From Process Fixers to System Designers: Career in the Age of AI

1. AI transformation starts small. It begins with one little, but important process being made reliable and fast, and the impact reaches further than the process itself
2. One automated process leads to the next, and eventually to redesigning the whole flow. That shift — from fixing a task to rethinking a system — is where the work changes in kind, not just in scale.
3. The people who drive AI shifts are fluent in both business and tech. They are deeply knowledgeable in both domains and see how the two interact in practice.
4. Experimenting builds the thinking, not just the skill. The more you try, the more connections you find between processes, systems, and business logic. That is how you start thinking in flows rather than tasks.
5. Vibe coding is good for learning, but it's not enough. The deeper you go into actual products and solutions, the more you need to know about the things under the hood and the less you can do with just a prompt.

Mark's team moves fast. In trade finance — payments, charges, fees, financings, treasury operations for large corporate and institutional clients — speed and accuracy are both required simultaneously. When he looked at where a simple AI-powered automation could make everything quicker and more reliable, the answer was unglamorous: data entry.

A customer sends a request as a scanned document, a photo, a PDF. A human reads it, interprets it, types the data in. This human is competent, experienced and responsible, just not as quick and fatigue-proof as AI agent. And at volume, even a small delay matters. AI reads the document in any format, extracts the data, and routes it directly into the system — consistently and instantly. The sensitive process becomes a reliable one, secured by constant human oversight.

"It might sound small," Mark says, "but as we are working with large corporate customers and big institutional clients, this small thing has an impact in this big economy." Trade finance offers a set of mechanisms that lets businesses transact across borders. Faster, more accurate data flows mean faster decisions, quicker financing, and better working international collaboration.

And at a certain point the question changes from how to handle a step more accurately to how the entire flow should actually work. That is the moment where individual automation tips into something larger.

Mark's career arrived there gradually. He came into banking interested in data, showed up whenever a process needed improving, and accumulated enough experience across the boundary of business and technology that he became the person his department turned to for both. The role — AI ambassador and workflow architect — followed from the very pattern of work.

What that produces is a person who thinks in systems: how the parts of a process connect, where the failure points are, what a more reliable design would look like. " The future is that you are designing the processes, have to understand and overlook them and use your creativity at the same time " he says. AI agents run the workflows, but they are only as good as the logic behind them, and writing that logic is human work — requiring a specific kind of human.

Asked about future-proof subjects to study, Mark is direct: "It's not anymore about only studying economics, only learning IT and coding. You have to understand your business, and you also must understand the AI models and the mathematics behind it and the data science. You really have to combine this."

Designing a reliable data extraction workflow for trade finance requires knowing what trade finance does, what the failure modes look like in practice, what regulations demand, and what the technology can handle. None of those questions yield to expertise from one side alone.

When it comes to vibe coding — Mark's response is immediate: do it. Every experiment builds a connection between a system and the business logic it serves, and that is how the more complex thinking develops.

But professional practice in a complex and regulated environment adds requirements that vibe coding alone does not cover. Code in banking must follow guidelines, meet the standards, and be understood by the person deploying it. "The AI writes the code — so what does it mean? You have to explain it to your managers, to your colleagues. And if you're not understanding this, it's a nonsense," Mark says. The tool cannot explain its own reasoning or take accountability for what it produces. That remains with the person who put it into use.

Experimentation is the right starting point. Deeper understanding accumulates from it: seeing how one process connects to the next, where the system is fragile, what a better design would look like. That is the mindset shift we all need, and it arrives the same way Mark got there — by staying curious about the next problem in the chain.