LLM

Some time ago, I felt specifically that the AI workflow in my app was not working as expected. This happened several times even with the same workflow I had done previously. I didn’t change anything, so my assumption is that the capabilities of the OpenAI gpt-4.1-nano model have changed. What happened to me was that the output from the gpt-4.1-nano LLM model often produced results that did not meet expectations. It gave the wrong time output, which is a major concern specifically for my application. Therefore, for the time being, I moved to gpt-5-nano because it is still relatively cheap, it worked, it is smarter, but a new problem appeared. ...

There are many expense tracking apps out there, but my main concern now is how to make expense tracking less manual—no need to fill in lots of data like item name, price, category, and so on. But before that, there’s an experience that became the reason why expense tracking is necessary: when I was abroad. I live in Indonesia, where the cost of living is relatively cheaper compared to other countries. The currency difference made it difficult for me to compare how much I had spent so far. ...

When we are working on a project, sometimes we forget some commands in the terminal, such as how to install packages, use options, or filter the use of the grep command for example. Sometimes it’s a pain, making us need to google, look at script documentation, or ask chatGPT if we want an instant. Even though what we are looking for is only a short terminal command just because we forgot the command. ...

Some time ago, I was working on an e-learning related project. Simply put, teachers can create questions and students can answer them there. I made the system like Google Forms. After the teacher creates the questions, there will be a link that students can access to answer the questions. During the creation process, I thought, why don’t I add AI to this system? I know… not all software needs AI, it depends on the needs. Here I will discuss the creation process, from ideation to implementation. ...