You can't learn AI by passively watching videos or reading reports.
I was fortunate to go through a Gen AI automation course led by the Heicoders Academy's co-founder and CTO (Wei Chen Beh). Truth be told, the hands-on experience completely changed my perspective; it forced me to move beyond theory and confront the real complexity behind what often looks deceptively simple.
Designing an automation workflow is far from straightforward. It requires careful consideration of the use cases, practical constraints, AI-related risks, and — most importantly — whether there is a tangible outcome that justifies the effort. These are not things that I would fully appreciate until I start building and experimenting something for myself.
And oh gosh, then there's the debugging, which can be painful at times. But that's precisely where my learning happens after going through these iteration and refinement.
One of the more surprising lessons for me was how quickly "free credits" get consumed as I include more advanced requirements.
For example, as part of the hands-on exercise, I used Manus AI to build a website for a Hong Kong IPO evaluator.
The problem I set out to solve was simple in my mind: how can we better assess IPO attractiveness in a structured and scalable way? I chose HK as my country scope.
However, I quickly realized I had been overly ambitious in trying to incorporate a dynamic evaluation model through the Yahoo Finance API call. My heart was definitely pounding as I saw my "free credits" get used up in minutes!
It's a good reminder that while AI agents can accelerate development significantly, they are not without cost.
Nonetheless, the entire experience using a tool like Manus and the automation course was incredibly rewarding. It reinforced a key lesson for me: the importance of rolling up my sleeves and actively experimenting with automation workflows.
P/S: A big shout out to my colleagues (James W. & Keryn Oh) who also joined me to have fun with these AI lessons. And thank you Armand Izzaq for dropping by as well!
You can't learn AI by passively watching videos or reading reports.
