This document is not about slowing down. It is about steering. Six messages about what AI actually does to work, organisation, and leadership, and what you as leaders need to do to take control of the process before it takes control of you.
The six messages:
- AI development requires active leadership
- Stabilisers and explorers need to collaborate for scalable innovation
- AI is the team’s colleague, not only the individual’s tool
- AI creates new possibilities for continuous organising
- AI is already in your systems
- Reverse the competence debt before it reverses you
1. AI development requires active leadership
Background
We do not need a new kind of leadership. We need a kind of leadership we have almost forgotten.
In the first months of the pandemic, leaders stepped forward in a way we rarely see otherwise. They were visible. They communicated often and honestly. They said: this is what we know, this is what we do not know and based on that we are taking this step now. They did not have all the answers, but they were present. That created a sense of calm, and trust in leadership grew.
That is exactly what is missing now. Employees can see that AI is changing jobs. They see cutbacks and reorganisations. They quietly wonder: will my job still exist in a year? They do not need perfect answers. They need a manager who is willing to step forward and walk alongside them through the major change that AI represents.
For managers to be able to do this, they need support. The leader is expected to be calm when others are anxious. Clear when the picture is unclear. Courageous when they themselves are afraid. That is an almost inhuman expectation, and yet we almost never talk about it.
Leaders need a space for honesty. A place where they can put words to what they genuinely do not know and what trade-offs they are wrestling with. That might be a leadership team with the right culture, a coach, or a network of peers. But it must exist. Because if it does not, the leader carries everything alone. And that will show in the organisation.
Leaders also need permission not to know. That sounds simple, but it is structurally difficult. Owners, boards, and markets reward decisiveness and certainty. That creates pressure to communicate more confidence than you actually feel.
We need leaders who lead the way they did at the start of the pandemic. But to make that possible, we also need to take care of the people who are doing the leading.
Example
A manager receives questions from employees about AI, future roles, and whether their skills will be enough. At the same time the manager feels uncertain and has received no clear direction from above. The result is silence, cautious answers, and an anxiety that grows within the team.
What you can do
Step 1. Create a space for honesty. Not a room where you are expected to have answers, but a place where you can put words to what you genuinely do not know. That might be a leadership team with the right culture, a coach, or a network of peers in similar positions. But it must exist somewhere.
Step 2. Tell it as it is. Be willing to describe reality to your employees. Outline different scenarios and point to different paths forward. Leaders do not need all the answers, but they do need to inspire confidence that we will navigate this transition together.
Step 3. Build structures that support leaders. Give leaders the opportunity to spend time on what actually makes a difference. Review the number of direct reports per manager. Simplify and automate administration.
Questions for leadership
- How are you currently communicating with employees about the impact of AI on the organisation, and what do you know about how that message lands?
- Where do leaders find support and honesty when they themselves are uncertain?
- What structures exist to give leaders the time and space to be a present leader?
2. Stabilisers and explorers need to collaborate for scalable innovation
Background
AI is increasing the pressure on both efficiency and renewal at the same time. The challenge now is to connect the existing business with the new possibilities. These interests often pull in opposite directions within the organisation, and that tension is reproduced in the leadership team.
Stabilisers, who helped build the existing business, are skilled at scaling through processes and KPIs. Explorers are skilled at adopting new technology and testing new products and ways of working. Stabilisers want control. Explorers want freedom. They need each other but often find it difficult to understand each other.
This is not a question of personality or personal chemistry. It is a question of leadership and organisation. Leadership’s task is not to choose between stabilisers and explorers, but to create conditions for both logics to operate at the same time. Only then can the organisation both protect what works today and build what is needed tomorrow.
Example
A company wants to use AI in customer service. The operations director wants clear processes, risk assessments, and KPIs before anything is introduced. Business development wants to put a small team to immediately test what value the technology can actually create. Both are right, but if leadership manages both tracks with the same logic, they risk locking each other out.
What you can do
Step 1. Make visible where stabilisers and explorers collide. Identify the questions, forums, and decisions where the two logics are already pulling in different directions. Where does friction arise around pace, risk, mandate, resources, or follow-up? As long as this remains invisible, leadership risks interpreting it as a personal issue when it is actually about different assignments.
Step 2. Design different assignments for different logics. Not everything should be managed in the same way. Decide what needs stability, clear KPIs, and control. Also decide what needs learning, testing, and greater freedom. Secure the reallocation of resources as results emerge.
Step 3. Build bridges between the logics. Create roles, forums, and development paths where people are exposed to both perspectives. Otherwise the polarisation reproduces itself.
Step 4. Develop collaboration within the leadership team. Work actively on the leadership team’s collaboration. Strengthen the shared assignment and the mutual dependency.
Questions for leadership
- Have you deliberately designed space for exploration, or does it compete for the same resources as the core business?
- What do internal development paths look like? Do they reproduce existing polarisation, or do they provide breadth and greater understanding? How do we build leaders who understand both logics, not just one?
- Who in the leadership team represents the logic of stability and who represents the logic of change, and how does that collaboration work?
