Are we building systems for the future, or simply digitizing the past?

Introduction: a question I keep coming back to

Over the past few years, I have increasingly found myself reflecting on a question that extends far beyond Data Governance or Data Management. It concerns the way large organizations approach change itself.

Does it still make sense to invest substantial financial resources, time, and human effort into organizational and technological models when there is a real possibility that the assumptions behind them will become outdated before those investments have fully paid off?

This question becomes particularly relevant in the context of the accelerating development of artificial intelligence.

Companies continue to invest in multi-year transformation programs, engage consulting firms, introduce new methodologies, define organizational responsibilities, implement platforms, and train employees.

All of these activities serve a purpose.

However, while such projects are being planned, approved, and implemented, technological capabilities continue to evolve.

What required a substantial development team a year ago may now be demonstrated through a functional prototype in a matter of days or weeks. What we consider an advanced solution today may become a standard feature of an existing platform within the next year.

Of course, a functional prototype is not the same as a reliable production-ready solution. But that distinction raises another important question: is technology itself changing, or is the entire economics of developing business solutions changing with it?

I am not suggesting that organizations should stop planning or investing.

I am suggesting that we need to examine more critically what we are building, why we are building it, and whether our current ways of working will remain justified as technological capabilities evolve.

1. The paradox of large transformation projects

Consider a typical Data Governance transformation initiative.

An organization identifies the need to improve its approach to data management. This is followed by assessments of the current state, workshops, maturity evaluations, responsibility definitions, policy development, control frameworks, technology selection, and implementation planning.

Such projects can take months, sometimes years.

Meanwhile, technology, market conditions, and automation capabilities continue to change.

By the time the project is completed, the organization may have a well-documented and technically functional system built around assumptions that were valid when the project began.

And this is where I see a potential problem.

It does not necessarily mean that the system is poorly designed. Nor does it automatically mean that the investment was a mistake.

But it is entirely possible that the same business outcome could now be achieved more simply, more quickly, or with lower operational costs using technology that became available during the project’s implementation.

This creates a paradox of digital transformation.

An organization can successfully complete a transformation project while simultaneously missing an opportunity to fundamentally change how the work is performed.

For this reason, I believe transformation success should be measured primarily through actual business outcomes rather than the completion of planned deliverables.

2. Data Governance is not the same as its traditional implementation

When discussing the future of Data Governance, the conversation often focuses on whether artificial intelligence will replace certain professionals or automate specific tasks.

I believe that question is too narrow.

A more interesting question is: will some of the activities we currently consider essential to Data Governance even need to exist in their present form?

Traditional data governance models involve defining policies, monitoring data quality, analyzing discrepancies, producing reports, managing incidents, and tracking corrective actions.

Many of these processes were developed in an environment where data analysis, control development, and information integration required significant human effort.

Artificial intelligence now creates opportunities for a different approach.

Imagine a system that continuously monitors data quality, identifies discrepancies, connects them to business processes, investigates potential root causes, and recommends corrective actions.

Instead of employees regularly reviewing hundreds or thousands of individual data quality issues, such a system could group problems by their underlying causes and highlight only those cases requiring human judgment or a business decision.

This is not merely the automation of an existing process. It is an opportunity to redesign the process itself.

Naturally, such a system must operate within clearly defined business rules, accountability structures, automation limits, and verification mechanisms. AI cannot arbitrarily determine which data is correct from a business perspective, nor can it independently assume accountability for the consequences of its decisions.

However, that does not change the fundamental argument.

The purpose of Data Governance remains relevant, but the way we achieve that purpose does not necessarily have to remain the same.

3. Digitizing existing processes is not necessarily transformation

This brings me to what I consider one of the most important questions.

When organizations introduce new technologies, they often begin with an existing business process. They analyze how an activity is currently performed, identify opportunities for automation, and develop a digital solution that makes the process more efficient.

This approach can deliver substantial benefits.

But it also has a limitation. It assumes that the existing process, with some improvements, should continue to exist.

