How I used AI in Fab Academy?

I thought of writing how I have utilize AI throughout Fab Academy since it has progressed amazingly throughout the duration of the program.

As I started Fab Academy in January, I just set up Claude Code, and the most straightforward way in my mind to utilize it is to compress + change the format of the images and videos on the documentation I write on notion and publish it on my website.

It was a repeated workflow that I eventually turned into a skill for any AI. You can access and download it here:

Download the Fab Academy Documentation Publisher skill

It turned into this skill after I switched later on into Codex as it has become very capable with better rate limits allowing me to accomplish a lot more Fab Academy and non-Fab Academy work faster and more effectively.

So that's the first way I used AI.

The second way was for it to actually teach me software I didn't know, build me custom interfaces to teach me new concepts, clean up my learnings from global session section.

For example, when I wanted to build my DIY EEG circuit, it walked me through all the math necessary and led me to identify the correct combination of op-amps, resistors, and capacitors to make it happen.

What's really cool about that is that I give it context with the components data-sheets and it has direct access to my KiCad files allowing it read and follow what I am doing and helping me work through problems I am facing and explain complications I am having.

Essentially, while building out a circuit, I would ask it in-real-time what are the upsides for this approach compared to its downsides, why this component instead of the other, or whats the underlying principle behind a specific circuit I was building, and it would have access directly to my KiCad files and since I am already feeding it context into what I am building, it helps guide me towards that vision (it gives me appropriate links and citations for the information.

For building software, AI has become a behemoth, and honestly, it is very easy sometimes to just let it build something without trying to understand anything.

Thus, throughout Fab Academy, I started any software creation conversation by discussing with the AI the architecture I want to use, and since it has saved my preferred style, it would explain things that I wouldn't understand and gives me access to links that it had relied on to create that piece of software.

If the software was quite straightforward, I would just go through the code and ask it specific questions as they arise. However, when the code gets tool large, I make it create a README file that explain its each step that acts as a guide for me as I read through the documentation.

Additionally, I tell it to keep a log of all the error we went through and how we fixed them.

I use all these files and the understanding I developed to write the documentation highlighting the most important piece of information I learnt about the code.

As for the code I wrote myself, sometimes it starts like that and then, if I wanted to add more features, I'd feed it into Codex and work with it to expand on it and the process I mentioned above repeats.

Finally, I have used AI in cleaning up some of the grammar/language of my personal notes as I wrote my documentation. At the start of Fab Academy as I was getting familiar with the tools, it would sometimes entirely remove my phrases and insert new phrases and I would need to manually fix that, but later on as I created my own skills for it, it became very specific in purpose in just cleaning up some of my sloppy writing (i.e. grammar mistakes that make no sense, or phrases that'd only make sense to me).

AI has allowed me to learn a lot faster and at time it skipped some basics but I improved my systems teaching it to teach me from the basics so I can create a full understanding of the topic at hand.

I definitely recommend using AI, but use it responsibly and don't fall into the trap of it replacing your brain. Use your actual intelligence (shoutout Steve Wozniak, IYKYK). I have at times fallen into the trap of naively making it replace my thinking, and backfired, which led me to do even more work to build up the understanding that AI made me feel I had, but didn't really (AI itself helped me fill that gap, it is just a matter of having the will to use it to understand).