3. AI is the team’s colleague, not only the individual’s tool
Background
AI is often introduced initially as an individual tool for increased personal efficiency. That creates silent silos where everyone optimises separately. The paradox is that one of AI’s greatest untapped potentials is to strengthen the entire team’s capacity to think and decide together.
Example
With AI as the individual’s tool, members of a leadership team each prepare individually for a strategy meeting. The CFO summarises figures with AI. The CHRO analyses employee data. The CMO brings in external intelligence. They arrive with their own AI-analysed perspective. Nobody has assembled the full picture. The meeting starts with people talking past each other.
With AI as the team’s colleague: the different inputs have been processed with the help of AI, which has identified tensions and contradictions between the perspectives before the meeting. The leadership team starts directly on the difficult questions.
What you can do
Step 1. Review how AI is actually used today. Is it an individual tool, or also a shared one? Where is the potential going forward?
Step 2. Design shared ways of working with AI. Not policies and guidelines, but concrete agreements within the team about how AI is used together. What do you share? What do you run jointly? Where do you need a human voice as a complement?
Step 3. Test AI as the team’s colleague in real situations. Start small. Let AI prepare a decision basis that the whole team sees before the meeting. Evaluate what that does to the quality of the conversation. Then use a transcription tool connected to AI that identifies questions the group did not ask but should have.
Questions for leadership
- Is AI making teams better or just individuals faster?
- Where do we need shared ways of working with AI rather than just individual solutions?
- How do we use AI to strengthen the quality of conversations and decisions, not just the pace of preparation?
4. AI creates new possibilities for continuous organising
Background
Since the beginning of industrialisation, organisations have been structured in “boxes” with department names and job titles. Complex and fast-moving challenges are bursting through these traditional ways of organising. It is simply too slow to redistribute resources when they are locked into a fixed departmental structure.
Organisations are therefore experimenting with new ways to organise themselves. New technology makes it possible to openly advertise tasks and assignments that employees can volunteer to take on. Resources are redistributed more quickly and collaboration across departmental boundaries becomes more natural. Every organisation is different, but AI now gives us the ability to organise in entirely new ways.
It is leadership’s role to challenge existing ways of organising and distributing resources.
Example
A global industrial company notices that competence is locked into business units while strategic initiatives are starved of resources. HR, IT, and business development test an internal platform where employees can see ongoing projects and express interest in tasks outside their regular role. Within six months, mobility has increased and several initiatives that had previously stalled have gained new momentum, without a single reorganisation.
What you can do
Step 1. Identify where the current structure is slowing things down. Where in the organisation does it become clear that the ordinary structure is not sufficient? Where does it lock both people and money into things that have already been decided, even though needs have changed?
Step 2. Run a smaller pilot. Carry out a pilot to see what questions arise and how the pilot could be scaled up.
Questions for leadership
- Where in the organisation are competence, time, mandate, and budget stuck in ways that are slowing down important initiatives?
- Which parts of the work would benefit from being organised more continuously, without first going through a large reorganisation?
- How can we create greater mobility in the distribution of resources, both people and budget, as needs change?
5. AI is already in your systems
Background
It is easy to think that the AI question is about future decisions. Which tool to procure and when to run a pilot. But existing system support for areas such as sales, finance, programming, and workforce planning has most likely already received AI functionality that is already influencing ways of working.
This is a leadership question because the technology does not only create efficiencies and new business opportunities. It also affects what becomes important in the work, what competencies are needed, and what you choose to measure and reward.
Example
Maria is a B2B sales representative. Her CRM automatically flags expiring contracts, prioritises the customer list, and suggests the next action. She delivers more than ever. But the role now requires something different from her: the ability to scrutinise and question what the system suggests, to understand the logic behind the priorities, and to explain them to colleagues when questions arise.
Adam works in financial reporting. The financial system compiles, analyses, and prepares communications for relevant managers. Adam’s value no longer lies in producing the underlying material, but in interpreting it, seeing what is missing, and taking responsibility for the conclusions when managers ask questions that require deep knowledge.
What you can do
Step 1. Map where AI already exists in your most important systems. Go through your most important systems — not the IT department’s list, but the systems where decisions are made and work is done. For each system: what AI features already exist, what do they affect, and which roles are changing?
Step 2. Challenge your system suppliers. Book meetings that are not about support or contract renewal, but about direction. What is happening in product development? How are other organisations using the system? How does it affect their ways of working? What competencies are needed?
Step 3. Review what you measure, develop, and reward. When AI takes over parts of the operational work, what creates value in a role also changes. You then need to review which performances you monitor, which competencies you develop, and what you reward. Otherwise you risk continuing to measure what the system does rather than what the person now needs to contribute.
Step 4. Ensure that learning does not disappear with the task. When AI takes over the production of a piece of work, the learning that happened in the doing often disappears with it. The person who no longer compiles the underlying material also stops understanding it in depth. Ensure that employees regularly encounter the raw material, not just AI’s conclusions. Build in moments where they practise interpreting, questioning, and explaining — not just delivering.
Questions for leadership
- Do we understand how AI in our most important systems is affecting ways of working?
