AI can take on activities that previously relied on people, changing where human judgement, knowledge and skills create the most value. That makes work redesign increasingly important. Rather than trying to predict the future role in advance, organisations can involve the people closest to the work to test new approaches, learn what works and then redesign roles and processes around what they discover.
A lot of the value of AI will come from redesigning how work gets done. But we can't fully design that future state in advance. The people doing the work need the trust, freedom and capability to experiment with new ways of working, learn through doing and help shape what comes next. That makes trust and learning more than support for AI adoption. They become part of how organisations discover what the future could look like.
For most of human history, we've been remarkably poor at predicting the consequences of technological change.
When electricity emerged, attention naturally fell on generation, transmission and the machinery it could power. When computers entered organisations, we concentrated on processors, memory and storage. When the internet began connecting our homes and workplaces, it was often described as a faster and more convenient way to exchange information.
In each case, we could see the technology more easily than we could imagine the behaviours that would eventually develop around it.
At Fusion5 we spend a lot of time thinking about how we build the capability to adapt to change, and we find ourselves reflecting on that pattern as we grapple with AI and how it impacts the way we work.
Much of the early conversation has been about the technology. Which platform? Which model? What should we automate? How many licences? Which use cases should deliver quick productivity wins?
Those questions still need answers, but we're seeing the conversation around AI adoption starting to change. Organisations are realising that simply turning on the tools doesn't transform the way we work. The bigger challenge is rethinking and redesigning the work itself and building the organisational capability to do that.
48% of leaders at organisations reinventing how work gets done report meaningful enterprise value from AI, compared with 13% at organisations focused on enablement.
A task that once took three hours might now take ten minutes. That's an efficiency, but it doesn't tell us what happens to the other two hours and fifty minutes. Do we serve more customers? Improve the quality of the work? Investigate harder problems? Develop people? Increase the amount of work expected or simply reduce headcount?
AI creates the possibility, but the choices we make around it determine the value it delivers.
And that takes us into much less certain territory. We aren't simply introducing new tools into an organisation whose jobs, processes and structures remain unchanged. We're reimagining the work itself.
The challenge is that we don't yet know what that future organisation will look like. And if we can't predict it, we can't design and control our way towards it. We must create it.
When you loosen control, trust matters more
Organisations naturally develop more controls as they grow.
We create policies, procedures, position descriptions, delegated authorities and approval processes. These create consistency and allow organisations to operate at scale.
But control also carries an assumption: we know the best way. That assumption becomes harder to sustain when the work itself is changing.
If we want people to experiment with AI, challenge how work has traditionally been done and find better ways to create value, we must give them some freedom to do it. The organisation moves from prescribing every step towards establishing the boundaries within which people can explore.
45% of AI users say it feels safer to focus on their current goals than to redesign work with AI.
Why trust matters when AI changes how we work
That doesn't mean removing control altogether. The boundaries around security, privacy, risk and accountability still need to be clear. But within those boundaries, people need enough freedom to explore, experiment and exercise judgement.
We trust people to operate within the boundaries we've established. We trust them to exercise judgement, to experiment responsibly, to admit when something hasn't worked and to share what they've learned.
That trust isn't something an organisation can switch on when it launches an AI programme.
Every organisation enters this period with a trust architecture that has been built over years through the behaviour of its leaders, the way decisions have been made, how openly information has been shared and what happens when people make mistakes.
AI will put that existing trust architecture under pressure.
47% of employees in the lowest quartile of organisational trust had intentionally withheld AI knowledge, workflows or techniques, compared with 14% in the highest-trust quartile.
People are being asked to experiment with technology while simultaneously hearing that it may fundamentally change their work. Leaders are being asked where this will ultimately lead when, in many cases, they don't know either.
Uncertainty is part of the workplace for many people right now, and you can't airbrush that away. But you can influence how openly you deal with it.
Trust doesn't require leaders to pretend they have answers they don't have. Sometimes the more credible response is to be transparent about how uncertain you are, while being clear about what you do know, what you're trying to achieve and how people will be involved in working it out.
That makes trust part of the invisible architecture that keeps people participating, experimenting and moving forward through uncertainty.
Trust therefore isn't a communication workstream sitting alongside the technology programme. It's one of the conditions that enables meaningful change to happen at all.
How do you redesign work for AI? Create first, codify second.
This changes how I think about redesigning work.
We've traditionally approached organisational change by imagining a future state and then creating the structures around it. We define roles, responsibilities and processes, codify them and then help people transition into a new model or role.
AI challenges that sequence because we don't yet know enough about what the future state will be.
A job is ultimately a container around many kinds of work. One person might build relationships, interpret information, make decisions, enter data, resolve conflict and exercise professional judgement. AI will affect each of those activities differently.
The skills employers seek are changing 116% faster in occupations most exposed to AI than in those least exposed.
So rather than asking whether an entire job survives, a more useful question is what happens to the work inside it. And we can’t answer all of that from a workshop or an organisational design exercise. People must start doing.
We're seeing this already at Fusion5. Some of the most significant changes to roles are emerging as people experiment with AI, solve different problems and take on new forms of work. The work starts to change first, and the role begins to evolve around it.
Take our Legal Contracts team as an example. As they've explored and experimented with how AI agents can support contract reviews, they've learned a significant amount about how the work itself changes. By doing the work differently, they've been able to see where value has moved, what AI can safely take on and where human skills and judgement become more important. That understanding has emerged through doing, rather than trying to define the future role in advance.
