Helping Teams Embrace Change: Solving Everyday Problems Together with AI

Start with the work people are trying to get through

After the session, several people asked whether they could listen to the replay. Those messages meant a lot to me. More than 70 people joined iCentra’s “The Governance Gap: Why Enterprise Transformation Fails Before It Scales,” including participants from Nigeria, Uganda, South Africa, and beyond. Speaking from Taiwan late at night, I listened to questions about older systems, customer habits, and employees’ concerns about change. One question stayed with me: where do you start when an organization does not have a transformational mindset? My answer was to start with employees’ own problems. This reflection draws on what I shared during the session, with some extended examples to explain how that approach could work.

Leaders may see transformation through the organization’s direction for the next three years. Employees may be thinking about the unfinished work waiting tomorrow morning. Will the new system make the process harder? Who will teach them? Who will help if AI makes a mistake? These questions deserve proper answers. I would invite a team to bring one task that regularly causes difficulty and explain what they have already tried. An improvement might mean entering information once instead of twice, or handling an exception without always waiting for the same manager. When employees help define the problem and check whether the change makes their work easier, they have a practical reason to participate. The organization’s ambition starts to connect with something they experience every day.

Use AI to explore perspectives, then ask people to check them

During the webinar, I spoke about including AI in the team’s way of working. That requires context, relevant rules, and clear limits on what it may do. People need to know when to take over. When thinking about empathy, I am interested in how AI can help us examine the difficulties different people face. For example, we could ask it to suggest what a new employee might find confusing in a process, or what a customer might need if a service stops halfway through. We would then discuss those suggestions with the people involved. Someone might say, “That part is fine. The difficulty is somewhere else.” That correction is useful. AI supplies possible perspectives, and people with direct experience help us check them. The exercise creates an opportunity to listen to colleagues whose concerns might otherwise be missed.

Rehearse the moment someone needs help

One example I shared was an AI customer service assistant. During a pilot, a project manager helps with difficult cases and IT quickly corrects the knowledge base. The service appears to work well. But after it expands to several locations, what happens if it uses an outdated refund policy? Who confirms the correct rule, updates the information, and supports the waiting customer? The employee at the counter needs to know what they can do and where to get help. Before expanding the service, I would bring policy, technology, and operations colleagues together to rehearse that situation. AI could suggest variations involving an unavailable manager, an offline system, or different local rules. The team would check which scenarios are realistic and agree on the response, including who coordinates it, who can make decisions, and when the service should pause.

Work through the options together

Rehearsal also gives teams a chance to practise making decisions together. An AI agent can organize questions and suggest several possible responses. People then verify the information, discuss the consequences, and have the person with the relevant authority confirm the way forward. Decisions outside that authority go to the appropriate level. The invoice-processing example I discussed follows the same logic. An agent may read invoices and prepare entries, but duplicate invoices, missing purchase orders, and changes to payment details require clear handling rules. Finance, security, technology, and operations each contribute knowledge. Initially, the agent could prepare an entry for a person to review. Once the results are reliable, the team can consider a wider role. These conversations help colleagues understand the reasons behind each other’s requirements and identify a solution they can carry out.

Respect the knowledge already in the organization

When answering a question about legacy banking technology, I drew on my previous experience. Colleagues who understand an existing system may know why a complicated process was introduced and remember exceptions that were never documented. We need their knowledge when deciding what to improve, what still works well, and how a new approach could fit the current environment. Inviting them into the discussion gives the team a better basis for those decisions. That respect should continue when the work produces results. If employees identified the problem, tested the solution, and spent time correcting AI’s mistakes, their contribution should be visible. Saying only that “AI improved efficiency” leaves out the people who helped make the improvement possible. Their experience belongs in both the design discussion and the account of what the team achieved.

Make the team’s contribution visible

This was a point I particularly wanted to make during the session: seeing work improve can build confidence, and knowing that your effort is valued gives you a reason to stay involved. Recognition can be specific. Explain whose observation exposed a gap, which suggestion reduced repeated work, or how a test helped prevent a problem. At the same time, people should feel able to discuss what remains difficult. Did the time saved create extra work for someone else? Was support available when an error occurred? Listening to those answers helps make a successful trial a basis for further cooperation. People have experienced their ideas being heard and their difficulties being addressed. That gives them something concrete to draw on when the next change is proposed, even if they were uncertain at the beginning.

Choose one problem for the next 90 days

My closing suggestion was to choose one business problem to work on over the next 90 days. Bring the relevant people together, agree on a clear measure of success, and use AI to explore and test possible solutions. Review the results every few weeks. At the end, share what worked, recognize the team’s contribution, and decide together what to improve or expand. Thank you to Taopheeq Babayeju, Alfred Okoh, our moderator Loom Ambe, and the iCentra team for making this exchange possible. I am also grateful to everyone who brought a workplace challenge into the discussion. If you are preparing to introduce a change, start by asking your colleagues what is difficult in their work and whether you can test an improvement together. When their effort makes a difference, make sure they hear it.