There are moments in a career when you feel the ground shifting beneath your feet, not with instability, but with the rumble of an approaching wave of innovation. For me, that wave is Agentic AI. My journey into this fascinating world has been a whirlwind of courses, articles, and late-night coding sessions, and I wanted to take a moment to document the path I’m carving, the tools I’m using, and the incredible things I’m learning.
A Rich Ecosystem of Learning Resources
My dive into AI has been powered by a fantastic blend of personal curiosity and invaluable support from my employer. On my own time, I’ve been engrossed in Ed Donner’s course on Agentic AI on Udemy, which has been a cornerstone of my learning. It was here I was introduced to an Anthropic article on “Building Effective Agents” that crystallized a core concept for me: simplicity is key. The advice to avoid unnecessary complexity and favor direct API connections over frameworks for better debugging was a game-changer.
This personal learning is powerfully augmented by my employer’s commitment to upskilling. With access to LinkedIn Learning, a vast library of O’Reilly books, and most notably, an enterprise license for Google Gemini, the field of possibilities has opened up dramatically. It’s a powerful statement when a company invests in its employees’ growth, and I firmly believe in seizing these opportunities.
From Core Principles to Practical Application
It’s one thing to read about AI, but it’s another to get your hands dirty. Replaying the foundation courses and walking through the ipynb notebooks from Ed Donner’s course helped solidify workflow patterns, especially the concept of orchestrating multiple LLMs and having one act as a “judge” to evaluate the others’ responses.
The real excitement, however, comes from applying these principles. I’ve already taken my first step by creating a custom agent in Gemini to help with my blogging activities! But the ambition doesn’t stop there. I’m actively planning to use a similar agent-based approach to evaluate MoB responses from our BPOs, a key task in my professional role.
Inspired by a lab in the course, I also built a career conversation bot. This project sparked another idea: reviving my personal website to serve as a professional and personal profile. I’ve successfully set up a subdomain and a WordPress instance, and even deployed a Hugging Face space on it. The next challenge is to figure out how to augment the generation, a topic I’m eager to explore further.
The Personal Side of Learning
I won’t pretend this journey is an easy one. As Ed Donner wisely mentioned at the start of his course, patience is paramount. There are moments when the concepts don’t click immediately, but the thrill of executing code and seeing it work is a powerful motivator. The ability to follow along, experiment, and even fail is where the real learning happens. I find myself constantly wanting to go back and review the material, but I’m pushing myself to complete the course first to get that all-important initial overview. It’s a continuous cycle of learning, applying, and revisiting, and frankly, it feels incredibly exciting.
My Evolving Tech Stack
For those interested in the nuts and bolts, here’s a snapshot of the tools I’m currently using:
| Environment | Technology |
|---|---|
| At Home | IDEs: Cursor, VS CodePackage Management: uv |
| At Work | AI Platform: Google Gemini EnterpriseLearning: Google Skills Platform |
The Journey Continues
The world of Agentic AI is vast, and there is so much more to learn. While I’m not under any illusion that mastery will happen overnight, I am encouraged, engaged, and determined to stick with it. Every small success, from a working script to a deployed agent, fuels the passion to keep going. I’m excited to see where this path leads and what I’ll build next.
Leave a Reply