AI changes more than the tech stack

Over the last year, many of my conversations about AI across business functions, IT organizations and executive teams have started in the same place:

“Which model are we using?”

“What's our AI platform strategy?”

“Should we standardize on Copilot, ChatGPT or something else?”

These are valid questions. But they are not the most important ones.

The biggest impact of AI is not on the technology stack. It is on the work stack.

We have been focusing on the wrong layer

Historically, technology transformations centered on systems.

We implemented ERP platforms to standardize transactions. We deployed CRM systems to manage customer relationships. We built workflows to move information from one team to another. Success was often measured by whether the technology was implemented on time, on budget and adopted by users.

AI shifts the conversation.

Instead of asking how people should use software, we are increasingly asking how software can participate in the work itself.

That is a fundamentally different problem.

The question is no longer, “How do we automate a process?”

The question is, “Which decisions, judgments, analyses and actions within that process can be delegated, assisted or augmented?”

That is a much bigger change than introducing another application into the environment.

The workflow is the unit of transformation

Most organizations are still thinking about AI at the tool level.

I believe the more useful lens is to think about tasks, decisions and outcomes within an end-to-end workflow.

Take a typical finance process. The obvious opportunities might be document extraction, invoice matching or report generation.

But when you observe how work actually gets done, you quickly discover that much of the effort is not spent entering data. It is spent interpreting exceptions, applying institutional knowledge, resolving ambiguity, reconciling conflicting information and making judgment calls.

These activities rarely exist in process maps. They are embedded in the heads of experienced employees.

AI forces organizations to make this hidden layer visible.

Before an agent can assist with a decision, we have to understand how that decision is made. Before a workflow can become autonomous, we have to codify the business rules that humans apply instinctively.

In many ways, AI is becoming one of the most consequential process discovery exercises companies have undertaken.

AI is exposing organizational knowledge debt

Every company has technical debt.

Most companies also have knowledge debt.

Knowledge debt accumulates when critical business logic exists only in tribal knowledge, email threads, undocumented workarounds or the experience of a handful of subject matter experts.

For years, organizations could operate this way because people bridged the gaps.

AI does not eliminate the need for expertise. It highlights where expertise is undocumented.

That is why many AI initiatives stall after the proof-of-concept stage. The technology works. The models work. The platform works.

What does not exist is a clear, documented understanding of how the business actually operates.

Organizations discover that the bottleneck is not AI maturity. It is process maturity.

The most consequential AI skill may not be prompting

There has been enormous focus on prompt engineering and model literacy. Those skills matter.

But I increasingly believe the more valuable capability is the ability to deconstruct work.

The people creating the biggest impact with AI are not necessarily the ones who know the most about models. They are the people who can look at a business process and ask:

• What decision is being made?
• What information is required?
• What rules govern the outcome?
• What exceptions exist?
• What level of confidence is acceptable?

In other words, they understand the system behind the system.

This is why process experts, operations leaders, analysts and business subject matter experts remain incredibly important in an AI-driven organization.

AI does not replace domain expertise. It amplifies it.

The boundary between business and technology is dissolving

For years, organizations separated technology teams from business teams.

Business defined requirements. Technology built solutions.

AI is blurring that boundary.

Successful AI initiatives require business experts, operations leaders, process owners, architects and engineers to work as a single design team.

The challenge is no longer about translating requirements into software. It is about translating human expertise into executable logic.

That requires a different way of working.

The organizations that move fastest will be the ones that stop thinking of AI as a technology project and start treating it as an operating-model transformation.

The objective is not automation. It is redesign.

This may sound counterintuitive, but I do not think the end goal is automation.

Automation is a byproduct. The real opportunity is redesign.

When organizations begin questioning every task, every handoff, every approval and every decision point, they inevitably discover that many processes exist simply because they always have.

AI gives us a reason to rethink them.

Some work will be automated. Some will be augmented. Some will disappear entirely. And entirely new forms of work will emerge around governance, oversight, orchestration and continuous improvement.

The organizations that create the most value will not necessarily be the ones that automate the most. They will be the ones that redesign the fastest.

Final thought

Every major technology wave changes how work is organized. AI is different in how directly it can participate in the work itself, including analysis, interpretation and judgment that previously required a person.

The conversations that matter most are no longer only about models, vendors or infrastructure. Those are important, but they are table stakes.

The more strategic question is this:

If intelligence becomes accessible on demand, how should work be redesigned?

The organizations that answer that question well will not just have a modern AI stack. They will have a fundamentally different operating model.

And that is where the real transformation begins.