Author: Clayton Pilat 

So, “Agentic AI” – I’ve seen this term pop up everywhere lately. At first, I thought it was just another tech industry buzzword? With the hype around LLMs and Generative AI, I’ve seen a lot of new (and made-up) buzzwords around LinkedIn. But the more I dug into it, the more I realised there’s actually some pretty cool stuff going on here. If you’re curious about what all the fuss is about, stick around. Let’s take a deep dive into Agentic AI and see what makes it tick.

So, What’s the Deal with Agentic AI?

Imagine an AI that doesn’t just sit around waiting for you to ask it questions. Instead, it’s proactive, makes decisions on its own, and even takes action to achieve its goals. That’s Agentic AI in a nutshell. It’s like upgrading from an Amazon Alexa that tells you the weather to a personal assistant that reminds you to grab an umbrella before you leave the house.

Agentic AI represents a pretty big leap forward in the world of artificial intelligence. Unlike traditional AI models that simply respond to inputs, these new systems are designed to:

  • Perceive their environment (kind of like having AI eyes and ears)
  • Make decisions autonomously (yep, they can think for themselves)
  • Take actions to influence their surroundings (they’re doers, not just thinkers)
  • Learn from the outcomes of these actions (which is something we could all work on)
Agentic overview : from Tarun Sharma
Agentic overview: from Tarun Sharma

The DNA of Agentic AI: What Makes These Systems Unique?

If we were to look at the DNA of Agentic AI, we’d find a unique genetic code that sets these systems apart from their basic LLM Chatbots. At the core of this DNA, you’d find one standout gene: goal-oriented behaviour. This isn’t just a feature; it’s the driving force behind everything these AI systems do. Imagine an AI that’s not just sitting there waiting for instructions but one that’s constantly thinking, “What do I need to do to achieve my objectives?” It’s like having a super-smart intern who doesn’t just complete tasks but actively looks for ways to meet and exceed the company’s goals.

But goal-oriented behaviour is just the start. The full genetic makeup of Agentic AI includes traits like adaptability (changing strategies on the fly), proactivity (taking initiative without being prompted), social skills (working well with humans and other AIs), and advanced reasoning (planning ahead and making complex decisions). It’s not just about having these traits; it’s about how they work together to create systems that are more autonomous, more capable, and more “intelligent” in a way that’s closer to how we think about human intelligence.

The real magic happens when you see these traits in action. Picture an Agentic AI personal assistant that proactively reorganises your schedule when it detects a conflict, adapts to your preferences over time, interacts with your contacts for scheduling, and makes smart decisions about which meetings are priorities. This combination of traits is what makes Agentic AI so exciting – and potentially game-changing. It’s not just about having smarter systems; it’s about having systems that can truly assist and augment human capabilities in meaningful ways.

Our Journey with Agentic AI: The Evolution of explAIn

Now, let’s talk about how we’ve been dipping our toes into the Agentic AI pool with our product, explAin. It’s been a long journey, and we’ve learnt a tonne along the way.

  1. The Simple Beginnings: explAIn started its life as a straightforward question-answering system. You’d ask it a question, and it would scour its knowledge base to give you the best answer it could find. It was pretty cool, but we knew it could be so much more.
  2. Adding Some Database Smarts: Next, we gave explAIn the ability to query SQL databases. This was like teaching our AI to read both books and live news feeds. Suddenly, it wasn’t just relying on pre-programmed info but could tap into dynamic, up-to-date data sources. This was a game-changer for providing real-time, accurate information.
  3. Learning to Do Stuff: The next big step was adding basic function calling. This is where things got really interesting. Suddenly, explAIn could not just know things, but do things too. It could perform calculations, format data, and even interact with other systems. It was like giving our AI hands to manipulate the digital world.
  4. The Agentic Evolution: Today, explAIn is a configurable, somewhat agentic experience. It’s like a Swiss Army knife of AI capabilities, adaptable to various applications. It can understand context, make decisions based on goals, and even take proactive actions in certain scenarios.

While we’re proud of how far explAIn has come, we know there’s still room for growth, especially in terms of autonomy. We’re constantly improving its decision-making capabilities and expanding its range of actions. It’s an exciting process, and every day, we’re discovering new potential applications.

