Last week I attended a conference where public organizations presented their digitization projects. What struck me: many projects support old processes with new technology. But the processes themselves? Those remained what they were: designed in a paper era, only now with a digital sauce. A room full of digital promises, but little real digitality. A digital future requires a digital beginning - not a digitized past.
And that difference matters: because only true digital design can provide agility, scale and intelligence in a data-driven society. Digitizing an existing process is very different from redesigning it into a digital process.
Digitizing, in too many cases, means optimizing the existing process with digital means. The thinking models remain the same. Management steers for efficiency and quick wins - not for fundamental redesign. Architecture is absent, data-driven innovation is missing. Underlying process thinking remains analog and linear. Digitization is a pragmatic translation of the past - not a design for the future.
Working truly digitally means: thinking from data and building on architecture. Not the process is central, but the dataflow. Processes are modular, adaptive and supportive. In my earlier blog The data flower: the beauty of data-centric thinking, I described this with a metaphor: the core is data, the petals are processes that create, edit or enrich data. Each process starts with data, uses data and ends with new data.
Such as a self-service portal that automatically verifies identity data. Or event-driven infrastructures in which processes start based on data mutations or API-driven decision-making without human intervention. These are examples of digital processes that are not merely optimized, but fundamentally redesigned.
Yet not everything is digitizable. Some processes are unavoidably physical: delivering goods, providing care, performing maintenance. You can't digitize those completely. But you can surround them with digital layers: tracking, planning, documentation, communication. The key question then is: are you process-centric or data-centric? Once you work with data, you can digitally (re)design a process.
And then there is the human being. Who is and remains analogous. Thinking, feeling, deciding - that doesn't happen binary. Digital systems register, analyze, optimize. The challenge is in the design of interaction: not automation instead of, but alongside humans. Augmented intelligence instead of artificial intelligence.
Many government projects carry the legacy of the past. Cleverly conceived processes from the 1990s - once modern, built under architecture - are technically obsolete, but still leading. Administrative reality is risk-averse, focused on control. And often the architect has disappeared. The scope for redesign has narrowed.
A "green pasture" approach, in which you set up a truly digital process in parallel with the existing one, takes managerial guts. And patience. And a willingness to be temporarily inefficient. But that space is rarely created. So we digitize the old. And remain trapped in the past. Without guts, no digital future. Without architecture, no foundation.
This is precisely where AI comes in. Because AI functions exclusively in a digital context - and makes painfully visible where processes are not yet. AI is the ultimate test: it makes visible where our systems are still analog or process-oriented.
An AI that predicts, analyzes or advises needs access to clean, current, standardized data. Not half-scanned PDFs or embedded Excel sheets. And at the same time, AI reminds us that humans are still needed: for interpretation, ethics and consideration. The future is in the balance. How do you design processes in which humans and machines work together?
Too often we think only in processes, but rarely in systems. Digitizing is process thinking with digital means. Digital design means systems thinking: data ánd processes, analog ánd digital. Not either-or, but both-and-one.
This requires a different type of model thinking. Not just a process model (who does what in what order), but especially a data and information model: who creates what information, who owns it, who are the users, who can do what, when and why? Information management is the bridge here: it connects the logic of processes with the structure of information and the underlying data model.
A powerful tool for this is IDEF - Integrated Definition - a methodology from the systems world that structures functions and information flows. It forces us to think in terms of system functionality: what should the system do, regardless of how it is implemented? From that functional specification, you develop a system definition. From that follows a product model (how do we realize the integrated whole), a production model (how do we build it) and finally a lifecycle model (how the system will be used, maintained and renewed during its lifecycle). See also my earlier article: A case for responsible digitization.
AI desperately needs this systems thinking. Because AI must be able to feed on structural, reliable, meaningful data. AI thrives not in a world of separate process fragments, but in a well-defined systems landscape. Therefore, we must stop developing "intelligent processes" and start designing intelligent systems.
Systems in which data flows, functions are logically structured and human interaction is consciously designed.
To digitize is a verb. Digital is a vision.
Those who truly want to be digital must dare to redesign. Not optimize what is there, but re-imagine what is possible. From data. With architecture. And with space for people.
A digital future requires a digital beginning - not a digitized past.
