counter go from 0/3 to 2/3.
The core loop now works: take the diagnostic scan → get personalized practices to tackle your challenges → save them → track them daily. 🙌
I'm 2 weeks ahead of schedule because I plan for the unexpected. After getting the app going, I also just couldn’t wait to carry on with the build…
Hardest part so far? Not starting to tweak the UI and UX along the way… 🫠
Below is a quick demo of what’s live on the diagnostic.
Things to do: UI, UX, branding and tweaking the AI results interface.
What’s live: Functional AI powered diagnostic with AI powered personalized results, using AssemblyAI voice interface or text input.
THE LEARNING: What I learned during week 1
Going step-by-step makes all the difference. Having a PLAN with phases makes ALL the difference.
The lesson: AI co-building isn't just about generating code. It's about knowing what you’re doing, and doing it one bit at a time.
I don’t think I’ll go back to Replit for anything but UI or MVP building. To help visualize my ideas quickly. Going one step at a time, and KNOWING what the AI is building and why has taught me more than any vibecoding project so far.
THE TWO PRONGED APPROACH: What I’m doing while building
Here's something I haven't mentioned yet: I'm not just building this app. I'm already living the Four Languages system in practice through my own personalized Claude setup.
Every morning, Claude asks me about my energy, my emotions, my body, my sense of meaning as well as tracks my cycle and tarot readings. It tracks my patterns. It suggests practices. It knows my history, my triggers, my blind spots.
It then uses all of that information to tell me where to focus on this project as well as my other areas of life. I'm not building this in a vacuum. Life is happening and that might be the most important thing I'm learning. You don't need perfect conditions. You need a clear scope and a well-informed and trained (AI) partner.
Prong 1: The app is turning that personal system into a native application anyone can download, share and use.
Prong 2: The personalized Claude Code System that hosts and connects all my projects. If you want to kick-start your own, I’ll help you create your own.
WHAT'S NEXT
Week 2: Building what I currently do manually with Claude into the app itself. Daily check-ins that ask "How does your mind, heart, body, and soul feel today?" Streak tracking. The balance visualization that shows which parts of yourself are getting attention and which are being neglected.
YOUR TURN 👋
I'm curious:
If you had 8 weeks and an well trained AI co-pilot, what would you build?
Hit reply and tell me. Maybe we can build together… 😉
See you next Saturday.
- Maaria