The journey with explAIn has taught us that developing Agentic AI isn’t just about adding features, it’s about creating a system that can think, learn, and act increasingly sophisticatedly. It’s a path of constant innovation, and we’re thrilled to be on this ride.

The Next Frontier: Multi-Agent Systems with AutoGen

As we’ve been levelling up our Agentic AI game, we stumbled upon this cool thing called AutoGen. It’s an open-source framework from Microsoft that lets us create entire ecosystems of AI agents. Picture a bunch of AI experts sitting around a table, each with their own specialty, working together to solve complex problems. That’s AutoGen in action.

AutoGen AI Agents
Autogen Agent Frameworks

The power of AutoGen lies in its flexibility and scalability. We’ve been experimenting with it, and it’s like playing with AI Legos. You can define various types of agents, each with its own knowledge base, decision-making processes, and interaction protocols. Then comes the fun part – orchestrating these agents to work together, sharing information and leveraging each other’s strengths to achieve common goals.

For instance, we’ve tinkered with scenarios where an AI agent acting as a lawyer converses with another AI agent specialised in finance. This interaction enables a more comprehensive analysis of issues that span legal and financial domains, leading to more nuanced and well-rounded solutions. It’s like having a virtual roundtable of experts at your fingertips. It’s like your Facebook group chat with friends, but the members can think and do things (like actually being able to organise a time to catch up…).

Backend example of agents interacting with each other
Backend Example of Agents Interacting With Each Other
Autogen Example for Linking Multiple Agents Together
Autogen Example for Linking Multiple Agents TogetherSource docs: https://microsoft.github.io/autogen/docs/Getting-Started

Our experimentation with AutoGen is teaching us valuable lessons about agent communication, task delegation, and collaborative problem-solving in AI systems. We’re learning how to create more complex, interactive, and truly autonomous AI ecosystems. This framework allows us to move beyond single-agent systems, opening up possibilities for AI applications that can handle intricate, multi-step tasks requiring diverse expertise. By incorporating AutoGen into our development process, we’re pushing the boundaries of what’s possible with Agentic AI. It’s exciting to think about the potential applications – from advanced decision-making systems in business to complex problem-solving in scientific research. The future of AI isn’t just about individual smart agents; it’s about creating efficient teams of AI agents working together to tackle the world’s most challenging problems.

What’s Next in the World of Agentic AI?

Imagine a world where AI systems can tackle complex decision-making challenges that even teams of human experts struggle with. We’re talking about AI that can juggle countless factors in fields like finance, healthcare, and urban planning. Picture an AI urban planner that can design a city layout considering everything from traffic flow and energy efficiency to social equity and environmental impact. That’s the kind of holistic thinking Agentic AI is poised to deliver.

But it’s not just about big, world-changing decisions. The future of Agentic AI is also deeply personal. We’re moving towards AI assistants that don’t just respond to your commands, but anticipate your needs and take care of things before you even ask. Imagine an AI that knows you have a big presentation coming up, so it automatically reschedules your less critical meetings, orders your favourite brain food for lunch, and even suggests some power poses to boost your confidence. It’s like having a super-smart, always-on personal assistant that knows you better than you know yourself.

Of course, as these AIs become smarter and more autonomous, we’re not ignoring the ethical aspects. The future of Agentic AI isn’t just about capability; it’s about responsibility too. We’re working hard to ensure these systems remain transparent, fair, and accountable. It’s crucial that as AI becomes more integrated into our lives and decision-making processes, we can trust it to act in our best interests and in line with our values. The road ahead for Agentic AI is about creating smarter, more adaptable, and trustworthy systems that can work alongside humans in increasingly sophisticated ways. We’re moving towards a future where AI isn’t just a tool we use, but a partner we collaborate with to solve problems, boost our productivity, and enhance our creativity.

It’s an incredibly exciting time to be in the field of AI. Every day, we’re pushing the boundaries of what’s possible, and the potential applications seem endless. From revolutionising scientific research to transforming how we interact with technology in our daily lives, Agentic AI is set to be a game-changer.

So, what do you think? Are you as excited about the potential of Agentic AI as we are? How do you see it changing your life or industry in the coming years?

To learn more about how we can help you with your AI, contact our friendly team for a no-obligation discussion HERE

 

Previous Post Next Post