Practice

Leading AI transformation

We work with the leaders and managers who have to turn an AI ambition into changes in how their teams work.

AI changes what people do and how teams work together. It also changes the manager’s role. Managers are usually the ones asked to lead that change, often without a method for doing it. Leading it well takes a view of where the value is, an understanding of what the technology can and cannot do, and knowledge of how people and organizations change.

The two questions

An AI program has to answer two questions: where the value is, and who has to learn something new. Programs usually put their budget and attention on the first one, with use cases, platform decisions and a steering group. The second is often handled as a communication task.

A communication task works when the knowledge already exists and only has to reach people. In an AI program, that knowledge often does not exist yet. People have to work out new ways of doing their jobs and decide what now counts as good output. A communication plan cannot do that for them.

That work often has no clear owner. The executive team owns the ambition and the technology function owns the platform. The part where people have to work differently usually ends up with the managers.

Why this is hard

An AI transformation needs several different kinds of expertise working on the same problem at once.

• A strategic view: Someone has to decide which parts of the business to change first, what that change is worth, and what to leave alone for now.
• A working grasp of the technology: Someone needs to know what these tools can do today, where they fail, and which failures matter for your work. The tools change quickly, so that knowledge has to be kept current.
• Risk and compliance, brought in early: Many of the use cases worth doing involve data that somebody is accountable for. If that person is only brought in at the end, a working pilot often stops there.
• People, and how organizations change: Roles, incentives, career paths and what people believe good work looks like all shift. This sits with human resources and learning, and it is usually the largest part of the work.
• Mechanisms: This work needs review points, criteria, named owners, and something that inspects whether any of it happened.
• Experiments, and the ones that fail: The answer usually cannot be specified in advance, so progress comes from running things and reading the result. That needs an agreed definition of an acceptable failure, settled before the first failure happens.
• Empathy, and staying in the conversation: People are being asked to change work they are good at, sometimes work that defines their professional and personal identity. A leader that shies away from that conversation will get a polite version of it and act on the wrong information.
• Culture, capability and resilience: The work runs over years. The organization has to be able to keep going after the first two or three attempts do not produce much, which is often the case.

Many organizations already have each of these somewhere. They sit in different functions, with their own budgets and reporting lines, and have rarely had to work on one problem together. This kind of transformation needs all the relevant departments and disciplines to come together.

What is different for a manager

Some managers have led a change before. It might have been a reorganization, a system migration, or a new operating model. AI transformation asks things of them that previous changes did not.

• You may not be the most expert person in the room: Managers have usually done the work they now manage. With AI tools, some of your people are further ahead than you are, and that will keep happening as the tools change.
• Your output changes from answers to questions: With a technical challenge, a manager assigns tasks, finds resources and removes obstacles. An adaptive challenge needs the manager to frame a question the team can work on, protect them while they work on it, and say clearly what is not changing.
• You have to name what people are losing: Rollouts tend to announce the gains and say little about what people will lose, which might be work they were good at and part of why they were valued. When that loss is not spelled out, people tend to object to the process. Those objections are hard to resolve because they are about the loss.
• You have to make it safe to be slower before faster: Working with an unfamiliar tool is slower at first. If quarterly targets penalize that, people will use the tool where it is safe and keep doing their important work the old way.
• You have to change what gets measured: People follow what they are measured and recognized for. If reviews and recognition still reward the old way of working, encouragement will not change much.

Two kinds of challenges

Start by working out which kind of problem you are facing. Most changes in an AI program mix the two, and the mix determines who needs to do the work.

• Technical challenges: Somebody already knows how. The knowledge exists, inside the organization or outside it, and it can be hired, bought or taught. The problem can be stated clearly and success measured against a known target. It feels like pressure: hard, sometimes exhausting, and orienting, because you know what good looks like and who is accountable.
• Adaptive challenges: No existing expertise contains the answer. Progress requires the people involved to change what they believe, what they value or how they work, so it cannot be handed to a specialist. It feels like distress: disorienting and personal, conflicts surface, people avoid it, and to some of them it feels like loss.

Technical challenges can be very hard. A heart bypass is a technical challenge. The surgeon needs years of training and thousands of repetitions, and the patient’s life is at stake. It counts as technical because the surgeon already knows how to do it, and the patient does not have to change any habits. Getting that same patient to eat, sleep and handle stress differently once they are home is the adaptive piece, and it is critical to the outcome. Many AI rollouts that stall have applied a technical answer to an adaptive challenge.

This is our own framework. Its starting point is Adaptive Leadership, from Ronald Heifetz and his colleagues at the Harvard Kennedy School, and we connected it to our AI value framework and to what we have learned working with leadership teams.

Adaptive challenges across the three value paths

We sort every engagement onto three value paths, and each path has a different mix of technical and adaptive challenges.