What if that assumption is no longer valid?

For example, suppose an organization has a process in which data quality control results are collected, analyzed, distributed to responsible teams, and tracked through several stages of corrective action.

We could develop an application to digitize that entire workflow. The result would likely provide better visibility, faster reporting, and more efficient tracking.

But perhaps we would have missed a more important question. Why do we need to organize the process this way in the first place?

Perhaps we no longer need so many reports. Perhaps we do not need the same escalation levels. Perhaps certain activities could be fully automated, with human involvement reserved for situations involving genuine business risk or decisions that require authorization.

In other words, perhaps we do not need a better digital version of the existing process. Perhaps we need an entirely different process.

There is a fundamental difference between using new technology to improve an old way of working and using new technology to question whether that way of working is still necessary.

I believe the ability to recognize this distinction will become one of the most important organizational capabilities in the years ahead.

4. Consulting models and the question of real business value

In this context, I cannot ignore the role of consulting firms, methodologies, and professional training.

I do not believe consultants are unnecessary, nor do I believe traditional methodologies have lost all their value. On the contrary, high-quality external expertise can be extremely valuable, particularly in complex organizations that must align business, regulatory, and technological requirements.

However, I believe we need to change how we evaluate the value of such engagements.

If an organization completes a multi-month project and receives a new methodology, defined responsibilities, documented processes, and numerous presentations, that does not automatically mean meaningful improvement has been achieved.

• Has data quality actually improved?
• Has the number of recurring problems decreased?
• Are incidents resolved faster?
• Have business risks been reduced?
• Have operational costs declined?
• Can the organization independently maintain and improve the implemented solution?

In my view, these questions should carry more weight than the volume of documentation delivered.

It is particularly important to reconsider engagement models in which substantial time is spent on activities that can now be automated or significantly accelerated.

This does not mean every workshop can be replaced by an AI agent, or that complex business decisions have suddenly become simple. It means that consulting methodologies should evolve alongside technological capabilities.

Otherwise, organizations risk continuing to pay for traditional service delivery models even when the possibilities for delivering those services have fundamentally changed.

5. The greatest risk may not be technological obsolescence

We often hear about the risk of companies falling behind their competitors technologically. That concern is justified.

However, I believe there is another, less visible risk. An organization may have access to advanced technology but remain unable to use it effectively because of its own processes, structures, and decision-making models.

Large companies operate through established budgeting cycles, investment procedures, accountability frameworks, and multi-year plans. These mechanisms exist for good reasons. Their purpose is to provide control, predictability, and responsible resource management.

But what happens when technological change accelerates significantly? What happens if, halfway through a multi-year project, a new approach emerges that can achieve the original objective in a fundamentally different way?

Can the organization change direction? Can it reconsider an investment that has already been approved? Can it acknowledge that part of the original project is no longer necessary?

Or will it continue because the budget has been allocated, the project approved, and the expected deliverables defined?

I am not suggesting that every organization behaves this way. But I believe the risk is real.

An organization’s ability to adapt may become more important than its ability to acquire the latest technology.

A company that can quickly recognize change and adjust its processes may create more value than one with superior technology operating within an outdated organizational model.

6. How should we make investment decisions when the future is unpredictable?

This leads to perhaps the most difficult question. If technology is evolving so rapidly, should companies postpone major investments?

I do not think so. Waiting for the perfect technology is not a realistic strategy because something new will always emerge.

At the same time, investing in rigid systems that are difficult to adapt does not seem sustainable in the long term.

For this reason, I would place greater emphasis on several principles.

First, we need to clearly distinguish the business objective from the technology used to achieve it. If the objective is to improve master data quality, that objective is relatively stable. However, the technology, architecture, and mechanisms used to implement data quality controls may change significantly.

Second, I believe major projects should be structured to demonstrate business value incrementally. Instead of designing an entire solution around assumptions extending several years into the future, organizations can begin with a limited experiment, measure the results, and use that evidence to guide further investment.