- Do we know what the most important human competencies are now that AI is taking over parts of the operational work?
- Are we still measuring and rewarding yesterday’s work, or what creates value now?
- How do we protect learning and deep knowledge as the system does more of the work?
6. Reverse the competence debt before it reverses you
Background
AI takes the simplest tasks first. Precisely the tasks where junior academics traditionally learned their profession. The learning ladder disappears, but it is not immediately obvious.
Senior employees hold deep knowledge and experience that needs to be transferred to junior colleagues. If that transfer does not happen, the competence leaves the organisation when senior employees retire or move on.
This is a leadership question. If AI changes how work is done, leadership must also change how learning, knowledge transfer, and development happen.
Example
A legal firm introduces AI for research and analysis. Junior employees deliver faster than ever. But they never encounter the senior legal partners in the difficult moments — in the situations where experience and judgement are decisive. After two years they have technical competence but lack the deep understanding of client relationships and business logic that the senior consultants carry. The organisation’s memory is no longer transferred automatically.
What you can do
Step 1. Make the invisible visible. Map where learning actually happens today — not in theory but in practice. Which tasks previously gave deep understanding of the organisation, the clients, and the business? Which of those is AI now taking over?
Step 2. Redesign how knowledge is transferred. Deep knowledge sits in the heads of senior employees and is no longer transferred automatically. That needs to be replaced by something intentional: structured conversations after difficult situations, side-by-side work in complex client meetings and decisions, shared reflection when something went wrong or unexpectedly right. Not as a programme, but as part of how the work is organised.
Step 3. Expose junior employees to the right situations early. Not more courses. Instead, deliberate choices about which meetings, which clients, and which decisions junior employees should be involved in — earlier and more systematically than before. Ensure that junior employees encounter both the logic of stability and the logic of change. Also explore how mentors, coaches, and AI coaches can support learning in everyday work, for both individuals and groups.
Questions for leadership
- Do you know where the organisation’s memory resides, and do you have a plan for how it is passed on?
- How do you ensure that junior employees encounter both logics — stability’s and change’s?
- How do you secure time and resources for knowledge transfer?
Sources
Anthropic: Labor Market Impacts — Deep analysis of AI’s actual impact on the labour market based on real user data.
Beyond Talking: Insight Work 2025 — Thousands of managers and employees asked about their greatest challenge. Lack of time consistently ranks number one. Three strategies identified: #protectfocus, #establishdirection, and #acttogether.
Deloitte: Reinventing Workforce Planning (2025) — 77% of managers report that the ability to flexibly move competencies to where they are needed is decisive.
Frida Pemer & Andreas Werr: Defusing Digital Disruption Through Creative Accumulation (2023) — Digital transformation requires organisations to actively build new competence in parallel with delivering. Learning must be embedded in daily work, not separated into training programmes.
Haier: How Haier Works (video) — Illustration of how Haier organises itself in micro-enterprises with their own profit and loss responsibility.
Harvard Business Review: The Fluid Future of Work (2025) — In industries where AI is replacing entry-level jobs, organisations are losing the environments where future leaders developed.
Home Depot: Magic Apron — AI built directly into the store associate’s workflow. Shows how AI changes work without the employee choosing it.
Humanocracy: Creating Organisations as Amazing as the People Inside Them — Framework for building organisations with a high degree of autonomy and decentralised decision-making.
i4CP: HR Priorities Study 2026 — The 26% who are actually scaling value from AI invest 70% of their resources in people and processes, not technology.
IMD: 2026 AI Trends — When AI handles scale and speed, human judgement becomes the real differentiator.
IMD: Ambidextrous Leadership (2025) — Successful leaders actively move between Perform and Transform. There is no longer a single best way to lead.
Josh Bersin: The Superworker (2025) — Organisations that succeed with AI learn to operate with fewer people and higher productivity, but that requires shared ways of working.
Larsson & Hatzigeorgiou: The Future of Labour — Analysis of how AI and technological disruption are transforming the labour market and which competencies are in demand.
Mattias Axelson: Dold potential (available in Swedish only) — Every organisation has untapped potential, but the core business consumes all the resources. AI can free up capacity for what otherwise never gets done.
McKinsey: Developing Human Leadership in the Age of AI (2026) — Three areas where only humans can lead: setting ambition and meaning, showing judgement and taking responsibility, and reading the room emotionally to mobilise people.
P&G x Harvard D³ Institute: AI as Cybernetic Teammate — Human and AI in team create multiplicative effects. The team’s collective output exceeds the sum of individuals’ AI use separately.
Salesforce: State of Sales 2024 — Sales representatives save on average 2 to 5 hours per week with AI, but freed-up time is immediately filled with higher expectations. 83% of sales teams with AI saw revenue growth versus 66% without.
Stanford Digital Economy Lab: Canaries in the Coal Mine (2025) — 22 to 25-year-olds in AI-exposed occupations saw a 6% decline in employment from 2022 to 2025. The 30-plus group saw 6 to 13% growth.
WEF: Future of Jobs Report 2025 — 170 million new roles and 92 million disappearing by 2030. Net gain globally, but the redistribution is not frictionless.