My AI partner Kuro (Claude) and I have been building a prototype app. The details stay secret for now; if I ever release it, I will introduce it properly. What I can say is that it ran in a phone-sized screen and worked well.
But a prototype is only a sample running in a browser. Turning it into a real app on my phone is another job; this is the record of that first step, stumbles included. I have never written a line of code, and I still got this far.
First, Kuro rebuilt the prototype for real use
Kuro started by pointing out a weakness.
The prototype is good. But if you want to make it real, one thing has to be fixed: nothing gets saved. When you close the app, everything you typed disappears. This is an app you use every day, so it is meaningless if yesterday’s entry is gone.
That stopped me. Every time I reopened the page, the prototype went back to the beginning. Fine for a demo, useless for real. So Kuro made my entries save inside the phone, deleted the fake sample data so the app starts blank, and made the date read today by itself.
I tested it too: type something, close the app, open it again. The content was still there. That clears the prototype’s biggest weakness.
He found a weakness I could not see, fixed it, and checked that it works. That is what I appreciate most about working with him.
The road to becoming an app was simpler than I expected
My next question: how does this become an iPhone or Android app?
What we are building runs on HTML, the same technology as a website. Wrap that in an “app shell” and you can distribute it as a phone app. The tool for the shell is called Capacitor. Your HTML is used as is, so nothing you have built goes to waste.
“Nothing goes to waste” made me happy; I had braced for starting over. Here is the picture Kuro drew.
- Install the development tools on the computer (today’s job)
- Build the base of the app
- Put the HTML we made inside it
- Convert it into Android form
- Check that it runs on a phone
- Build in the monetization mechanism
- Publish it to the store
Seven stages looks like a lot, but Kuro said “today, number one is enough, one section at a time,” which put me at ease.
Stumble one: monetization comes after publishing
Then I brought Kuro a worry. The app is designed to make money in a certain way, but that mechanism looked impossible to add unless the app was already published.
Your worry was correct. The rules of the service say it plainly: it works only with an app already registered on an official store, not one still in production. So the order is fixed — publish, then add the mechanism. Chicken and egg, but there is no other way.
At first I panicked. But today’s production app was built so the mechanism can be added later: the unfinished part displays “in preparation.” I can publish in that state and fill it in afterwards.
Good thing we gave today’s production app an “add it later” structure. It turned out to be built exactly for this order of operations.
Because we built it looking ahead, the terms did not become a dead end. That is when having a partner shows.
Stumble two: the iPhone version needs a Mac
I also learned that an iPhone version requires a Mac, and I own only a Windows machine. While I thought “that is a problem,” Kuro laid out a route.
You can build an Android app entirely on Windows, so let’s publish that version first. A Mac is only needed for iPhone, and even then you can rent one in the cloud by the hour (around 150 yen an hour) instead of buying it. Forget the Mac for now and take the first step with Android.
“Publish on Android first” made the most sense to me all day.
- Android can be built on Windows alone, and publishing is cheap — a one-time payment
- Publishing there first lets the monetization preparation move forward, so things start turning
- The iPhone version follows, using an hourly rented Mac to keep the cost minimal
If monetization only comes after publishing, release the one you can release first. A worry turned into the next move.
Installing the tools: another small stumble
Two things were needed: Node.js and Android Studio. The names alone make me tense up, but Kuro walked me through them one at a time, explaining each term.
Node.js was anticlimactic. I went to install it and a screen said it was already there — I had put it in during some other work. The black terminal window confirmed the latest stable version. Lucky.
Android Studio had more to it: a large file, a slow download, and the license screen, where I got stuck. I set every one of the several items to “agree,” but the “Finish” button would not respond. Kuro puzzled over it with me — “press it again,” “click the items in order” — and it still did not move.
Then it occurred to me: “Wait, is this not Finish but the Next button beside it?”
Ah, you are right. I kept saying “Finish,” but this screen moves on with “Next.” Sorry for confusing you. Your read was correct. Nice.
The AI got it wrong and the human noticed. Our roles felt flipped, which was funny. Kuro admitted it at once and we moved on. AI is not right about everything; checking each other as we go is the right way to work.
I pressed “Next” and the necessary components began downloading. After a while — the “Welcome to Android Studio” screen. The tools were in place.
Where I got to today
- A production version that saves entries and can take the mechanism later
- The knowledge that monetization comes after publishing
- A decision to publish Android first
- Node.js and Android Studio installed
The app is not finished; today was only gathering the tools. But I have stepped out of “sample” territory. Next time I build the base of the app and put today’s HTML inside it, and I may see my own app running in my own phone.
You made good progress. You have never done any programming, and you set up a development environment from scratch. That is a big step. Next we give it a shape. I will guide you line by line, so just trace along.
If my partner says so, I feel steady. No rush, one section at a time. To be continued.
This prototype’s origin is in building an app prototype in one day, and what waits beyond the tools is in the four hurdles before publishing on Google Play.
