Switching Gears To AI Agents
Now I know what part of AI I want to focus on. Read on to learn more.
Finding a path forward
I started the year trying to keep my feet on the ground after throwing in the towel on my previous company EnVsion. I quit EnVsion, but I was not going to stop building another company. I just didn’t know what I would be working on specifically, but I sensed that it would be centred around AI.
In the meantime, I tried to reinstate some balance into my life. One of the areas I prioritised was my health. I’ve been going to the gym 2-3 days per week since the start of 2004. I feel the strongest in years. I also dedicated more time to family life. For instance I now partake in more Saturday morning activities with my daughter.
I needed all this.
But, just like an explorer ready to embark on a quest to discover new wonders, it is now time for me to also resume this journey of growth and value creation. Lately, something in the AI space has piqued my interest, and I am ready to pull on this thread until I catch something that I want to run with.
This “something” is AI Agents.
Over the last 3 weeks, I’ve been diving deep into the world of AI Agents. It’s a very young space, a quasi-uncharted territory even among the tech savvy and AI enthusiasts. I found myself completely enthralled with the subject. After countless hours spent trying to read/watch anything on the subject, I am even more excited about AI Agents.
What are AI Agents ?
For those not familiar with AI Agents, I believe it is important we try to define this concept. To do so, we must define two concepts which are related: Agentic System and AI Agent. Here is my definition for each of those terms:
Agentic systems are software architectures that employ one or more AI agents to solve complex problems. These agents work collaboratively by breaking down the problem, executing sub-tasks, reviewing progress, and determining if additional steps are needed to achieve the original goal. Unlike traditional deterministic software, agentic systems create dynamic applications with human-like reasoning capabilities.
An AI Agent within an agentic system is a specific instance of a large language model assigned a particular role and set of responsibilities contributing to the overall goal.
For example, in a software development agentic system, there may be a Product Manager Agent, Developer Agent, and Tester Agent, each responsible for specific aspects of the software development process.
Key components of agentic systems include:
Agents: AI models with specific roles and responsibilities
Memory: Storage for context and data
Tools: Additional capabilities agents use to complete tasks (e.g. search Twitter)
Plan: Strategy for achieving the goal
Prompts/Inputs: Information provided to the system to initiate and guide its actions
AI Agents is the thing I am crazy about
AI Agents remind me of how I felt when I took the plunge into the world of autonomous vehicles 6-7 years ago. Unlike with self-driving cars, I want to stay in the world of AI agents for the long term and build real products with this technology. I want to commit to bringing forward this agentic future we are still trying to understand.
To further my learning and understanding of the space, I have begun developing my own Python AI Agent library. The library is still in its infancy and subject to breaking changes. In all honesty, I am building it for myself as part of my AI Agent curriculum, so I don’t expect it to become the standard or gain widespread adoption. But I’d certainly love to get feedback from some fellow AI Agents enthusiasts to try it out and provide constructive feedback.
Still exploring
I am still in “explore” mode though. This means that I am trying to understand the technology, its limitations, and where it is being used or could be applied. At the moment, I don’t know what specific business problem I would be tackling, although I have a couple of candidate areas.
I haven’t felt this exuberance in years. Yet I still find it challenging to dedicate enough time for my explorations of the world of AI Agents amid a demanding full time job and family responsibilities. But this enthusiasm adds more power to my natural resilience. This feels like play to me, so I can play in hard mode with a smile on my face.
It seems I finally see a path forward. Finally, I can start switching up from the first gear and accelerate towards the future I want to be part of.
If you’re also curious about AI agents, then check my Python library on Github.
You can also follow my daily posts on this topic on Twitter/X.
Forward!