Third, we should evaluate how adaptable a solution really is. If a more capable AI model becomes available next year, can we integrate it without rebuilding the entire system? If a business process changes, can we adjust the relevant rules and controls? If regulatory requirements evolve, can we demonstrate that the system remains compliant?

Finally, I believe organizations should pay much greater attention to the future cost of change. Knowing the initial implementation cost is not enough. We also need to understand how much it will cost to adapt the solution tomorrow.

Of course, modularity and flexibility are not free. Their value must be proportional to actual business needs and risks.

The objective should not be to build the most complex architecture imaginable in anticipation of every possible future scenario. The objective should be to avoid unnecessary dependencies that could prevent an organization from adopting better solutions when they become available.

7. What does this mean for data management professionals?

These questions are not relevant only to companies. They also concern those of us working professionally in Data Governance, Data Management, and Data Quality.

If artificial intelligence becomes increasingly capable of writing SQL queries, analyzing data, generating documentation, identifying anomalies, and recommending business rules, we need to reconsider how the value of our professional skills will evolve.

I do not believe the answer lies in trying to master every new AI tool. That is not a sustainable long-term strategy.

What matters more is understanding business problems, system architecture, organizational processes, and how technology can generate measurable value.

A professional who knows how to produce a data quality report certainly possesses useful skills. But a professional who can identify why data quality problems occur, design a system that prevents them, and demonstrate the business value of that solution plays a much broader role.

This does not mean such professionals will be immune to future automation. It means their value will increasingly depend less on knowledge of individual tools and more on the ability to connect business needs, data, technology, and responsible decision-making.

I believe this combination will become increasingly important.

8. Should we reconsider the organizational structure of Data Governance itself?

This is a question I would particularly like to emphasize.

If artificial intelligence significantly changes how controls, analysis, documentation, and data monitoring are performed, it is reasonable to ask whether the organizational structure supporting these activities should also change.

Large organizations often have separate functions responsible for data quality, data management, analytics, reporting, controls, and compliance. There are valid reasons for such divisions.

However, if operational activities become increasingly automated, it may become necessary to redefine the boundaries between these functions.

Some responsibilities may remain unchanged while the methods used to fulfill them evolve significantly. New roles may emerge around the supervision of automated processes, validation of AI-generated decisions, risk management, and assurance of data reliability. And some activities that are currently considered essential may gradually lose their operational importance.

I would not attempt to predict exactly what such an organization should look like. But I believe it would be a mistake to assume that artificial intelligence will transform technology while leaving organizational models untouched.

If the way work is performed changes, we must also be prepared to reconsider how that work is organized.

9. Instead of a conclusion: are we asking the right questions?

When I think about the future of Data Governance and Data Management, I find myself becoming less interested in predicting which tool or platform will be the best in a few years.

I am far more interested in how we can build organizations that will not need to undergo another major transformation every time a significant technological breakthrough occurs.

That is why I believe we should ask a few simple questions before committing to any major investment.

• What business problem are we actually trying to solve?
• Is our current way of working still justified?
• Are we attempting to improve a process that may no longer be necessary in its present form?
• Can we demonstrate business value before committing to a large-scale investment?
• And will the solution we build today allow the organization to adopt better technology tomorrow without repeating the entire project?

I have no illusions that these questions have simple answers. Large organizations cannot change at the same speed as AI models evolve. Regulatory obligations, security requirements, complex integrations, and accountability structures cannot simply be bypassed.

But that is precisely why I believe we need to rethink how we approach transformation.

We do not need to stop building systems for the long term. We need to stop assuming that the methods we use today to achieve business objectives will remain optimal in the future.

Perhaps the greatest risk is not that the technology we implement today will become obsolete in a few years. Perhaps the greatest risk is that we will continue developing, funding, and maintaining ways of working that new technology has already made unnecessarily complex.

And that brings me back to the question I started with.

Are we building systems for the future, or are we simply creating increasingly sophisticated digital versions of the past?

I believe it is time we started asking that question before major transformation projects begin, rather than after they have been completed.