Agents and Agency

It all started with a discussion. Someone raised the question: "Isn’t ‘AI agent’ just a fancy marketing term for chatbot?" It seemed a simple enough point at first. After all, chatbots are everywhere—they answer our queries, help with customer support, and manage appointments. But as the conversation unfolded, I found myself digging deeper, questioning what we really mean by “agent” and “agency” in the digital sense.

At the same time, I stumbled upon a fascinating example: a team of AI agents running a hedge fund. This wasn’t a simple chatbot responding to questions—it was a system of interconnected AI agents, each with its own specialised role, working together to analyse market data, assess risks, and make trading decisions. It was a vivid example of how far we’ve come from the static chatbots of just a few years ago.

That got me thinking: what’s really happening here? Are we witnessing the birth of something fundamentally new? And what does this mean for how we work, delegate, and lead in a digital world?

The AI Agent: A New Kind of Worker?

The term "agent" implies more than just a tool. It carries the idea of autonomy—a system that can act on your behalf, making decisions and taking action without constant supervision. But unlike humans, these agents are powered by Large Language Models (LLMs), which give them the ability to understand complex instructions, maintain context, and even collaborate with other agents.

This is where the distinction between chatbots and AI agents becomes clear:

• A chatbot is reactive, responding to predefined inputs.
• An AI agent is proactive, capable of initiating actions and collaborating with other systems.

When LLMs entered the scene, they added a layer of intelligence that chatbots could never achieve. These systems can now:

• Interpret nuanced instructions: Moving beyond rigid scripts to understand complex and ambiguous human language.
• Access external tools: Connecting to APIs to retrieve data, make calculations, or complete tasks.
• Make decisions: Based on probabilistic reasoning, they can suggest or even act on the best course of action.

But the hedge fund example I’d seen wasn’t just about one smart agent. It was about teams of agents—each with a specialised role—working together to achieve a shared goal.

A Team of Agents: The Hedge Fund Example

Let me take you back to that hedge fund. Here’s how it worked:

1. Market Data Agent: The scout. It gathered raw market data and calculated technical signals, preparing the groundwork.
2. Quant Agent: The strategist. It analysed the data and decided whether to buy, sell, or hold.
3. Risk Manager Agent: The sentinel. It evaluated the trading signal against risk thresholds.
4. Portfolio Manager Agent: The leader. It synthesised all the inputs to make the final trading decision.

Each agent operated independently but relied on the outputs of the others to perform its role. It was a textbook example of collaboration, not just automation. And yet, the system as a whole behaved like a cohesive team, much like human professionals working together to solve a complex problem.

This example struck me as more than just clever programming. It felt like a new paradigm for how we could think about AI: not as isolated tools, but as ecosystems of intelligence, capable of scaling human workflows in ways we’ve never seen before.

Agency and Control: Lessons in Leadership

The hedge fund story also surfaced deeper questions about control and delegation. As humans, we’re used to delegating tasks to others—but we still retain ultimate responsibility.

AI agents introduce a similar dynamic:

• How much agency are you willing to give away?
• How do you maintain control over the system without micromanaging every decision?

This reminded me of situational leadership

: the idea that you adapt your leadership style depending on the maturity of the team. With AI agents, it’s a similar process:

1. Directing: When the agent is first implemented, you monitor it closely, providing specific instructions.
2. Coaching: As the agent gains proficiency, you guide it with periodic feedback.
3. Supporting: You step back, allowing the agent to act autonomously while staying available for advice.
4. Delegating: The agent operates independently, requiring only occasional oversight.

The difference is that with AI agents, this process is compressed. They learn and adapt much faster than humans, but they still require careful calibration. And the ultimate responsibility—ethical and operational—remains with you.

The Bigger Picture: Ecosystems of Intelligence

The hedge fund example shows what happens when agents collaborate sequentially. But what if they worked in parallel? What if multiple agents tackled different aspects of a problem simultaneously, sharing insights in real time?

This is where the concept of digital ecosystems emerges. Instead of linear workflows, we can imagine:

• Parallel Collaboration: Agents working on various facets of a task simultaneously, such as one analysing data, another generating ideas, and a third optimising outcomes.
• Meta-Agents: A higher-level coordinator (human or AI) that ensures alignment, resolves conflicts, and refines strategies.
• Emergent Properties: When agents collaborate, the system gains capabilities no single agent could achieve, such as adaptability, robustness, and creativity.

This ecosystem model is particularly exciting because it mirrors how humans work in teams—leveraging diverse strengths while aiming for a shared goal.

My Personal Reflection

As I thought more about this, I realised how much this connects to my own work and aspirations. Like many, I want to express myself, collaborate meaningfully, and balance control with delegation. AI agents offer a glimpse of what’s possible: systems that amplify our capabilities without replacing our humanity.

But they also challenge us to rethink what it means to lead. Just as the portfolio manager in the hedge fund example synthesises inputs to make decisions, we must act as stewards of these digital ecosystems. We set the vision, guide the process, and ensure that the agents operate in alignment with our goals and values.

At the heart of this shift is a question of trust: How do we trust systems that act autonomously? For me, the answer lies in periodic reflection and recalibration. Just as you review a team’s performance, you must review your agents—ensuring they continue to serve your intent.

Leveraging AI Chatbots for RSV Education: A Partnership with ReSViNET

Discover how we partnered with ReSViNET to build an AI-powered chatbot that provides global RSV education, available via web and WhatsApp.

Web StrategiesRob Hoeijmakers

The New Paradigm: AI as Collaborative Partners

We are entering an era where AI agents are not just tools but co-creators. They will handle the mundane and the complex, freeing us to focus on what truly matters. But this also demands a shift in mindset:

• From controlling to guiding.
• From managing tasks to orchestrating systems.
• From using tools to collaborating with digital partners.

This is not just a technical evolution—it’s a cultural one. And as someone navigating this change, I find it both daunting and exhilarating. The future will belong to those who can balance agency and control, embracing AI not as a threat, but as a partner in creativity, productivity, and leadership.

Conclusion

The story of AI agents running a hedge fund is just one example of how digital agents are evolving beyond simple chatbots. They are becoming part of collaborative, intelligent ecosystems that augment our abilities and reshape our workflows. The real power lies not just in what these agents can do alone, but in how they work together—and with us—to create something greater.

The challenge ahead is to find the right balance of delegation, control, and trust. As we embrace AI agents as partners, we must remember that the ultimate responsibility for their actions rests with us. By doing so thoughtfully, we can unlock new possibilities and redefine how we create, lead, and connect in the digital age.

P.s. this has turned out to be a longer article then expected, it mostly means I am still working on more clear insights. And I will dig in deeper. Great to see you made it to here, thank you, and consider to subscribe.