AI Is Writing the Code—but Who Is Designing the System?
By Pablo M. Rivera | Hawaii, Colorado & East Haven, CT
I have been using AI coding tools almost every day for the past year. They are incredible. I can generate functions, debug issues, and build components in a fraction of the time it used to take. But something has been bothering me, and I think the industry needs to talk about it.
We are getting really good at producing code and really bad at understanding how it all fits together.
Missing Pieces in the Puzzle
Last month I was reviewing a system that had been built fast, mostly with AI assistance. Every individual piece worked. Clean code. Tests passing. Well-documented functions. But when I stepped back and looked at the whole thing, there were gaps everywhere. A payment flow that did not account for refunds. An API integration that fell apart when two users hit it simultaneously. A dashboard pulling from a data source that nobody was maintaining.
Each piece had been built by someone who only saw their own corner. Nobody was looking at the picture on the box.
Whether I am working from Hawaii, Colorado, or Connecticut, I keep seeing this same pattern across every organization I consult with. It points to something bigger than a few missed requirements.
Prompting Before Thinking in Systems
There is a generation of developers right now jumping straight into AI tools without building the foundational understanding of how technology actually works under the hood. Networking, databases, distributed systems, failure modes—the stuff that is not glamorous but gives you the ability to look at a complex system and immediately spot what is wrong or what is missing.
They can get AI to spin up a microservice in minutes. But ask them why it should be a microservice instead of a library, or what happens when the service it depends on changes its API contract, or how the thing behaves when traffic spikes 10x, and you get a blank stare.
They skipped the training that builds architectural intuition.
Why Operations Leaders See This Differently
I did not come to technology through a traditional CS path. I spent 25 years running operations—managing 120 technicians across 12 states from Hawaii to Connecticut, overseeing billions in assets, coordinating supply chains and field teams and vendor relationships across markets in Colorado and beyond. When you run a national operation, you cannot think about one market at a time. You have to see how the warehouse affects the field crew, how the field crew affects the customer, and how the customer affects the revenue that funds the whole thing.
That way of thinking—seeing flows and dependencies and failure points across an entire system—is exactly what is needed in technology architecture. And it is exactly what is disappearing.
When I look at a system today, I do not see individual services or databases. I see where data moves, where processes will break, where one team’s shortcut creates a six-month headache for another team. Most people building with AI do not think this way because they never had to.
The Correction That Is Coming
The hype right now is all about speed. Faster code, more features, shorter sprints. Companies are celebrating how quickly they can ship. But they are accumulating what I call integration debt—every AI-generated component that works in isolation but creates friction at the boundaries between services, between teams, between systems.
Organizations are going to look around and realize they have more code than ever and fewer people who understand how any of it connects. When that correction happens, the most valuable people will not be the fastest coders. They will be the ones who can look at the whole board and see what is missing.
Systems thinkers. Architects. People trained to hold complexity in their heads and make judgment calls about trade-offs that AI simply cannot make.
That gap is real and growing. Whether you are building teams in Hawaii, scaling operations in Colorado, or running distributed systems from East Haven, the need for big-picture thinkers has never been greater.
Pablo M. Rivera is a bilingual operations executive and technologist based in Hawaii, Colorado, and East Haven, CT, with 25 years of experience bridging operations leadership and technology strategy. Connect on LinkedIn.
Top comments (0)