• Boosting productivity: This path is mostly technical. The workflows are understood and the tools are proven, which is why many organizations start here. The adaptive part is small but important: somebody is being asked to give up a craft they were proud of, and rollout plans rarely address it. The leader’s move: Use the tools in your own work first, visibly, before asking others to change theirs.
• Creating new value: This path is roughly half technical and half adaptive. Building the product is a technical challenge. Deciding what deserves to exist, who it is for, and which current commitments it affects is an adaptive challenge. The leader’s move: Define what a good experiment looks like before anyone runs one, protect the work from short-term pressure, and stop the bets that are not working.
• Driving disruption: This path is mostly adaptive. There is no reference implementation for a business model that does not exist yet, and the people who would have to let go of the current one are the same people holding the decision. The leader’s move: Discuss the disruption openly, define what does not change and what to let go of, and hold the tension without resolving it too soon.

The AI value framework in full

Three moves a leader makes

A leader uses all three, often within a single conversation and more than once.

• Diagnose the work: Split the challenge into its technical and adaptive parts. For the adaptive part, ask who has to learn something and what they have to give up. Write the challenge as one sentence with a verb in it. Repeat this every quarter, because the line moves as the tools improve.
• Regulate the heat: Keep enough pressure on people to move, but not so much that they freeze. Raise it by naming what is at stake and keeping difficult questions open. Lower it by breaking the work into smaller pieces, setting a pace, and sequencing what comes first. Many managers are comfortable with only one of the two.
• Give the work back: Frame the challenge without handing over the answer, and ask who needs to learn what. If your sentence still contains the solution, you have delegated the task and kept the problem. You stay responsible for the standard, the deadline, the pace and the protection, and you back the experiment when it does not go to plan.

How adaptive challenges become technical

Adaptive challenges become technical over time. People test, experiment and eventually turn new ways of working into routine. Once something is routine, it can be specified and verified, a best practice can be written down, and AI can take over more of it. That frees up capacity.

Leaders then have to decide where that capacity goes. Using it for things that were not possible before usually creates more value than automating more of what people already do.

Because the boundary keeps moving, managers need to repeat the diagnosis regularly.

We run the framework as a working session. Participants bring challenges they are dealing with, sort them together, and each person leaves with one sentence to act on.

• For a leadership team: The group that set the ambition sorts the three or four changes it is driving. The output is an agreement about which parts have a known answer and which do not, who owns each, and where to raise or lower the heat.
• For a management layer: We run it as a short series for the managers who lead the work. Each session works on current challenges from their own teams. The managers practice what they need to do on those challenges and build a shared vocabulary they can use with each other.
• Inside a program you already run: If you already run a management or talent program, the workshop can become a module within it, using the measures, vocabulary and cases you already have.

Where this connects

• The leader’s role on each path: A table sets out what AI can do, what the team does and what the leader does on each of the three value paths. The leader’s role in AI transformation
• Where the value is: Our AI value framework scores use cases onto the three paths and names an owner and a first result for each. The AI value framework
• Enablement and adoption: Training, champions across functions, and governance help people work with AI safely. Implementation and adoption
• Mechanisms that hold: Review points, criteria and routines keep a change going after the launch. Innovation mechanisms
• Sessions that start it: Keynotes and workshops help a leadership team see what is possible before it commits to change. Inspiration and activation

Who this is for

This work is for executive teams who have set an AI ambition and cannot yet see it in the work, and for heads of function whose rollout slowed after launch. It is also for managers asked to lead a change they were not prepared for, and for human resources and learning teams whose leadership programs do not yet cover it.

Common questions

What is the difference between AI training and AI adoption?

Training moves knowledge that already exists to people who do not have it yet, and it is often necessary. Adoption is what happens afterwards: people decide whether to change how they work, what to stop doing by hand, and whether they trust the output enough to put their name on it. A program that funds only training can end with trained people whose work has not changed.

Why is leading an AI transformation harder than other change programs?

It needs several kinds of expertise at once: strategy, a current understanding of the technology, risk and compliance judgement, knowledge of people and organizational change, and mechanisms that make changes last. These usually have different owners. The destination is also unclear, and some of the people being led know more about the tools than their manager.

Our AI rollout stalled. Where do we start?

Start by sorting the specific thing that stalled. Describe it in a single sentence, then ask whether somebody already knows how to solve it, which people need to learn something new, and what they would lose in the process. If it stalled on the adaptive part, a better tool or a stronger mandate is unlikely to restart it, and the next attempt needs to start with the people who have to change.

Who should be in the room?

The people who own the problem should be in the room, which usually means the leadership team for the framing and the managers for the work. Sending a delegate is a common way this goes wrong. When the adaptive part is handed to a specialist group, it turns back into a technical project and tends to stall again.

How long before we see anything?

A first sort takes a single session, and teams often start discussing their work differently the same week. It takes a few months of practice on real decisions before managers use the moves without help. Changes in how the organization handles new challenges show up over several quarters.

We shape each engagement around the value you are aiming for and the people who will lead the change. We also adjust it to how many of your challenges turn out to be adaptive.