Tools and their learning curve
I’ve been learning more about AI since early January 2025. A year later, I see how much it’s an ongoing process.
The discovery phase
I didn’t pay much attention to AI before that and it’s only after a colleague (shoutout to Jacques) that I started to be more intrigued by the options it could offer me. Sure, I used ChatGPT and was impressed by Image generation, but at this point it was a bit more impressive.
What got me more into it was the possibility of having a thinking buddy I could chat with for “rubber duck” sessions and think more deeply about problems and figure out solutions. That’s when I started tinkering with Perplexity and Claude.
I found value in each of them, but I wasn’t really using those as much as I could’ve at this point.
As I was working with them more and more, I started thinking about building new tools. I didn’t know it at that time, but it was the start of a very big journey for me.
The honeymoon phase
Around February 2025, I decided to start building a new tool for work to help out. The tool itself is irrelevant here, the growth from the process is the important key.
As I saw that Claude was pretty good at discussions, I started chatting about a tool that could do XYZ. I was then bombarded with ideas and options on how this could work. I was amazed by it: it was fast and the building blocks were clear. Although I didn’t know how to integrate those parts together yet and I wasn’t clear on how they actually worked, I started building.
I found myself digging deeper and deeper in the code examples and outputs. In order to fully understand what I was provided, I decided to type everything I was presented instead of the usual copy-paste. This proved to be very helpful as I discovered some bugs here and there and saw my various skill gaps. This was amazing… until it wasn’t. I thought to myself: “Oh man, I’ll be done in like 3 weeks at this pace!”.
Oh, was I wrong!
The reality check phase
The more I asked questions and realized this or that just didn’t work, I found out more about context window size and how hallucinations can just break the whole flow and mess up what you thought was just perfectly organized data.
Don’t get me wrong, this is great info and awesome to learn more about it, but at the time, it felt like a giant waste of time. I then realized I was slowly starting to slip into the “do that for me” instead of “work with me on this” mentality. Using AI wasn’t about figuring things out anymore and it was more about getting things done.
Finalizing projects is great, but it shouldn’t be the only outcome here: we should grow as well through it like any other learning we do throughout our lives.
The final outcome
It ended up taking me a year to finish that “little” project by working on it part-time throughout that period. I learned a LOT about AI itself, backend and frontend too.
So, it was a great adventure filled with ups and downs, a lot of trials and errors, and an overall satisfaction of having learned more about “how” to use a tool “for” me instead of letting it use me.
Tools are not our enemies if we learn how to use them and that’s the reason we always have to keep learning and grow.