So that was the idea. It came while I was kind of about to finish round 2 of my trading bot. During my journey to create a trading bot a few years back, I also created a very primitive web based frond end to support my manual trading activities. This time round I thought: let’s give it a try and make this “idea” more commonly available. Although my GUI is publicly available already. I guess even a few people tried to install it, but it’s still very technical. Requires installation of various files, framework etc.
So I thought: An iOS App might make this journey way easier.
And here we go.
But how to start? I had put some time already into XCode, macOS, programming years back, before Swift came along. At that time I merely made a few prototypes. Cocoa programming was too complicated. Not really the language itself, more the knowledge about the frameworks etc.
Incidentally I was chatting with chatGPT about some issues with my trading bot. So I asked him, it’s a him for me: How much time I’d need to invest to learn to program an iOS App. He thought in maybe 2-3 months I could have an early version ready maybe. 2-3 months??
Okay. I downloaded the latest version of Xcode and went through some very easy tutorials. And, don’t ask me how, I figured out Apple Intelligence in Xcode. And let me tell you: It is fantastic!
You can tell the AI what you are looking for and it does it for you, right away. Unbelievable. Less than a week after my start I have 60% of my app ready and working:
- Spashscreen (although this I made without help of AI)
- WebSocketClient (connecting with Bitstamp websocket)
- BitstampClient (implementing Bitstamp REST API)
- App Settings (Bitstamp Accounts API key&secrets, watched markets, and more)
- LiveView (Live trades, summary of every minutes: amount of trades/ vWAP/Nominal Volume)
What still is open:
- Orders View (possibility to track and change open orders, the actual heart of the app)
- Trades View (possibility to track recent trades, outcomes, change of wealth etc.)
- Balances View (very easy, just 2-3 calls to Bitstamp API and displaying the results)
So these Bulletpoints above I’ll track in their own posts in the coming weeks.
AI saved me weeks of work, but it was not always foolproof. Eg. the WebsocketClient AI created crashed every now and then, eventually it would crash after the first few incoming messages. So I did my own research and found a cool framework. Implemented it and since then the Websocket client ist 100% stable.
There were 2-3 other cases where I needed to tell AI how to improve. Some errors he could not figure out how to fix.
Eg. initially I had active markets combined with Bitstamp Accounts (API key&secret). For me it’s only one crypto I’m interested in. So that fits. Eventually I figured: I want the accounts separated from the markets. And telling AI to implement he totally fucked up. about 50 errors throughout the project. Stuff he didn’t change etc. So I implemented the separation of accounts and markets myself, fixed all the files around that and got it working. BUT: I could not have done all this without the first implementation from AI. Figuring out all by myself: it had taken me weeks, not hours.