Fatima Al-Rashid manages a 20-person claims operations team at an insurance company. When her director told her the team would adopt an AI tool to triage and pre-fill routine claims, Fatima knew exactly what was coming, because she had been on the team for nine years before she managed it. The last big system rollout, five years earlier, had been followed by a quiet round of layoffs. Her team had long memories. At the first mention of AI, one of her most experienced adjusters, Robert, crossed his arms and said, "So this is how they replace us." Fatima did not have an enterprise change office or a polished communications budget. She had her team, her credibility, and the conversations she would have one at a time. This lesson follows how she led them through it.

Why people resist, and why their fears are not silly

Fatima's first instinct was to reassure everyone that nothing bad would happen. She caught herself. The fears were not irrational, and pretending they were would have cost her the trust she needed. AI can change jobs. It can reduce the autonomy of deciding how to work. It can make mistakes. Dismissing those realities would have made her sound either naive or dishonest, and her team would have spotted either one instantly.

So she mapped the actual sources of resistance on her team. Fear of displacement was the loudest: "will I still have a job?" Loss of autonomy was real too, because adjusters who had decided every case themselves would now be reviewing AI suggestions. Some worried about quality, that rushing claims through AI would let errors reach customers. Some, like Robert, distrusted leadership because of the layoffs five years back. And several were simply change-fatigued, having lived through rollouts that promised the world and delivered headaches. Each of these needed a different response, and Fatima's job was not to argue people out of their feelings but to address what was underneath them honestly.

Worked example: mapping the 20 people before saying a word

Before her first team conversation, Fatima did something that changed how she spent her energy. She sat down with a notepad and sorted all 20 people on her team into five groups, based on where each actually stood. This is a stakeholder map, and it kept her from the classic mistake of pouring effort into people who would never move while ignoring the people she could actually win.

• Champions (3 people). Already enthusiastic. Priya had been asking for better tools for a year. These three did not need convincing; they needed a job to do.
• Persuadables (8 people). Open but uncertain. The largest group, and Fatima's real target. They needed information and reassurance, and good communication would tip them toward support.
• Skeptics (5 people). Doubtful, with concrete concerns about quality and workload. Not opposed, but they needed to be heard and answered.
• Resisters (2 people). Actively opposed, Robert among them. Fatima decided not to spend most of her energy trying to convert them, only to keep them from blocking everyone else.
• Fence-sitters (2 people). Genuinely indifferent, waiting to see which way the wind blew. They would follow the majority, so winning the persuadables would pull them along.

The map made her strategy obvious. Eight persuadables plus two fence-sitters meant ten of twenty people were genuinely movable, and that is where she would invest. Her three champions became her amplifiers. Her five skeptics got listening time and honest answers. Her two resisters got respect and boundaries, but not her whole week. Most managers burn out arguing with the two resisters and neglect the ten people they could actually reach. Fatima refused to make that trade.

Worked example: using ADKAR to guide the team through

Knowing who stood where told Fatima nothing about what to do next. For that she used ADKAR, a change model that names the five things a person needs, in order, to make a change stick: Awareness of why the change is needed, Desire to support it, Knowledge of how to do it, Ability to actually do it, and Reinforcement to keep it from sliding back. ADKAR is useful precisely because it is sequential. You cannot give someone Knowledge if they have no Desire, and you cannot build Desire if they have no Awareness of why anything is changing.

Fatima ran a quick honest assessment of where her team sat on each step, scoring the team's average readiness from 1 to 5:

• Awareness: 2 of 5. People had heard "AI is coming" but did not understand why or what problem it solved. This was her starting gap.
• Desire: 2 of 5. Low, dragged down by the layoff memory. This was the hardest step and the one tied to trust.
• Knowledge: 1 of 5. Almost nobody knew how the tool actually worked yet.
• Ability: 1 of 5. Naturally, since no one had touched it.
• Reinforcement: not yet relevant.

The scores told her not to lead with training. Knowledge and Ability would be wasted while Awareness and Desire sat at 2. So she sequenced her plan. First she built Awareness by explaining, plainly and specifically, why the change was happening: claims were taking an average of two hours to reach an adjuster, customers were frustrated, and the AI would handle the routine pre-fill so adjusters could spend their time on the complex cases that actually needed human judgment. Then she worked on Desire, which meant tackling the layoff fear head-on rather than around it. Only once those rose did she move the team into Knowledge through training, then Ability through a parallel-run period, and finally Reinforcement by celebrating and sustaining the new way of working. Six weeks in, she reassessed: Awareness had reached 4, Desire 3, and she shifted her weight to the training phase.

Crafting the message: empathy first, then honesty

Fatima opened her first team meeting not with a slide deck but with an acknowledgment. "I know many of you are worried about what this means for your jobs. Those worries are reasonable, and given what happened a few years ago, I understand exactly why they are there. Let me be honest about what is changing and what is not." Starting with empathy, and naming the layoff history out loud instead of dodging it, bought her the credibility to say anything else.

Then she was honest about the change rather than pretending nothing was happening. "This tool will pre-fill routine claims, the straightforward auto-glass and minor-fender cases. That means fewer hours on data entry. It also means more of your time goes to the complicated, disputed, human cases, which is the work most of you actually find interesting." Honesty about real change is far more credible than a promise that nothing will change, which nobody believes anyway.

