Stop and give time to thinking
After I made the first version of Second Thought, I started the project page. This was a good opportunity to stop and improve the architecture, plan the next steps and functionality, and test what I did. This gave me time to think about how I worked and how to improve my working efficiency.
Quality over features
I like to create new functions. During development I always have a lot of ideas about how to improve the product. It is enjoyable and interesting. But more features don’t mean a better product.
When I started working on error handling, improved the tests or made the app more stable, I realized that I like to work on quality. It brings problems that I should solve. I have to think about the next steps connected to the project and how to handle them. I should also know that I don’t know everything, and the code shouldn’t be prepared for everything. I mean, if I know what functionality I want and I plan it, I will have knowledge about what I want. To me, this means that during development this knowledge helps me make better trade-offs. I know what I want, and I will prepare the code for that ahead of time. I also know what I don’t want or haven’t planned yet. So during development I will know that the code shouldn’t be prepared for unknowns. For example, I will know that I don’t have to make an interface for multiple implementations if only one is planned.
Work with AI trade-offs
I used Claude for the project. Here is my trade-off. It relates to how I used it while I made the first version.
Advantages:
- Syntax: I didn’t know TypeScript, and I didn’t have to learn it.
- Explanations: I got a lot of explanations when I needed them.
- Coding speed: The AI wrote the code and tests fast.
- Decisions: I had time to make decisions about what to do next and how to do it.
Disadvantages:
- Speed: Sometimes it was so fast that I didn’t want to stop and understand what happened (but I did). It is an easy trap to become lazy.
- Easy to use: Prompting what I want is easy. Sometimes I just wanted to leave it to the AI to do it the way it wanted.
- Syntax: I don’t really understand TypeScript.
- Missed coding: I like coding and improving the code. If the AI does it, I can’t enjoy it.
Consequences for myself:
- AI helps, but it is important how I use it.
- If I want quality, I can achieve it, but I can do it in a different way.
- AI usage improvement: I learned that it is important to know where I should put myself in development. AI can do a lot of things, and I have to know where my decisions help the most during the project.
- Doubts & questions: AI makes mistakes, as we do, but it makes them confidently. It is easy to just accept them. But it is useful to stop and ask questions, check things from different sources or ask for sources.
Architecture before implementation
After I stopped to make the project site, I had time to understand the architecture. Unfortunately, at first I only planned the functionality and then started development. With this method I didn’t have time to step back and see the bigger picture. I made a drawing of the architecture. While I was drawing it, I had time to think about it. This led me to plan the next architecture version of the app. It also revealed some weaknesses and possibilities. I like architectures, and I decided that I will plan them before implementing a project and stop to improve them from time to time before continuing.