There is a familiar pattern in how organisations approach major change. They invest in technology, process redesign, and a training plan. They commission a change management workstream, usually late in the programme when the design is already set. They measure adoption at go-live and call it done.
This pattern has always been inadequate. For AI, it is a route to failure at a scale most organisations have not yet reckoned with.
Thanks for reading The Operating Model Dispatch! Subscribe for free to receive new posts and support my work.
Done properly, the work of understanding your operating model before you redesign it is the beginning of the change itself.
What landscape mapping is actually doing
When you map your operating model landscape before redesigning it, you are not only gathering evidence. You are establishing a current state definition: a clear, honest picture of how the organisation actually functions, which is almost never the same as how it is supposed to function on paper. You are finding the people who hold the informal knowledge, who own the undocumented handoffs, who have built the workarounds that keep things moving.
This matters for a reason that goes beyond analytical rigour. The work of mapping your current state does not begin with the answer. No one commissioning this work right now knows with confidence where they want their organisation to be once AI is embedded in it. No one knows where AI will be in two or three years, or what it will mean for work, for roles, for organisations, or for any of us. The mapping starts from where the organisation genuinely is, not where it believes itself to be. It surfaces what is working and what is not, where the foundations are sound and where they need to be rebuilt. It establishes the starting point from which deliberate design can begin. The goal is to understand what exists, so that whatever comes next is based on something real rather than something assumed.
This applies as much to organisations that are functioning well as to those that know something is wrong. Even a well-designed operating model built for today will need its foundations understood and mapped before AI can be introduced into it deliberately. You cannot build reliably on a foundation you have never seen clearly.
A good mapping process also does something an audit never does. People are not only asked to describe how things work. They are asked what they could contribute if the conditions were right, what is slowing them down, what is wasting their time and effort. That is not only better analysis. It is the beginning of the change.
Why involvement is doing the change work
In most programmes, change management appears in the implementation plan. That is not managing change. It is managing the fallout from a design process that excluded the people it was designed for.
When it starts at the beginning with the mapping, the people who were interviewed have been heard. The problems they named have been taken seriously. The gaps between what they are asked to do and what the system actually enables them to do have been made visible. They were asked what they could do if things were easier, faster, better. They were asked what they know and what they could contribute that the current operating model is not drawing on. They were included in understanding where the organisation is, before anyone has decided where it is going.
This is change happening with people, not to them. And that distinction is the difference between a programme that holds and one that unravels.
People can hold uncertainty, even significant uncertainty, if they feel genuinely included in the process of navigating it. They do not need to know on Day 1 where the organisation is going to end up. No one does, and the honest answer in most AI adoption efforts is that the destination is genuinely unknown. What people need is to know that they have been listened to, that the people designing the change understand their reality, and that the redesign is being built on that understanding rather than imported from a template and imposed.
When those conditions are met, something specific happens: people are bought in before the design is finished. Not to a specific outcome, because the outcome is not yet clear, but to the process. They understand why the change is happening because they contributed to understanding what needs to change. They recognise themselves in the result because their experience shaped it. That buy-in is the operational foundation for adoption.
The fear that leads to sabotage can be real. Not everyone will come with you, and pretending otherwise is a form of planning failure. Some resistance is practical: people who see a genuine problem with the design and are right to surface it. That is useful. Some resistance is political: people with interests in the current model who will work against the new one regardless of how it is designed. That requires a different response. But a significant portion of resistance in AI programmes specifically is fear. It is fear of what AI means for this role, this career, this sense of professional identity and value. People who feel that fear and also feel that the change was designed without them do not just fail to adopt. They may actively undermine the effort.
Early and genuine involvement does not eliminate that fear. In some cases the fear is well-founded: AI will change what some roles look like and eliminate others, and honesty about that is part of what builds trust. But involvement changes the relationship to the fear. People who have been part of the landscape mapping, who have been asked what they find difficult and what they could do with better tools and fewer constraints, are not navigating the change as something that is happening to them. They are part of a process that began with their experience and is being built around their reality. That does not make every fear go away. It makes the fear addressable.
Human-centricity is a design requirement, not a value statement
Human-centricity is sometimes treated as a tone: be empathetic, communicate well, involve people. That matters, but the argument here is more structural than that.
