A friend asked me recently what I'd been spending so much time working on. I told her I'd been experimenting with AI agents. She nodded, thought about it for a second, and then asked the obvious question:
"So... ChatGPT?"
At first, that felt like an easy question to answer. I started explaining that an agent can use tools, work through tasks over time, read files, modify code, and interact with systems around it. Some frameworks even coordinate multiple agents, each with different roles and responsibilities. The explanation wasn't wrong, but the longer I talked, the less satisfied I became with it. I realized I wasn't struggling to explain the technology. I was struggling with the metaphor.
Most of the language we use to describe modern AI systems comes from organizations. We talk about workers, managers, planners, reviewers, and teams. We discuss delegation, coordination, communication, and hierarchy. If you spend enough time reading about agent architectures, it starts to sound less like software engineering and more like organizational design. At first that seemed perfectly reasonable. Agents behave enough like people that the comparison feels natural.
Lately, though, I've started wondering whether that metaphor is quietly shaping how we think about the problem itself.
Looking at the Work
Most conversations about AI focus on the worker. Which model should we use? Which framework is best? How many agents should be involved? What roles should they have? How should they coordinate? These are useful questions, and it's easy to understand why they've become the center of the discussion. After all, the workers are the visible part of the system.
What I've found myself paying attention to, however, is not the worker but the work itself.
When I step back and look at what is actually happening, I don't really see an organization. I see work moving through a process. Information enters the system, decisions get made, tasks move forward, stall, loop, branch, and occasionally have to be done again. The thing that increasingly captures my attention isn't the individual agent performing a task. It's the flow of the work itself and the way that flow changes over time.
That distinction may sound subtle, but I think it leads to a different set of questions.
If we view AI primarily as a workforce problem, then the obvious goal is to build better workers. Smarter models, better prompting techniques, improved coordination between agents, and more sophisticated frameworks all become natural areas of focus. Much of the current AI ecosystem is focused on exactly those things.
What I find interesting is that many of the frustrations people encounter with AI don't actually feel like failures of intelligence. The models are already astonishingly capable. They can write software, summarize research, analyze documents, generate designs, and solve problems that would have seemed extraordinary only a few years ago. Yet despite all of that capability, the same kinds of problems continue to appear. Context gets lost. Work gets repeated. Requirements drift. Effort accumulates without producing meaningful progress. Teams of agents sometimes create more complexity than clarity.
Those don't strike me as intelligence problems. They strike me as process problems.
A Different Tradition of Thinking
Once I started looking at AI through that lens, I found myself thinking less about artificial intelligence and more about systems. Manufacturing, operations, quality management, and reliability engineering have spent decades studying how work moves through complex systems. They ask questions about waste, variation, bottlenecks, feedback loops, and early warning signals. When problems emerge, the goal is not simply to identify who made a mistake. The goal is to understand what the system is producing and why.
The more I think about AI, the more relevant those questions seem.
That's why many discussions about agents feel incomplete to me. We spend enormous amounts of time debating models, frameworks, team structures, and architectures. Those discussions matter, but they all tend to assume that the worker is the center of the story. Increasingly, I'm convinced that the system deserves at least as much attention.
A factory can employ brilliant workers and still generate waste if the process is poorly designed. A software team can be filled with talented engineers and still struggle if the workflow is unhealthy. And an AI system can contain remarkably capable models while still produce disappointing outcomes if the surrounding process is fragile.
The Factory Behind the Workers
Maybe that's why I keep finding myself drawn toward concepts like continuity, feedback loops, drift, supervision, operational signals, and process health. Those ideas feel less like questions about intelligence and more like questions about systems. They focus less on the capabilities of individual workers and more on the health of the environment those workers operate within.
To be clear, I don't think the workers stop mattering. Better models matter. Better tools matter. Better frameworks matter. But I increasingly suspect we're reaching a point where making the workers smarter is only part of the story.
The system matters too.
I don't know whether the factory metaphor is ultimately the right one. What I do know is that the more I work with AI, the more I find myself paying attention to the movement of the work rather than the characteristics of the worker. And if that's the right direction, then some of the next breakthroughs in AI may come not from making the workers smarter, but from understanding the systems they're working within.
Top comments (0)