My AI Journey: From Prompting to Agents

This post will talk about my journey with AI as someone who experienced the rise of ChatGPT and other fields of AI in the past few years. I want to talk about how this blog could be integrated with a new personal project I will explore about Agentic AI. 

This blog has always been about documenting my journey, both in finance and in life. This project is the next step in my technical journey, and I’m excited to share all about it. Do read on if you are someone interested in the capabilities of AI and Machine Learning, through the eyes of this young person who watched it all unfold.

The Spark (2022)

I remember the "magic" of 2022, when models like ChatGPT and Stable Diffusion were new and hot. These models burst into the scene and made it felt like a new era of technology had arrived overnight. 
 
I was, and still am, a user of this new "magic". I spent hours on interest groups online, just trying to keep up with the new developments and use cases in image generation. 
 
But as a computer science student, "magic" isn't a good enough answer. I do need to understand and be able to explain, "How does this actually work?" 
 

Exploration of AI (2023 - 2025)  

I started university in 2023, and moved on from simply playing with AI as a user, to being interested in applying AI in different fields. 
  1. Automatic Speech Recognition (ASR), using models like OpenAI's Whisper to achieve fast and accurate inference when I participated in a hackathon
  2. Computer Vision (CV), using models like YOLO for CV tasks like object detection in my credit-bearing internship.
  3. Cloud Platforms, specifically Google Cloud Platform and learnt more about its capabilities in building and deploying AI solutions for a wide variety of tasks. My first exposure was using Google Colab to run inference and training for Stable Diffusion models.
  4. Machine Learning Fundamentals, learnt about neural networks (NNs) and the math behind them, and the common NN architectures.

Building With AI (Now)

Now, I am at the start of the next phase, which is to build with AI. I am building not just models, but agents. 
 
What's the difference? A model is a simple tool, passively waiting for you to give it a prompt, before giving you a response. An agent is active. You give it a goal, and it can reason, plan, and take multiple steps to achieve that goal on its own. I am using the  "Agentic Design Patterns" from Antonio Gulli as my reference for learning.
 
My new personal project is to build an "Agentic Blogger" system to help me with this very blog. My idea is to create a multi-agent system that:
  1. Scouts the internet for "signals", like business news or built-up public sentiments about the market and economy.
  2. Filters out the noise and identifies topics that can be turned into relevant blog posts.
  3. Drafts a post that analyzes the "signal" with our Dividend Investing mindset and breaks down relevant financial concepts for a broad audience to understand.

Conclusion

This is a long-term learning project which sits perfectly at the intersection of two things I am passionate about. This isn't about replacing my own voice; it's about automating as much "grunt work" as possible so I can focus on the job of analysis. 
 
In my journey so far as a student, I have been learning and experiencing without building anything of my own. This is something I want to develop patiently from the ground up, and by the end of it, I can say that I built a reliable and systematic tool of my own using cutting edge AI.  
 
Disclaimer: This blog post is for entertainment and educational purposes only. It does not constitute financial advice. Please do your own due diligence. 

Comments

Post a Comment

Popular posts from this blog

2025 Portfolio Review and Why I Don't Limit The Number Of Stocks Owned

2025 is in the books with only 2 and a half trading days left.  2025 is the year I initiated my dividend portfolio and embarked on this hopefully decades-long journey with dividend investing. My portfolio was steady and already printed cash like it was supposed to. In 2025 my  Net Assets  increased by 45%, of which around two thirds can be attributed to personal savings and one third can be attributed to investment returns (capital gains + dividends). Here is a breakdown of the number of dividends I can expect to receive by month next year: Jan: None Feb (3): Kimly, SGX, Suntec Mar (6): Keppel DC REIT, AIMS APAC REIT, PLife, CICT, MIT, MLT Apr (1): DBS May (7): DBS, OCBC, UOB, CLI, CDG, SGX, Suntec Jun (3): AIMS APAC REIT, MIT, MLT Jul (1): Kimly Aug (6): DBS, OCBC, UOB, CICT, CDG, Suntec Sep (5): Keppel DC REIT, AIMS APAC REIT, MIT, MLT, PLife Oct (1): SGX Nov (3): DBS, SGX, Suntec, Dec (3): AIMS APAC REIT, MIT, MLT I currently hold 15 individual counters, giving me...
Read more

Decoding DBS's FY2025 Earnings Call

Image
 This article was written with reference to DBS Media Briefing Transcript and Analyst Call Transcript for FY2025, which can be found at https://www.dbs.com/investors/financials/quarterly-financials During the DBS Full Year 2025 earnings call, CEO Tan Su Shan described the year using two specific words: A "Perfect Storm." She is entirely correct about the macro environment that our Singapore banks had to operate in. Benchmark interest rates (SORA and HIBOR) crashed by almost 2 percentage points. The Singapore Dollar remained brutally strong, hurting the translation of foreign earnings. Finally, the new 15% global minimum tax kicked in, slapping the bank with an additional SGD 400 million in tax expenses. DBS headline net profit fell 3% to SGD 11 billion. But if  you look at the actual operating engine, pre-tax profit rose to a record high of $13.1 billion . Total income grew 3% to a record $22.9 billion .  Deep Deep Deposits When interest rates drop, the textbook resp...
Read more

Decoding UOB’s FY2025 Earnings Call

Image
Writing this with reference to the Earnings Call Transcript posted here: https://finance.yahoo.com/quote/UOVEY/earnings/UOVEY-H2-2025-earnings_call-415512.html UOB just released its Full Year 2025 results, reporting a net profit of SGD 4.7 billion . Operating profit saw a slight 4% dip to SGD 7.7 billion . Here is my breakdown of the latest earnings call, and why UOB remains a core position in my portfolio Falling Rates But Rising Fee Income The market priced in severe NIM compression this year, but UOB's margins have proven stickier than anticipated. While full-year NIM settled at 1.89% , the downward trajectory has slowed. In fact, Q4 NIM ticked up to 1.84% (from 1.82% in Q3) , and management noted that as of end-January 2026, the exit NIM remains stable at 1.82% . This resilience is a direct result of active funding cost management . UOB successfully drove double-digit CASA (Current Account Savings Account) growth in its wholesale banking division . This influx of cheap deposit...
Read more