An operating model is not a set of processes and systems. It is the way human beings coordinate to turn strategic intent into execution. The processes and systems exist to support that coordination. When the design does not reflect how people actually work, think, and make decisions, the processes and systems are working against the people rather than with them. Adoption fails because the change is asking people to operate in ways the design has not made possible.
A design that reflects that reality does not require people to abandon what they know and adopt something foreign. It takes what works and builds on it deliberately, while addressing what does not. People recognise themselves in the result because they were part of producing it.
This is also where enterprise-wide change has to be designed in from the beginning, rather than added as a final workstream. The people who use a redesigned workflow are embedded in a wider organisation. Their managers, their adjacent teams, and the senior leaders who set the conditions for their work all have a role in whether the change holds. If that wider context has not been mapped and designed for, the change holds only in controlled conditions and unravels when it meets the broader organisation.
Why AI changes the stakes entirely
Previous enterprise technology changes were significant. ERP implementations, CRM rollouts, Agile transformations: all of them required genuine organisational change and all of them produced real casualties when the change was managed poorly. But they were bounded. They changed how people worked within the organisation. They did not change the world outside it.
AI is different in kind.
The closest parallel in recent history is the shift to ubiquitous internet. The internet changed how people lived: how they accessed information, how they formed relationships, how they understood the world and their place in it. It altered the conditions of daily life for almost everyone, and the organisations implementing internet-era changes were doing so into a workforce that was also living through that broader transformation outside of work. But even that comparison undersells what is happening now.
The internet changed the conditions of daily life. AI is raising questions about the nature of daily life itself. What it means to create, to decide, to learn, to contribute. Who benefits from the productivity gains and who bears the displacement costs. Whether the models making consequential decisions can be interrogated, and by whom. What the energy and environmental cost of this technology at scale actually means against the sustainability commitments organisations have made. What people entering the workforce now can reasonably expect their working lives to look like, and whether the future being built is one that was designed with them in mind or simply handed to them.
These are not fringe concerns held by a minority. They are the substance of a civilisational conversation that is happening in living rooms, lecture theatres, newsrooms, and parliaments. Your people are part of that conversation. They arrive at work with those questions already active, already partially formed, already shaping how they feel about what their organisation is asking them to do. This is not background noise. It is the context inside which every AI programme is operating, whether the programme acknowledges it or not.
The landscape mapping work calls for precision here. It surfaces what people actually think and fear and believe about AI in the context of this organisation and this role. It distinguishes between resistance that reflects a genuine design problem and anxiety that reflects the broader disruption. It identifies where the concerns are about this change and where they are about the larger one. Without that distinction, change management is working in the dark.
The stakes of getting this wrong are also categorically higher than they were for earlier enterprise changes. Organisations that fail to adopt AI effectively are not just behind on a technology implementation. They are falling behind in a shift that is not going to reverse. The gap will compound. The organisations that are getting this right are not the ones with the most sophisticated technology. They are the ones that understood what they were actually asking of their people, built the operating model to support it, and designed the change from the foundation up rather than from the plan down.
What this means in practice
The argument is not complicated, but it is consistently resisted because it requires the current state work to happen before the design is set, and most programmes want to begin designing before they have finished understanding. The mapping does not need to be exhaustive before work begins. A broad understanding of the whole, deepened progressively as each part is redesigned, is both sufficient and more honest than a design built on untested assumptions.
Map the operating model landscape as it actually is, not as it is supposed to be. That means the people who execute the work, not only the people who describe it. It means the informal systems alongside the formal ones. It means the workarounds, the load-bearing individuals, the undocumented handoffs that hold the organisation together. It also means asking the questions that surface what people could do, not only what is going wrong.
Design the operating model to be human-centric and AI-enabled in that order. Human-centric first, because the operating model exists to coordinate people. AI-enabled second, because AI is most powerful when it is introduced into a foundation that has been deliberately designed to absorb it.
Build enterprise change into the design from the beginning as a genuine account of what is being asked of people across the whole organisation, at every level, and what the conditions are for them to do it.
Your people are navigating the largest technological shift of their lifetimes, and they are doing it at home and at work simultaneously. The fear is understandable and in places well-founded. The uncertainty is shared by everyone, including the people designing the programmes. Offering that honesty early, and meaning it, is one of the few things that actually builds real trust.
The people you are asking to change need to feel that this was built with them. If that condition has not been met, you already know how this ends.