That suggests a different sequence for how we think about redesigning work: create first, then codify.
Give people the capability, tools, boundaries and permission to explore how their work can be done differently. Observe what emerges and understand which experiments create value. Then formalise the successful patterns into roles, processes and organisational structures.
You can't predict the new role without doing the new work. This is one area where mid-market organisations may have an interesting advantage.
Mid-market organisations don't always have the resources of a large enterprise, but many retain something more organic in the way they operate. There are fewer layers between the person doing the work and somebody with the authority to change it. Ideas can travel more quickly, and leaders remain closer to customers and operations.
That proximity can be valuable when the answer must emerge from the work itself, but it requires organisations to resist the instinct to codify too soon. Control works when we can predict the answer. When we can't, we need to create enough freedom to discover it.
Why learning becomes critical as AI changes work
If new work must be discovered through doing, then learning takes on a different role too.
Traditionally, organisational learning has often followed the decision. A new system is selected, a new process is designed, or a new strategy is agreed, and learning helps people develop the knowledge and skills required to operate within it.
AI is different because the destination is still moving.
There is no complete playbook that an L&D team can package up and teach to everybody. Capability must develop through practice. People try something, observe the result, reflect, adjust and try again.
Learning isn't merely preparation for organisational change. It's the mechanism through which change occurs, which changes what we need to develop in people.
Technical fluency matters, but so do curiosity, collaboration, judgement and the confidence to work with uncertainty. Trust helps create the conditions in which those capabilities can be exercised. We need people who can help create the answer, not simply execute against one that has already been determined.
At Fusion5, that's part of the thinking behind our own organisation-wide learning. We've deliberately gone back to foundations such as understanding and building trust. Trust isn't a new organisational capability, and many organisations might reasonably feel it's something they already have. But AI is placing new demands on it.
As AI becomes more deeply embedded in our work, we're asking people to experiment more, exercise judgement in new ways and help reshape work when none of us can be entirely certain where that will lead. If we're going to ask more of trust, we need to make sure we understand what it looks like in practice, can recognise when it's under pressure, and continue to strengthen it.
There is, however, a very practical challenge for mid-market organisations. Learning through doing requires time.
Lean organisations rarely have large teams with spare capacity for experimentation. People have customers to serve, targets to meet and day jobs that haven't disappeared simply because they're also being asked to rethink how those jobs might work in future.
Managers therefore become important. If we tell people to experiment but continue to manage every hour against yesterday's measures of productivity, we shouldn't be surprised if experimentation struggles to happen.
Leaders need to recognise that asking people to help create new ways of working while measuring them entirely against the old ones sends competing signals. Creating the future of work requires some capacity to create it.
That may be one of the tensions mid-market organisations need to manage most deliberately: we have the proximity and agility to change quickly, but we don't always have the capacity of a larger organisation.
Why mid-market organisations have an advantage in redesigning work
There's a lot of pressure around AI to move quickly.
We have no doubt this change will happen faster than many of the technological shifts that preceded it. But the full impact, from the technology itself through to how work, roles and organisations evolve around it, may take longer to play out than we're sometimes being told.
If work is genuinely being redesigned, there is still a process of creation, observation, learning and eventually codification. New organisational structures need to emerge around what we've learned, and that takes time.
For mid-market organisations, the answer isn’t to replicate an enterprise AI transformation playbook with fewer people and a smaller budget.
We should recognise the advantages we already have and use them deliberately. The proximity and agility that many mid-market organisations retain can be particularly valuable when new ways of working need to be discovered rather than prescribed.
But those advantages only matter if we create the conditions to use them. That means investing in the invisible architecture of trust, giving people enough freedom to experiment and developing the capabilities that help them create rather than simply follow.
It also means making room for learning and resisting the temptation to define the future organisation before we've done enough to understand what it needs to become.
AI is an extraordinary new capability. But technology alone won't tell us what to do with it.
That part we still have to create.
FAQs
How is AI work redesign changing the way organisations operate?
Why is trust important when introducing AI into the workplace?
People are more likely to share what they know, try new ways of working and acknowledge uncertainty when they feel trusted. That matters with AI because much of the knowledge needed to redesign work sits with the people doing it today. Trust creates the conditions for people to participate openly in that change rather than feeling that change is simply being done to them.
Why is learning important for successful AI adoption?
Learning is more than preparation for an AI adoption strategy. When organisations cannot fully predict how roles and processes will change, people need to learn through using AI, testing new ways of working and reflecting on what creates value. In that sense, learning becomes part of how AI-driven change happens, not simply something that supports it.
How should organisations approach redesigning work for AI?
Start with the work rather than trying to define the future role in advance. With the right guardrails in place, give people room to explore where AI can help, where human judgement remains important and how activities might be done differently. As better ways of working emerge, organisations can start formalising what they've learned. Create first, then codify what works.
Baden U'Ren | Head of Learning and Development
Baden leads Learning and Development at Fusion5, designing and delivering enterprise-wide programmes. He has a particular interest in building practical AI fluency while strengthening the leadership, communication and decision-making capabilities that help organisations adapt and change.
Rebekah Pine | Director of People and Culture
Rebekah leads Fusion5’s People and Culture function, bringing together the teams and initiatives that shape the employee experience, from attracting and hiring great people to supporting their development and growth throughout their careers.
In this article