She connected the change to a purpose her team cared about. Customers waiting two hours for a first response was a source of the angry calls her adjusters hated taking. Cutting that wait was something they could want for their own sake, not just the company's. And she outlined the transition concretely, because uncertainty is what breeds the worst anxiety: three weeks of training, then a two-month parallel run with the old and new processes side by side, then full cutover, with peer mentoring and one-on-one help available the whole way.

Addressing the specific objections, one at a time

General reassurance does not work; specific answers do. Fatima prepared honest responses to the objections she knew were coming, and she did not sugar-coat them.

To "AI will eliminate my job," she did not promise the impossible. She said what was actually true: the team's volume was growing, no positions were being cut as a result of this tool, the work was being shifted from routine entry to higher-value judgment, and if anything ever did change she would tell people early and manage it responsibly. To "AI makes mistakes, I do not trust it," she agreed: "It does, which is exactly why it does not decide anything. It suggests, you review, you have the final say, and nothing reaches a customer without a human approving it." To "I am too old to learn this," she pointed to the training and support and to the fact that the tool was built to be simple, no technical background required. And to the sharpest one, "this is just about cutting headcount," she was straight: efficiency does help the business, she would not pretend otherwise, but the driver here was faster service and freeing adjusters from drudge work, and she would be judged by whether that turned out to be true.

Building a coalition and letting results talk

Fatima knew her own voice had a ceiling. She was management, and to a skeptic that made her suspect by default. So she put her three champions to work. Priya ran the AI tool in a two-week pilot, then stood up in a team meeting and described, in plain adjuster language, what was better and what was annoying about it. Peer-to-peer honesty from someone who did the same job landed in a way Fatima's reassurance never could.

Then she let results do the heavy lifting. After the pilot, she shared real numbers with the team: the pilot group was clearing roughly 30 percent more claims with the same headcount, first-response time had dropped from two hours to about 15 minutes, and customer complaints on those claim types were down. She was careful not to oversell. She presented the numbers plainly and let them speak, because the fastest way to lose a skeptic is to promise more than you deliver. Results decreased skepticism more than any speech could.

Leading through the resistance that remains

Even with a good plan, resistance did not vanish, and Fatima learned to expect it rather than be wounded by it. She kept listening, asking "what specifically worries you about this?" because sometimes a skeptic's concern revealed a genuine flaw she needed to fix, like a claim type the AI handled badly. She communicated consistently, repeating her message across team meetings, one-on-ones, and the champions' voices, because people need to hear something several times from several sources before it lands. She acknowledged progress when adjusters adapted, naming it specifically.

And with her two resisters she held a firm, respectful boundary. With Robert she was direct: "I understand this is hard, and I am not going to pretend your concerns are baseless. But this is the direction we are going, and I will support you in adapting to it." She did not let his resistance hold the other eighteen hostage. Interestingly, when the parallel run showed the tool was not coming for his job and his routine workload genuinely eased, Robert softened on his own. Results reached him where arguments never would have. When the change finally delivered, Fatima celebrated it openly and credited the team's flexibility, because reinforcement is what keeps a change from quietly sliding back.

Three ways Fatima could have made it worse

Looking back, Fatima could name the three mistakes she had watched other managers make, each of which she had felt tempted by in the moment. The first is dismissing concerns: "people always resist change, they'll get over it." That single sentence would have confirmed every fear her team held about leadership not caring, and deepened the resistance instead of easing it. The antidote was the opposite reflex, to listen and answer substantively, even when the concern was inconvenient.

The second is overselling. It would have been easy to stand up and promise that the AI would fix everything and everyone would love it. The first time the tool fumbled a tricky claim, which it did, that promise would have shattered her credibility and handed the resisters their proof. By deliberately under-promising on the pilot numbers and letting the results slightly exceed what she had claimed, she built trust instead of spending it. Under-promise and over-deliver is not a slogan here; it is how a skeptical team decides whether to believe you next time.

The third is communicating once and assuming the job is done. One all-hands announcement feels like communication to the manager who gives it, but it barely registers with people who are anxious and distracted. Fatima repeated her message across formats and weeks, invited questions every time, and accepted that hearing something once is not the same as believing it. People absorb a hard message slowly, from several directions, before it finally settles.

Key Takeaways

• Take the fears seriously. AI can change jobs, reduce autonomy, and make mistakes. Dismissing those concerns as irrational destroys the trust you need; addressing them honestly is what builds it.
• Map your people before you spend your energy. Sort stakeholders into champions, persuadables, skeptics, resisters, and fence-sitters. The persuadables and fence-sitters are where movement happens; do not burn your week arguing with the two resisters.
• Use ADKAR in order. People need Awareness, then Desire, then Knowledge, then Ability, then Reinforcement. Assess where the team actually sits and do not start training (Knowledge) while Awareness and Desire are still low.
• Open with empathy, then be honest about real change. Acknowledge concerns first, including any painful history, then tell the truth about what is changing. Honesty is more credible than a promise that nothing will change.
• Answer specific objections specifically. "AI will take my job," "it makes mistakes," "I am too old," and "this is just about headcount" each deserve a prepared, honest, non-sugar-coated response.
• Let peers and results carry the message. A champion who does the same job is more persuasive than a manager, and real numbers, presented without overselling, move skeptics more than any speech.
• Expect resistance and hold respectful boundaries. Listen for genuine flaws, communicate consistently, and with the few who will not move, be firm and respectful without letting them block everyone else.
• Reinforce the win. When the change delivers, celebrate it and credit the team, so the new way of working sticks instead of sliding back.