The Dawn of Agentic AI: A Personal Journey

In my current place of work I've tried to engage strongly with AI whether it be taking courses on prompt engineering, GitHub Copilot (code/chat/cli - in all forms) or inventing my own wrapper around OpenAI's open source Whisper voice-to-text model and combining with Small Language Models locally.

I've tried to push the boundaries, however this week is different. This week marks a significant milestone in my professional journey — one that could very well be a pivotal point in the history of how we work. I've had the privilege of completing a free online course on agentic AI from Andrew Ng's deeplearning.ai in collaboration with crewai.com, and it's not an exaggeration to say that it has been a transformative experience.

Agentic AI, a new term that may still sound foreign to some, is poised to redefine the landscape of programming and business operations. Unlike procedural programming — a familiar territory where I've spent many years — agentic AI introduces a paradigm shift towards more dynamic, autonomous workflows. It's a leap from the traditional, step-by-step approach to one that is more intuitive and self-directed.

The course itself was absolutely both shocking and eye-opening, unraveling the intricacies of agent-based systems - something new to me. These systems are designed to act by themselves, interacting with other AI agents to achieve a common goal. This is a stark contrast to the rigid sequences of procedural programming, which require explicit instructions for every scenario. Starting with the simplest use case of a blog post writer agent interacting with an editor correcting its mistakes, this course was very different from traditional "prompt engineering". Taken to its extreme, being able to use tools like internet search, web scraping, across different AI agents, who interact - for example to customise a initial cold email to a potential sales contact after researching them (automatically), is something I've just simply never seen before.

In the past I was to be excited by the newest technology on offer by Microsoft, and the challenge was to learn it and use it. Now, the challenge is to determine how to find scenarios to solve using agent like AIs - with so many to choose from it's almost analysis paralysis!

What's truly exciting is the potential agentic AI holds for businesses. It promises a future where companies can use traditional programming methodologies along with heavy-duty research-like agents. Honestly, it's a skill set that, I believe, will soon become a staple in the toolkit of every forward-thinking business.

I'm not going to lie to you. I was in two minds as to whether to post this. On one hand, this is literally the future that I've witnessed in front of me, and there's a certain advantage to being ahead of the curve, but on the other hand, I want to mark this week for historical purposes as it’s when I personally discovered agentic AI. In my opinion, this is honestly truly transformational, and before you know it, we will all be talking about this as a staple for all businesses worldwide.

For me, it's clear that I'm witnessing a game changer. The way we approach problems, develop solutions, and conduct business is about to transform. And when we look back years from now, I will likely see this moment as the turning point that set us on a path to a more intelligent, autonomous world...

Enrol for free here at DeepLearning.AI: https://www.deeplearning.ai/short-courses/multi-ai-agent-systems-with-crewai/