Many years ago, we built a multi-agent environment using an open-source technology called JADE. Back then, there were no cloud platforms, no large language models, and certainly no pre-built agent environments. We ran everything locally—and every time we powered it up, the lights in the building flickered. It was a thrilling reminder that we were literally drawing power to simulate intelligence.
That early experiment opened a doorway into complex adaptive systems—worlds where behavior is non-linear, emergent, and often unpredictable. We quickly discovered that the real challenge wasn’t the agents themselves, but the architecture of the environment they lived in. To make it work, we needed a deep understanding of business rules and pathways; without that foundation, even the most sophisticated agent would collapse into chaos.
Because no tools existed to model that kind of complexity, we built our own. We wrapped our logic into what we called a belief system—a framework that defined how an agent would reason, decide, and act. Once encapsulated, each agent entered the environment to Sense → Think → Act → Learn. This circular feedback mechanism—its beliefs, rules, and interactions—was the heartbeat of the system.
The difficult part was building this within a fully decentralized environment, where every agent was an autonomous entity capable of self-organization. Each had to collaborate, cooperate, or coordinate with others without any central control. Thinking this through stretched the limits of our then-linear mindset. Traditional modeling techniques simply couldn’t describe the behavior of such an adaptive, distributed system. Hence, we had to invent our own.
That experience still shapes my view of how business designers and toolmakers must evolve. Most enterprise tools—process modeling, workflow management, value chain mapping—still assume linearity. They describe cause and effect in straight lines, but the modern business system behaves far more like the London Underground: complex, interlinked, and occasionally delayed by mysterious forces.
As we step deeper into the multi-agent era, success will depend less on the brilliance of individual agents and more on the architecture of interaction—the system of systems they inhabit. Modern agents operate within graphs: networks of nodes, edges, and states, where knowledge, intent, and action intersect. To navigate that world, we must learn to design like it—non-linear, adaptive, and richly connected.
Humans, fortunately, are wired for this kind of complexity. We connect dots, improvise, and find strange, inventive routes through problems. Yet our tools and frameworks still lag behind our natural capacity. The disciplines of business design, business architecture, and business analysis remain mostly linear: process maps, value streams, sequential flows.The business architects, designers, and analysts of the future will need tools—and mental models—that can move fluidly between modes of thought. We need to model flows and feedback, not just steps and stages. We need architectures that describe interaction, not just execution.
What’s needed is a multi-model mindset, knowledge-set, skillset and toolset—integrating multiple ways of seeing and simulating complexity. I’ve seen hints of this in platforms like AnyLogic, which combines system dynamics, agent-based modeling, and discrete event simulation. It’s a promising direction, but even there, the potential of integrating these three paradigms hasn’t been fully realized. Most tools still force users to choose one lens at a time, rather than blending them into a unified model capable of representing true autonomy and emergence.
It’s time for that to change. To stay relevant in a world where AI is accelerating faster than we can redraw our diagrams, we need to embrace multi-model thinking, systems pattern design, and dynamic ways of working. The future of intelligent enterprise design won’t be built on straight lines—it will be built on networks, feedback loops, and emergent structures that can learn and adapt as quickly as the machines we create.